511TESOL QUARTERLY Vol 36 No 4 Winter 2002

Cultural Representationsof Rhetorical ConventionsThe Effects on Reading RecallHSI-CHIN JANET CHUProvidence UniversityTaiwan Republic of China

JANET SWAFFAR and DAVIDA H CHARNEYThe University of Texas at AustinAustin Texas United States

This study explores whether culture-speci c rhetorical conventionsaffect the reading recall of Chinese EFL college students at two gradelevels Four English passages veri ed as using Chinese rhetorical con-ventions were modi ed into four counterpart versions re ecting En-glish rhetorical conventions One hundred twenty Taiwanese freshmenand 120 seniors read two of the four passages one in each rhetoricalconvention After each reading students completed a passage percep-tion questionnaire and an immediate-recall test One week later partici-pants completed a delayed-recall test and a topic assessment question-naire Although students failed to perceive rhetorical differences differentrhetorical convention had a signi cant overall effect on Chinese stu-dentsrsquo reading comprehension in both immediate and delayed recallMoreover post hoc comparisons revealed that two topics among thefour re ected in the eight passages showed more impact from rhetoricalconvention than did the others Analysis of questionnaire data sug-gested that factors such as topic interest and topic familiarity moderatedthe effect of rhetorical convention The study concludes with sugges-tions for future research and classroom implications

The question of whether discourse conventions affect reading com-prehension has become increasingly important in an era of interna-

tional telecommunication dominated by texts that use English or West-ern rhetorical conventionsmdashtypically foregrounding focal topics andsupported by discourse markers specifying a linear ow of ideas and atext intent Given the very different rhetorical strategies identi ed ascharacteristic of non-Western cultures some researchers have speculatedthat comprehending texts written in rhetorical conventions unfamiliar

512 TESOL QUARTERLY

to the reader may hinge in part on familiarity or unfamiliarity with atextrsquos communicative styles (Eggington 1987 Hinds 1984 Kaplan 1966Ricento 1989) To address this issue the present study looked at whetherrhetorical style in and of itself affected the recall of Chinese studentsreading otherwise identical English language texts

RESEARCH ON EFFECTS OFRHETORICAL CONVENTIONS

Models of Reading Comprehension

Two of the most widely accepted models of reading comprehensionnotably those of Rumelhart (1977) and Kintsch (1998) agree thatcomprehension is a multicomponential interactive process One resultof such interactivity seems to be what Stanovich (1980) calls compensatoryprocessing ldquoa de cit in any knowledge source [that] results in a heavierreliance on other knowledge sources regardless of their level in theprocessing hierarchyrdquo (p 63)

For Kintsch (1998) the product of the comprehension can be a textbase model a situation model or something in between depending ontask demands (p 292) In this construction-integration model readersassemble textual units in two ways as recall and as situations applicableto their lives In either case when they experience problems in puttingtextual information into a coherent pattern readers call on generalknowledge and draw inferences elaborating on the actual text baseKintsch emphasizes that although integrating prior knowledge withtextual propositions still in working memory is automatic it takes upmental capacity Moreover because the integration process allows for allsorts of knowledge to come into play one feature of that process spreadactivation strengthens related items and suppresses unrelated ones (pp98ndash99)

The result is a text base that structures mental representation ofmicro- and macropropositions hierarchically These propositions repre-sent the readerrsquos recall of corresponding text and they enable reproduc-tion tasks such as recall and summary In the situation model theinformation carried in the text has been modi ed and embedded into areaderrsquos existing knowledge structure resulting in learning Althoughtraces of the original propositions and structure of text have been lost(eg a reader might say ldquoI canrsquot remember where I heardread thatrdquo)the macrostructure or rhetorical logic the author has chosen in uencesthe readerrsquos original processing (Weaver amp Kintsch 1991)

Kintschrsquos (1998) insights suggest that the macrostructures or rhetori-

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 513

cal conventions in the text are not only vital to textual comprehensionbut also essential for the readersrsquo intake of information and possiblereconstruction of the text Macrostructures do more than inform con-struction of macropropositions They also enable retrieval of proposi-tions and their integration into long-term memory (van Dijk amp Kintsch1983) When readers process a text with unfamiliar macrostructures thisunfamiliarity might in uence the construction of both text base and textsituation Whether the task facing readers demands recall or applicationde ciencies in their knowledge of rhetorical conventionsmdashthe macro-structures that reveal textual organizationmdashcould overtax readersrsquo syn-thesizing capabilities

Another way in which Kintsch (1998) speaks to the signi cance ofrhetorical conventions is by emphasizing that readers with a sense ofcoherence different from that suggested in the discourse features of thetext might be connecting propositions in ways different from thoseintended by the author Schnotzrsquos (1984) subjects who read two textswith the same content but contrasting thematic relationships seem toillustrate Kintschrsquos assertion that readers introduce inferential changesto the text base when its content is reorganized Schnotz describes thesedifferences as two distinct types of organization from which a reader maychoose The processes resulting from that choice will result in ldquodifferentknowledge structures for both organization types although the textcontent may be the samerdquo (p 71) Similarly Goetz and Armbruster(1980) conclude that connected discourse is easier to comprehend thanunrelated and disconnected content because the reader is able toldquoorganize and interrelate elements in the textrdquo (p 206) They suggestthat the textrsquos logical structure when perceived by the reader facilitatesthat readerrsquos schematizing process The reader is consequently able toanticipate text content and construct meaning in a relational frameworkthereby retaining content in memory as an integrated story rather thanas disconnected pieces

Studies Examining the Impact ofRhetorical Structure on L2 Reading

L1 research has established that whenever a mismatch between textualorganization and reader expectations occurs readers tend to distort atextrsquos meaning their processing ef ciency is impaired and their reten-tion is short-circuited (Brennan Connie amp Winograd 1986 Kintsch ampYarbrough 1982 Meyer Brandt amp Bluth 1980) With regard to L2readers research on the impact of rhetorical structure sometimesreferred to as formal schema (eg Barnitz 1986 Carrell 1984) also

514 TESOL QUARTERLY

supports the claim that familiarity with rhetorical conventions plays arole in recall However most studies on text structure in L2 reading havefocused on how the logical organization of textual information typical inWestern writing in uences text processing and recall (Carrell 1984)Tian (1990) replicating Carrellrsquos 1984 study in Singapore introducedthe additional variable of home language group (Chinese Malay orTamil) As in Carrellrsquos study differences in rhetorical structure affectedthe readersrsquo recall Their native language however did not seem totrigger a different recall pattern for different rhetorical structures Tianconjectured that the homogeneity of the language environment inschools in Singapore leveled and neutralized the home language effectsHowever Carrell (1992) found a relationship between her ESL ChineseJapanese Arabic Malaysian Turkish Spanish and Korean readersrsquoawareness of text structure and their recall

Considered as a whole most studies on the effects of text structurehave been conducted in an ESL context with a focus on the effects oflogical organization rather than on the effects of rhetorical conventionviewed as a cultural phenomenon Whether L2 readersrsquo experience witha culture-speci c L1 rhetoric plays a role in L2 reading among readers atdifferent pro ciency levels has yet to be explored Work comparingstudentsrsquo background knowledge and interest in a topic (Carrell amp Wise1998) and the interaction between content schemata and formal sche-mata (Carrell 1987) points in this direction For this reason the presentstudy focuses on potential differences in recall of texts re ecting Westernand non-Western rhetorical conventions

Additionally many studies of L2 reading comprehension suggest apositive relationship between increased language pro ciency and abilityto recognize discourse cues (eg Cooper 1984) thus avoiding the short-circuits that are more common among readers with de cient languageskills (Clarke 1980 Cummins 1979) Therefore this study includedboth college freshmen and seniors who were English majors

RHETORICAL STRUCTURE IN CHINESEEXPOSITORY WRITING

Qi-Cheng-Zhuan-He

An approach to Asiatic writing qi-cheng-zhuan-he (ki-shoAring -ten-ketsu inJapanese and ki-sung-chon-kyual in Korean) is a prevalent though not theonly model used in analysis of Chinese texts It was originally employedby Fan-heng a scholar in the Yuen Dynasty to analyze poetry and waslater applied to structuring essays Though relevant for narrative and

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 515

poetry as well as public speaking it is most frequently applied inexpository writing (Kojima 1972) Literally qi means beginningmdashtheopening of a topic Cheng means followingmdashelaborating the openingZhuan means turning or changing Here the writer expresses anotherpoint of view turns to an example or develops the idea further Hemeans wrapping up when the writer provides the highest level ofgeneralization

Qi-cheng-zhuan-he is not a rigidly formalized structure the number ofwords and paragraphs as well as the amount of information devoted toeach of the four parts can vary disproportionately As long as the order ofpresentation remains the essence of the structure is there (Kojima1972) Hence qi-cheng-zhuan-he designates a conceptualreasoning se-quence in overall organization

Because qi-cheng-zhuan-he is simply one model for rhetorical sequenc-ing texts that use this model also exhibit other features commonly foundin Chinese writing These features differ markedly from those commonin Western writing (for examples see Appendix A) A tally of suchdifferences frequently starts with the Western preference for deductivestyle and the Chinese tendency toward inductive approaches particu-larly in expository writing The practice of putting the main thesis of atext before supporting ideas violates a Chinese readerrsquos expectation forwhat Kachru (1998) calls a ldquodelayed introduction of purposerdquo orldquodelayed topic statementrdquo (p 55)

Chen (1986) in comparing the expository discourse structure ofEnglish and Chinese passages found that Chinese paragraphs tended tointroduce more subtopics than did their Western counterparts Kirkpatrick(1995) quoting Wu (1988) suggests how inductive reasoning mightcondition Chinese rhetorical schemata Wu posits that inductive reason-ing is felt to be ldquonaturalrdquo whereas deductive reasoning requires aninformation sequence that goes against ldquoa natural sequencerdquo (p 281) asperceived by Chinese readers

A second difference researchers have noted between Western andChinese rhetorical conventions is closely related to the way main andsubordinate information is sequenced in each Discourse cues forsentential and intersentential cohesion tend not to occur as often inChinese as they do in English writing (Normant 1986) Further com-pounding this Chinese tendency toward indirect rather than directpresentation of authorial intent Jensen (1998) stated is that subtletyanalogy and metaphor highly valued writing techniques in classicalChinese tend to hide the writerrsquos intents He attributed this tendencytoward indirect allusion to the Chinese cultural heritage in rhetoric

As a result of these differences researchers have predicted thatChinese EFL readers because they have rhetorical expectations thatdiffer from those of Western readers might experience problems

516 TESOL QUARTERLY

comprehending a passage written with Western rhetorical conventionsMain ideas for example might not be recognized as such due to theirearly presentation and thus may fail to facilitate comprehension

Consequently Chinese EFL readers could be at a disadvantage whenreading English texts written by native speakers of English EFL readerswho predict a Chinese discourse structure may nd a mismatch betweentheir expectations and the macrostructures (the organization of contentand the sequence of ideas) presented in the Western text This gapbetween the Chinese readerrsquos anticipation of text development and thetextrsquos actual macrostrutures might result in lower comprehension andretention of an English text written in an English rhetorical structureConversely a text written in the rhetorical structure common in Chi-nesemdashthat is according to conventions such as those of qi-cheng-zhuan-he mdashmight facilitate recall for Chinese readers by providing them withfamiliar organizational macrostructures

The Role of Rhetorical Conventions inTaiwanese High Schools

In the past Taiwanese studentsrsquo exposure to English reading andwriting before they entered college focused more on the lexical andsyntactic features of a passage than on its rhetorical conventions Theirreading in Chinese on the other hand frequently dealt with theschemata of Chinese rhetorical conventions In terms of expositorywriting in their native language the classical Chinese rhetorical struc-ture qi-cheng-zhuan-he exerts a signi cant in uence as a macrostructurefor text analysis

For the students who participated in this study work with Englishlanguage texts did not emphasize the macrostructures of those passagesUntil 1999 when competitive marketing of textbooks was introducedthe traditional English textbooks commonly used in Taiwanese highschools offered little explicit training in the rhetorical organization oftexts Although reading formed the core component of these textbooksexercises for reading passages emphasized word- and sentence-levelmeanings rather than rhetorical organization Most of the readings werewritten originally by native speakers of English but were adapted oredited to control for length and to simplify vocabulary and syntax

The reading aids that commonly accompanied the reading passages inthe traditional textbooks were phonetic transcription and a glossary ofvocabulary translation and interpretation of idioms and phrases Cover-age of the rhetorical aspect of the reading was limited Reading andwriting exercises typically consisted of cloze vocabulary matching phrase

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 517

substitution translation questions on factual content and essay ques-tions on content exercises only rarely focused on the development ofideas and the discourse markers signaling the textual organization ofthose ideas Importantly for this study these textbooks did not explicitlycall studentsrsquo attention to rhetorical structures common in Englishwriting

In reading in Chinese on the other hand rhetorical structures play asigni cant role in secondary school instruction High school studentsrsquocourses in Chinese involve intensive reading of classical Chinese andmemorization of classical texts as part of the literary tradition A keypractice is the paraphrasing of classical texts into modern ChineseBecause text reproduction tasks require the construction of a text basestudents must inevitably attend to the global structure of the textconsciously or subconsciously in the retention and retrieval processesPossibly then students in this study developed robust text-structureschemata based on rhetorical conventions often found in Chinesewriting such as the above-mentioned qi-cheng-zhuan-he

The English Majorrsquos Literacy Experience at College in Taiwan

In Taiwan English majors such as those who participated in this studyenroll in content courses in linguistics and English literature in whichthey use collections such as The Norton Anthology of American Literature(Abrams 1996) that have been compiled for native-English-speakingstudents In the freshman year most courses focus on language skills afew are introductory courses in linguistics and literature As studentsmove on to the sophomore junior and senior years the proportion oflanguage skill courses decreases and content courses in linguistics andliterature increase Because this study was conducted while the freshmenwere in their rst semester and the seniors in their seventh semester thefreshmenrsquos exposure to authentic texts was much more limited than thatof the seniors

Regardless of grade level the texts in linguistics and literature thatEnglish majors read in college are very different from those they read inhigh school English classes The readings are considerably longer andmore demanding in content and style To increase their reading speedand their ability to cope with large quantities of EFL reading Englishmajors must either resort to translations or shift their reading style to amore macrotextual level

Whether reading English texts in British or US anthologies orChinese translations of those texts English majors in Taiwan are readinginformation presented in English rhetorical structure After 3 years ofsuch exposure students in their senior year may well be able to process

518 TESOL QUARTERLY

these texts as ef ciently as they can a text in Chinese rhetorical structureparticularly if the passage in question is expository or presents theauthorrsquos point of view typical for a classical essay written in the style of qi-cheng-zhuan-he No research has ever been conducted to test thatassumption

Chinese rhetorical structures found in the commonly used styles ofexpository texts contrast sharply with the rhetorical premises of Westernwriting Consequently the impact of culture-speci c rhetorical structureson the reading comprehension of Chinese students may be particularlyprofound for this type of writing Although researchers have looked atlanguage pro ciency (Carrell 1991) and text structure (Carrell 1992) asvariables to our knowledge no study has looked at both simultaneously1

This study examined the recall of readers at two different learning levelsmdashfreshmen and seniorsmdashwho read texts having identical content butrepresenting two distinct culture-speci c rhetorical conventions Thestudy poses the following questions (a) Does reading an L2 text thatfollows L1 rhetorical conventions affect EFL studentsrsquo recall and percep-tions of the text differently from reading the same text in L2 conventions(b) Do effects of rhetorical convention depend on the readerrsquos grade level

METHOD

Participants

To establish text selection criteria and choose suitable measurementsfor the main study we conducted a pilot study in June 1998 withfreshman English majors at Providence University a private university inTaiwan A total of 240 English majors at the same university participatedin the main study on a voluntary basis in December 1998 Half of theparticipants (98 females and 22 males) were recruited from fourfreshman sections and half (102 females and 18 males) were from foursenior classes The freshmen had had at least 6 years of formal EFLlanguage education in Taiwan which as noted above focused more onlanguage features than on the rhetorical aspects of English and includedrelatively limited access to extensive reading of naturally occurringEnglish prose written by Western authors for Western readers

In contrast the seniors had had 3 years of extensive exposure toEnglish texts written for native speakers of English in linguistics and

1 Work on the effects of rhetorical convention on reading comprehension has been done inother Asian languages notably Korean (Eggington 1987) and Japanese (Hinds 1987 Ricento1989)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 519

literature courses Presumably then the seniors had not only greater L2language pro ciency than did the freshmen but also considerably moreexposure to English rhetorical style Both ESL groups had had a formalMandarin Chinese education with a focus on classical Chinese while inhigh school and college and thus had been exposed to a rich L1environment with texts featuring traditional rhetorical conventions

Materials

Reading Passages

Four passages in English that followed Chinese rhetorical conventions(hereafter C versions) were selected from a Taiwanese bilingual magazineand modi ed to conform to English rhetorical conventions (hereafter Eversions) All the passages (Hsiao 1993a 1993b 1993c 1994) wereeditorials originally written in Chinese by a Chinese author and trans-lated into English for the magazine Editorials were chosen because thisgenre tends to use conventions that re ect features of qi-cheng-zhuan-heThe passages were judged as representative of qi-cheng-zhuan-he by sevenprofessors in the Department of Chinese Literature at a private universityin Taiwan

The four passages dealt with sociological issues of contemporaryTaiwanese society the unintended consequences of childcare for moth-ers with careers (ldquoChildcarerdquo Hsiao 1993c) the crisis one charityhospital in Taiwan faced because nonindigent patients abused its re-sources (ldquoCharityrdquo Hsiao 1993b) gender issues for male nurses inTaiwan (ldquoMale Nursesrdquo Hsiao 1994) and the legal problems facingparents who wanted to start alternative schools in Taiwan (ldquoSchoolingrdquoHsiao 1993a)

The E versions were developed in consultation with native-English-speaking faculty at a US university The second author revised thepassages to cue the organization of information in line with Westernrhetorical conventions Revisions involved both the sequencing of tex-tual chunks and the cueing of the ideas with discourse markers (fordetailed descriptions of this process and the resultant texts see Chu1999 pp 137ndash159) The sequencing principles were1 putting the topic or commentthesis and argument at the beginning

of the essay2 putting background information about the topic at the beginning of

the essay3 modifying general statements so that the argument changed from

inductive to deductive and was explicitly marked as such

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

512 TESOL QUARTERLY

to the reader may hinge in part on familiarity or unfamiliarity with atextrsquos communicative styles (Eggington 1987 Hinds 1984 Kaplan 1966Ricento 1989) To address this issue the present study looked at whetherrhetorical style in and of itself affected the recall of Chinese studentsreading otherwise identical English language texts

RESEARCH ON EFFECTS OFRHETORICAL CONVENTIONS

Models of Reading Comprehension

Two of the most widely accepted models of reading comprehensionnotably those of Rumelhart (1977) and Kintsch (1998) agree thatcomprehension is a multicomponential interactive process One resultof such interactivity seems to be what Stanovich (1980) calls compensatoryprocessing ldquoa de cit in any knowledge source [that] results in a heavierreliance on other knowledge sources regardless of their level in theprocessing hierarchyrdquo (p 63)

For Kintsch (1998) the product of the comprehension can be a textbase model a situation model or something in between depending ontask demands (p 292) In this construction-integration model readersassemble textual units in two ways as recall and as situations applicableto their lives In either case when they experience problems in puttingtextual information into a coherent pattern readers call on generalknowledge and draw inferences elaborating on the actual text baseKintsch emphasizes that although integrating prior knowledge withtextual propositions still in working memory is automatic it takes upmental capacity Moreover because the integration process allows for allsorts of knowledge to come into play one feature of that process spreadactivation strengthens related items and suppresses unrelated ones (pp98ndash99)

The result is a text base that structures mental representation ofmicro- and macropropositions hierarchically These propositions repre-sent the readerrsquos recall of corresponding text and they enable reproduc-tion tasks such as recall and summary In the situation model theinformation carried in the text has been modi ed and embedded into areaderrsquos existing knowledge structure resulting in learning Althoughtraces of the original propositions and structure of text have been lost(eg a reader might say ldquoI canrsquot remember where I heardread thatrdquo)the macrostructure or rhetorical logic the author has chosen in uencesthe readerrsquos original processing (Weaver amp Kintsch 1991)

Kintschrsquos (1998) insights suggest that the macrostructures or rhetori-

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 513

cal conventions in the text are not only vital to textual comprehensionbut also essential for the readersrsquo intake of information and possiblereconstruction of the text Macrostructures do more than inform con-struction of macropropositions They also enable retrieval of proposi-tions and their integration into long-term memory (van Dijk amp Kintsch1983) When readers process a text with unfamiliar macrostructures thisunfamiliarity might in uence the construction of both text base and textsituation Whether the task facing readers demands recall or applicationde ciencies in their knowledge of rhetorical conventionsmdashthe macro-structures that reveal textual organizationmdashcould overtax readersrsquo syn-thesizing capabilities

Another way in which Kintsch (1998) speaks to the signi cance ofrhetorical conventions is by emphasizing that readers with a sense ofcoherence different from that suggested in the discourse features of thetext might be connecting propositions in ways different from thoseintended by the author Schnotzrsquos (1984) subjects who read two textswith the same content but contrasting thematic relationships seem toillustrate Kintschrsquos assertion that readers introduce inferential changesto the text base when its content is reorganized Schnotz describes thesedifferences as two distinct types of organization from which a reader maychoose The processes resulting from that choice will result in ldquodifferentknowledge structures for both organization types although the textcontent may be the samerdquo (p 71) Similarly Goetz and Armbruster(1980) conclude that connected discourse is easier to comprehend thanunrelated and disconnected content because the reader is able toldquoorganize and interrelate elements in the textrdquo (p 206) They suggestthat the textrsquos logical structure when perceived by the reader facilitatesthat readerrsquos schematizing process The reader is consequently able toanticipate text content and construct meaning in a relational frameworkthereby retaining content in memory as an integrated story rather thanas disconnected pieces

Studies Examining the Impact ofRhetorical Structure on L2 Reading

L1 research has established that whenever a mismatch between textualorganization and reader expectations occurs readers tend to distort atextrsquos meaning their processing ef ciency is impaired and their reten-tion is short-circuited (Brennan Connie amp Winograd 1986 Kintsch ampYarbrough 1982 Meyer Brandt amp Bluth 1980) With regard to L2readers research on the impact of rhetorical structure sometimesreferred to as formal schema (eg Barnitz 1986 Carrell 1984) also

514 TESOL QUARTERLY

supports the claim that familiarity with rhetorical conventions plays arole in recall However most studies on text structure in L2 reading havefocused on how the logical organization of textual information typical inWestern writing in uences text processing and recall (Carrell 1984)Tian (1990) replicating Carrellrsquos 1984 study in Singapore introducedthe additional variable of home language group (Chinese Malay orTamil) As in Carrellrsquos study differences in rhetorical structure affectedthe readersrsquo recall Their native language however did not seem totrigger a different recall pattern for different rhetorical structures Tianconjectured that the homogeneity of the language environment inschools in Singapore leveled and neutralized the home language effectsHowever Carrell (1992) found a relationship between her ESL ChineseJapanese Arabic Malaysian Turkish Spanish and Korean readersrsquoawareness of text structure and their recall

Considered as a whole most studies on the effects of text structurehave been conducted in an ESL context with a focus on the effects oflogical organization rather than on the effects of rhetorical conventionviewed as a cultural phenomenon Whether L2 readersrsquo experience witha culture-speci c L1 rhetoric plays a role in L2 reading among readers atdifferent pro ciency levels has yet to be explored Work comparingstudentsrsquo background knowledge and interest in a topic (Carrell amp Wise1998) and the interaction between content schemata and formal sche-mata (Carrell 1987) points in this direction For this reason the presentstudy focuses on potential differences in recall of texts re ecting Westernand non-Western rhetorical conventions

Additionally many studies of L2 reading comprehension suggest apositive relationship between increased language pro ciency and abilityto recognize discourse cues (eg Cooper 1984) thus avoiding the short-circuits that are more common among readers with de cient languageskills (Clarke 1980 Cummins 1979) Therefore this study includedboth college freshmen and seniors who were English majors

RHETORICAL STRUCTURE IN CHINESEEXPOSITORY WRITING

Qi-Cheng-Zhuan-He

An approach to Asiatic writing qi-cheng-zhuan-he (ki-shoAring -ten-ketsu inJapanese and ki-sung-chon-kyual in Korean) is a prevalent though not theonly model used in analysis of Chinese texts It was originally employedby Fan-heng a scholar in the Yuen Dynasty to analyze poetry and waslater applied to structuring essays Though relevant for narrative and

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 515

poetry as well as public speaking it is most frequently applied inexpository writing (Kojima 1972) Literally qi means beginningmdashtheopening of a topic Cheng means followingmdashelaborating the openingZhuan means turning or changing Here the writer expresses anotherpoint of view turns to an example or develops the idea further Hemeans wrapping up when the writer provides the highest level ofgeneralization

Qi-cheng-zhuan-he is not a rigidly formalized structure the number ofwords and paragraphs as well as the amount of information devoted toeach of the four parts can vary disproportionately As long as the order ofpresentation remains the essence of the structure is there (Kojima1972) Hence qi-cheng-zhuan-he designates a conceptualreasoning se-quence in overall organization

Because qi-cheng-zhuan-he is simply one model for rhetorical sequenc-ing texts that use this model also exhibit other features commonly foundin Chinese writing These features differ markedly from those commonin Western writing (for examples see Appendix A) A tally of suchdifferences frequently starts with the Western preference for deductivestyle and the Chinese tendency toward inductive approaches particu-larly in expository writing The practice of putting the main thesis of atext before supporting ideas violates a Chinese readerrsquos expectation forwhat Kachru (1998) calls a ldquodelayed introduction of purposerdquo orldquodelayed topic statementrdquo (p 55)

Chen (1986) in comparing the expository discourse structure ofEnglish and Chinese passages found that Chinese paragraphs tended tointroduce more subtopics than did their Western counterparts Kirkpatrick(1995) quoting Wu (1988) suggests how inductive reasoning mightcondition Chinese rhetorical schemata Wu posits that inductive reason-ing is felt to be ldquonaturalrdquo whereas deductive reasoning requires aninformation sequence that goes against ldquoa natural sequencerdquo (p 281) asperceived by Chinese readers

A second difference researchers have noted between Western andChinese rhetorical conventions is closely related to the way main andsubordinate information is sequenced in each Discourse cues forsentential and intersentential cohesion tend not to occur as often inChinese as they do in English writing (Normant 1986) Further com-pounding this Chinese tendency toward indirect rather than directpresentation of authorial intent Jensen (1998) stated is that subtletyanalogy and metaphor highly valued writing techniques in classicalChinese tend to hide the writerrsquos intents He attributed this tendencytoward indirect allusion to the Chinese cultural heritage in rhetoric

As a result of these differences researchers have predicted thatChinese EFL readers because they have rhetorical expectations thatdiffer from those of Western readers might experience problems

516 TESOL QUARTERLY

comprehending a passage written with Western rhetorical conventionsMain ideas for example might not be recognized as such due to theirearly presentation and thus may fail to facilitate comprehension

Consequently Chinese EFL readers could be at a disadvantage whenreading English texts written by native speakers of English EFL readerswho predict a Chinese discourse structure may nd a mismatch betweentheir expectations and the macrostructures (the organization of contentand the sequence of ideas) presented in the Western text This gapbetween the Chinese readerrsquos anticipation of text development and thetextrsquos actual macrostrutures might result in lower comprehension andretention of an English text written in an English rhetorical structureConversely a text written in the rhetorical structure common in Chi-nesemdashthat is according to conventions such as those of qi-cheng-zhuan-he mdashmight facilitate recall for Chinese readers by providing them withfamiliar organizational macrostructures

The Role of Rhetorical Conventions inTaiwanese High Schools

In the past Taiwanese studentsrsquo exposure to English reading andwriting before they entered college focused more on the lexical andsyntactic features of a passage than on its rhetorical conventions Theirreading in Chinese on the other hand frequently dealt with theschemata of Chinese rhetorical conventions In terms of expositorywriting in their native language the classical Chinese rhetorical struc-ture qi-cheng-zhuan-he exerts a signi cant in uence as a macrostructurefor text analysis

For the students who participated in this study work with Englishlanguage texts did not emphasize the macrostructures of those passagesUntil 1999 when competitive marketing of textbooks was introducedthe traditional English textbooks commonly used in Taiwanese highschools offered little explicit training in the rhetorical organization oftexts Although reading formed the core component of these textbooksexercises for reading passages emphasized word- and sentence-levelmeanings rather than rhetorical organization Most of the readings werewritten originally by native speakers of English but were adapted oredited to control for length and to simplify vocabulary and syntax

The reading aids that commonly accompanied the reading passages inthe traditional textbooks were phonetic transcription and a glossary ofvocabulary translation and interpretation of idioms and phrases Cover-age of the rhetorical aspect of the reading was limited Reading andwriting exercises typically consisted of cloze vocabulary matching phrase

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 517

substitution translation questions on factual content and essay ques-tions on content exercises only rarely focused on the development ofideas and the discourse markers signaling the textual organization ofthose ideas Importantly for this study these textbooks did not explicitlycall studentsrsquo attention to rhetorical structures common in Englishwriting

In reading in Chinese on the other hand rhetorical structures play asigni cant role in secondary school instruction High school studentsrsquocourses in Chinese involve intensive reading of classical Chinese andmemorization of classical texts as part of the literary tradition A keypractice is the paraphrasing of classical texts into modern ChineseBecause text reproduction tasks require the construction of a text basestudents must inevitably attend to the global structure of the textconsciously or subconsciously in the retention and retrieval processesPossibly then students in this study developed robust text-structureschemata based on rhetorical conventions often found in Chinesewriting such as the above-mentioned qi-cheng-zhuan-he

The English Majorrsquos Literacy Experience at College in Taiwan

In Taiwan English majors such as those who participated in this studyenroll in content courses in linguistics and English literature in whichthey use collections such as The Norton Anthology of American Literature(Abrams 1996) that have been compiled for native-English-speakingstudents In the freshman year most courses focus on language skills afew are introductory courses in linguistics and literature As studentsmove on to the sophomore junior and senior years the proportion oflanguage skill courses decreases and content courses in linguistics andliterature increase Because this study was conducted while the freshmenwere in their rst semester and the seniors in their seventh semester thefreshmenrsquos exposure to authentic texts was much more limited than thatof the seniors

Regardless of grade level the texts in linguistics and literature thatEnglish majors read in college are very different from those they read inhigh school English classes The readings are considerably longer andmore demanding in content and style To increase their reading speedand their ability to cope with large quantities of EFL reading Englishmajors must either resort to translations or shift their reading style to amore macrotextual level

Whether reading English texts in British or US anthologies orChinese translations of those texts English majors in Taiwan are readinginformation presented in English rhetorical structure After 3 years ofsuch exposure students in their senior year may well be able to process

518 TESOL QUARTERLY

these texts as ef ciently as they can a text in Chinese rhetorical structureparticularly if the passage in question is expository or presents theauthorrsquos point of view typical for a classical essay written in the style of qi-cheng-zhuan-he No research has ever been conducted to test thatassumption

Chinese rhetorical structures found in the commonly used styles ofexpository texts contrast sharply with the rhetorical premises of Westernwriting Consequently the impact of culture-speci c rhetorical structureson the reading comprehension of Chinese students may be particularlyprofound for this type of writing Although researchers have looked atlanguage pro ciency (Carrell 1991) and text structure (Carrell 1992) asvariables to our knowledge no study has looked at both simultaneously1

This study examined the recall of readers at two different learning levelsmdashfreshmen and seniorsmdashwho read texts having identical content butrepresenting two distinct culture-speci c rhetorical conventions Thestudy poses the following questions (a) Does reading an L2 text thatfollows L1 rhetorical conventions affect EFL studentsrsquo recall and percep-tions of the text differently from reading the same text in L2 conventions(b) Do effects of rhetorical convention depend on the readerrsquos grade level

METHOD

Participants

To establish text selection criteria and choose suitable measurementsfor the main study we conducted a pilot study in June 1998 withfreshman English majors at Providence University a private university inTaiwan A total of 240 English majors at the same university participatedin the main study on a voluntary basis in December 1998 Half of theparticipants (98 females and 22 males) were recruited from fourfreshman sections and half (102 females and 18 males) were from foursenior classes The freshmen had had at least 6 years of formal EFLlanguage education in Taiwan which as noted above focused more onlanguage features than on the rhetorical aspects of English and includedrelatively limited access to extensive reading of naturally occurringEnglish prose written by Western authors for Western readers

In contrast the seniors had had 3 years of extensive exposure toEnglish texts written for native speakers of English in linguistics and

1 Work on the effects of rhetorical convention on reading comprehension has been done inother Asian languages notably Korean (Eggington 1987) and Japanese (Hinds 1987 Ricento1989)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 519

literature courses Presumably then the seniors had not only greater L2language pro ciency than did the freshmen but also considerably moreexposure to English rhetorical style Both ESL groups had had a formalMandarin Chinese education with a focus on classical Chinese while inhigh school and college and thus had been exposed to a rich L1environment with texts featuring traditional rhetorical conventions

Materials

Reading Passages

Four passages in English that followed Chinese rhetorical conventions(hereafter C versions) were selected from a Taiwanese bilingual magazineand modi ed to conform to English rhetorical conventions (hereafter Eversions) All the passages (Hsiao 1993a 1993b 1993c 1994) wereeditorials originally written in Chinese by a Chinese author and trans-lated into English for the magazine Editorials were chosen because thisgenre tends to use conventions that re ect features of qi-cheng-zhuan-heThe passages were judged as representative of qi-cheng-zhuan-he by sevenprofessors in the Department of Chinese Literature at a private universityin Taiwan

The four passages dealt with sociological issues of contemporaryTaiwanese society the unintended consequences of childcare for moth-ers with careers (ldquoChildcarerdquo Hsiao 1993c) the crisis one charityhospital in Taiwan faced because nonindigent patients abused its re-sources (ldquoCharityrdquo Hsiao 1993b) gender issues for male nurses inTaiwan (ldquoMale Nursesrdquo Hsiao 1994) and the legal problems facingparents who wanted to start alternative schools in Taiwan (ldquoSchoolingrdquoHsiao 1993a)

The E versions were developed in consultation with native-English-speaking faculty at a US university The second author revised thepassages to cue the organization of information in line with Westernrhetorical conventions Revisions involved both the sequencing of tex-tual chunks and the cueing of the ideas with discourse markers (fordetailed descriptions of this process and the resultant texts see Chu1999 pp 137ndash159) The sequencing principles were1 putting the topic or commentthesis and argument at the beginning

of the essay2 putting background information about the topic at the beginning of

the essay3 modifying general statements so that the argument changed from

inductive to deductive and was explicitly marked as such

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 513

cal conventions in the text are not only vital to textual comprehensionbut also essential for the readersrsquo intake of information and possiblereconstruction of the text Macrostructures do more than inform con-struction of macropropositions They also enable retrieval of proposi-tions and their integration into long-term memory (van Dijk amp Kintsch1983) When readers process a text with unfamiliar macrostructures thisunfamiliarity might in uence the construction of both text base and textsituation Whether the task facing readers demands recall or applicationde ciencies in their knowledge of rhetorical conventionsmdashthe macro-structures that reveal textual organizationmdashcould overtax readersrsquo syn-thesizing capabilities

Another way in which Kintsch (1998) speaks to the signi cance ofrhetorical conventions is by emphasizing that readers with a sense ofcoherence different from that suggested in the discourse features of thetext might be connecting propositions in ways different from thoseintended by the author Schnotzrsquos (1984) subjects who read two textswith the same content but contrasting thematic relationships seem toillustrate Kintschrsquos assertion that readers introduce inferential changesto the text base when its content is reorganized Schnotz describes thesedifferences as two distinct types of organization from which a reader maychoose The processes resulting from that choice will result in ldquodifferentknowledge structures for both organization types although the textcontent may be the samerdquo (p 71) Similarly Goetz and Armbruster(1980) conclude that connected discourse is easier to comprehend thanunrelated and disconnected content because the reader is able toldquoorganize and interrelate elements in the textrdquo (p 206) They suggestthat the textrsquos logical structure when perceived by the reader facilitatesthat readerrsquos schematizing process The reader is consequently able toanticipate text content and construct meaning in a relational frameworkthereby retaining content in memory as an integrated story rather thanas disconnected pieces

Studies Examining the Impact ofRhetorical Structure on L2 Reading

L1 research has established that whenever a mismatch between textualorganization and reader expectations occurs readers tend to distort atextrsquos meaning their processing ef ciency is impaired and their reten-tion is short-circuited (Brennan Connie amp Winograd 1986 Kintsch ampYarbrough 1982 Meyer Brandt amp Bluth 1980) With regard to L2readers research on the impact of rhetorical structure sometimesreferred to as formal schema (eg Barnitz 1986 Carrell 1984) also

514 TESOL QUARTERLY

supports the claim that familiarity with rhetorical conventions plays arole in recall However most studies on text structure in L2 reading havefocused on how the logical organization of textual information typical inWestern writing in uences text processing and recall (Carrell 1984)Tian (1990) replicating Carrellrsquos 1984 study in Singapore introducedthe additional variable of home language group (Chinese Malay orTamil) As in Carrellrsquos study differences in rhetorical structure affectedthe readersrsquo recall Their native language however did not seem totrigger a different recall pattern for different rhetorical structures Tianconjectured that the homogeneity of the language environment inschools in Singapore leveled and neutralized the home language effectsHowever Carrell (1992) found a relationship between her ESL ChineseJapanese Arabic Malaysian Turkish Spanish and Korean readersrsquoawareness of text structure and their recall

Considered as a whole most studies on the effects of text structurehave been conducted in an ESL context with a focus on the effects oflogical organization rather than on the effects of rhetorical conventionviewed as a cultural phenomenon Whether L2 readersrsquo experience witha culture-speci c L1 rhetoric plays a role in L2 reading among readers atdifferent pro ciency levels has yet to be explored Work comparingstudentsrsquo background knowledge and interest in a topic (Carrell amp Wise1998) and the interaction between content schemata and formal sche-mata (Carrell 1987) points in this direction For this reason the presentstudy focuses on potential differences in recall of texts re ecting Westernand non-Western rhetorical conventions

Additionally many studies of L2 reading comprehension suggest apositive relationship between increased language pro ciency and abilityto recognize discourse cues (eg Cooper 1984) thus avoiding the short-circuits that are more common among readers with de cient languageskills (Clarke 1980 Cummins 1979) Therefore this study includedboth college freshmen and seniors who were English majors

RHETORICAL STRUCTURE IN CHINESEEXPOSITORY WRITING

Qi-Cheng-Zhuan-He

An approach to Asiatic writing qi-cheng-zhuan-he (ki-shoAring -ten-ketsu inJapanese and ki-sung-chon-kyual in Korean) is a prevalent though not theonly model used in analysis of Chinese texts It was originally employedby Fan-heng a scholar in the Yuen Dynasty to analyze poetry and waslater applied to structuring essays Though relevant for narrative and

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 515

poetry as well as public speaking it is most frequently applied inexpository writing (Kojima 1972) Literally qi means beginningmdashtheopening of a topic Cheng means followingmdashelaborating the openingZhuan means turning or changing Here the writer expresses anotherpoint of view turns to an example or develops the idea further Hemeans wrapping up when the writer provides the highest level ofgeneralization

Qi-cheng-zhuan-he is not a rigidly formalized structure the number ofwords and paragraphs as well as the amount of information devoted toeach of the four parts can vary disproportionately As long as the order ofpresentation remains the essence of the structure is there (Kojima1972) Hence qi-cheng-zhuan-he designates a conceptualreasoning se-quence in overall organization

Because qi-cheng-zhuan-he is simply one model for rhetorical sequenc-ing texts that use this model also exhibit other features commonly foundin Chinese writing These features differ markedly from those commonin Western writing (for examples see Appendix A) A tally of suchdifferences frequently starts with the Western preference for deductivestyle and the Chinese tendency toward inductive approaches particu-larly in expository writing The practice of putting the main thesis of atext before supporting ideas violates a Chinese readerrsquos expectation forwhat Kachru (1998) calls a ldquodelayed introduction of purposerdquo orldquodelayed topic statementrdquo (p 55)

Chen (1986) in comparing the expository discourse structure ofEnglish and Chinese passages found that Chinese paragraphs tended tointroduce more subtopics than did their Western counterparts Kirkpatrick(1995) quoting Wu (1988) suggests how inductive reasoning mightcondition Chinese rhetorical schemata Wu posits that inductive reason-ing is felt to be ldquonaturalrdquo whereas deductive reasoning requires aninformation sequence that goes against ldquoa natural sequencerdquo (p 281) asperceived by Chinese readers

A second difference researchers have noted between Western andChinese rhetorical conventions is closely related to the way main andsubordinate information is sequenced in each Discourse cues forsentential and intersentential cohesion tend not to occur as often inChinese as they do in English writing (Normant 1986) Further com-pounding this Chinese tendency toward indirect rather than directpresentation of authorial intent Jensen (1998) stated is that subtletyanalogy and metaphor highly valued writing techniques in classicalChinese tend to hide the writerrsquos intents He attributed this tendencytoward indirect allusion to the Chinese cultural heritage in rhetoric

As a result of these differences researchers have predicted thatChinese EFL readers because they have rhetorical expectations thatdiffer from those of Western readers might experience problems

516 TESOL QUARTERLY

comprehending a passage written with Western rhetorical conventionsMain ideas for example might not be recognized as such due to theirearly presentation and thus may fail to facilitate comprehension

Consequently Chinese EFL readers could be at a disadvantage whenreading English texts written by native speakers of English EFL readerswho predict a Chinese discourse structure may nd a mismatch betweentheir expectations and the macrostructures (the organization of contentand the sequence of ideas) presented in the Western text This gapbetween the Chinese readerrsquos anticipation of text development and thetextrsquos actual macrostrutures might result in lower comprehension andretention of an English text written in an English rhetorical structureConversely a text written in the rhetorical structure common in Chi-nesemdashthat is according to conventions such as those of qi-cheng-zhuan-he mdashmight facilitate recall for Chinese readers by providing them withfamiliar organizational macrostructures

The Role of Rhetorical Conventions inTaiwanese High Schools

In the past Taiwanese studentsrsquo exposure to English reading andwriting before they entered college focused more on the lexical andsyntactic features of a passage than on its rhetorical conventions Theirreading in Chinese on the other hand frequently dealt with theschemata of Chinese rhetorical conventions In terms of expositorywriting in their native language the classical Chinese rhetorical struc-ture qi-cheng-zhuan-he exerts a signi cant in uence as a macrostructurefor text analysis

For the students who participated in this study work with Englishlanguage texts did not emphasize the macrostructures of those passagesUntil 1999 when competitive marketing of textbooks was introducedthe traditional English textbooks commonly used in Taiwanese highschools offered little explicit training in the rhetorical organization oftexts Although reading formed the core component of these textbooksexercises for reading passages emphasized word- and sentence-levelmeanings rather than rhetorical organization Most of the readings werewritten originally by native speakers of English but were adapted oredited to control for length and to simplify vocabulary and syntax

The reading aids that commonly accompanied the reading passages inthe traditional textbooks were phonetic transcription and a glossary ofvocabulary translation and interpretation of idioms and phrases Cover-age of the rhetorical aspect of the reading was limited Reading andwriting exercises typically consisted of cloze vocabulary matching phrase

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 517

substitution translation questions on factual content and essay ques-tions on content exercises only rarely focused on the development ofideas and the discourse markers signaling the textual organization ofthose ideas Importantly for this study these textbooks did not explicitlycall studentsrsquo attention to rhetorical structures common in Englishwriting

In reading in Chinese on the other hand rhetorical structures play asigni cant role in secondary school instruction High school studentsrsquocourses in Chinese involve intensive reading of classical Chinese andmemorization of classical texts as part of the literary tradition A keypractice is the paraphrasing of classical texts into modern ChineseBecause text reproduction tasks require the construction of a text basestudents must inevitably attend to the global structure of the textconsciously or subconsciously in the retention and retrieval processesPossibly then students in this study developed robust text-structureschemata based on rhetorical conventions often found in Chinesewriting such as the above-mentioned qi-cheng-zhuan-he

The English Majorrsquos Literacy Experience at College in Taiwan

In Taiwan English majors such as those who participated in this studyenroll in content courses in linguistics and English literature in whichthey use collections such as The Norton Anthology of American Literature(Abrams 1996) that have been compiled for native-English-speakingstudents In the freshman year most courses focus on language skills afew are introductory courses in linguistics and literature As studentsmove on to the sophomore junior and senior years the proportion oflanguage skill courses decreases and content courses in linguistics andliterature increase Because this study was conducted while the freshmenwere in their rst semester and the seniors in their seventh semester thefreshmenrsquos exposure to authentic texts was much more limited than thatof the seniors

Regardless of grade level the texts in linguistics and literature thatEnglish majors read in college are very different from those they read inhigh school English classes The readings are considerably longer andmore demanding in content and style To increase their reading speedand their ability to cope with large quantities of EFL reading Englishmajors must either resort to translations or shift their reading style to amore macrotextual level

Whether reading English texts in British or US anthologies orChinese translations of those texts English majors in Taiwan are readinginformation presented in English rhetorical structure After 3 years ofsuch exposure students in their senior year may well be able to process

518 TESOL QUARTERLY

these texts as ef ciently as they can a text in Chinese rhetorical structureparticularly if the passage in question is expository or presents theauthorrsquos point of view typical for a classical essay written in the style of qi-cheng-zhuan-he No research has ever been conducted to test thatassumption

Chinese rhetorical structures found in the commonly used styles ofexpository texts contrast sharply with the rhetorical premises of Westernwriting Consequently the impact of culture-speci c rhetorical structureson the reading comprehension of Chinese students may be particularlyprofound for this type of writing Although researchers have looked atlanguage pro ciency (Carrell 1991) and text structure (Carrell 1992) asvariables to our knowledge no study has looked at both simultaneously1

This study examined the recall of readers at two different learning levelsmdashfreshmen and seniorsmdashwho read texts having identical content butrepresenting two distinct culture-speci c rhetorical conventions Thestudy poses the following questions (a) Does reading an L2 text thatfollows L1 rhetorical conventions affect EFL studentsrsquo recall and percep-tions of the text differently from reading the same text in L2 conventions(b) Do effects of rhetorical convention depend on the readerrsquos grade level

METHOD

Participants

To establish text selection criteria and choose suitable measurementsfor the main study we conducted a pilot study in June 1998 withfreshman English majors at Providence University a private university inTaiwan A total of 240 English majors at the same university participatedin the main study on a voluntary basis in December 1998 Half of theparticipants (98 females and 22 males) were recruited from fourfreshman sections and half (102 females and 18 males) were from foursenior classes The freshmen had had at least 6 years of formal EFLlanguage education in Taiwan which as noted above focused more onlanguage features than on the rhetorical aspects of English and includedrelatively limited access to extensive reading of naturally occurringEnglish prose written by Western authors for Western readers

In contrast the seniors had had 3 years of extensive exposure toEnglish texts written for native speakers of English in linguistics and

1 Work on the effects of rhetorical convention on reading comprehension has been done inother Asian languages notably Korean (Eggington 1987) and Japanese (Hinds 1987 Ricento1989)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 519

literature courses Presumably then the seniors had not only greater L2language pro ciency than did the freshmen but also considerably moreexposure to English rhetorical style Both ESL groups had had a formalMandarin Chinese education with a focus on classical Chinese while inhigh school and college and thus had been exposed to a rich L1environment with texts featuring traditional rhetorical conventions

Materials

Reading Passages

Four passages in English that followed Chinese rhetorical conventions(hereafter C versions) were selected from a Taiwanese bilingual magazineand modi ed to conform to English rhetorical conventions (hereafter Eversions) All the passages (Hsiao 1993a 1993b 1993c 1994) wereeditorials originally written in Chinese by a Chinese author and trans-lated into English for the magazine Editorials were chosen because thisgenre tends to use conventions that re ect features of qi-cheng-zhuan-heThe passages were judged as representative of qi-cheng-zhuan-he by sevenprofessors in the Department of Chinese Literature at a private universityin Taiwan

The four passages dealt with sociological issues of contemporaryTaiwanese society the unintended consequences of childcare for moth-ers with careers (ldquoChildcarerdquo Hsiao 1993c) the crisis one charityhospital in Taiwan faced because nonindigent patients abused its re-sources (ldquoCharityrdquo Hsiao 1993b) gender issues for male nurses inTaiwan (ldquoMale Nursesrdquo Hsiao 1994) and the legal problems facingparents who wanted to start alternative schools in Taiwan (ldquoSchoolingrdquoHsiao 1993a)

The E versions were developed in consultation with native-English-speaking faculty at a US university The second author revised thepassages to cue the organization of information in line with Westernrhetorical conventions Revisions involved both the sequencing of tex-tual chunks and the cueing of the ideas with discourse markers (fordetailed descriptions of this process and the resultant texts see Chu1999 pp 137ndash159) The sequencing principles were1 putting the topic or commentthesis and argument at the beginning

of the essay2 putting background information about the topic at the beginning of

the essay3 modifying general statements so that the argument changed from

inductive to deductive and was explicitly marked as such

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

514 TESOL QUARTERLY

supports the claim that familiarity with rhetorical conventions plays arole in recall However most studies on text structure in L2 reading havefocused on how the logical organization of textual information typical inWestern writing in uences text processing and recall (Carrell 1984)Tian (1990) replicating Carrellrsquos 1984 study in Singapore introducedthe additional variable of home language group (Chinese Malay orTamil) As in Carrellrsquos study differences in rhetorical structure affectedthe readersrsquo recall Their native language however did not seem totrigger a different recall pattern for different rhetorical structures Tianconjectured that the homogeneity of the language environment inschools in Singapore leveled and neutralized the home language effectsHowever Carrell (1992) found a relationship between her ESL ChineseJapanese Arabic Malaysian Turkish Spanish and Korean readersrsquoawareness of text structure and their recall

Considered as a whole most studies on the effects of text structurehave been conducted in an ESL context with a focus on the effects oflogical organization rather than on the effects of rhetorical conventionviewed as a cultural phenomenon Whether L2 readersrsquo experience witha culture-speci c L1 rhetoric plays a role in L2 reading among readers atdifferent pro ciency levels has yet to be explored Work comparingstudentsrsquo background knowledge and interest in a topic (Carrell amp Wise1998) and the interaction between content schemata and formal sche-mata (Carrell 1987) points in this direction For this reason the presentstudy focuses on potential differences in recall of texts re ecting Westernand non-Western rhetorical conventions

Additionally many studies of L2 reading comprehension suggest apositive relationship between increased language pro ciency and abilityto recognize discourse cues (eg Cooper 1984) thus avoiding the short-circuits that are more common among readers with de cient languageskills (Clarke 1980 Cummins 1979) Therefore this study includedboth college freshmen and seniors who were English majors

RHETORICAL STRUCTURE IN CHINESEEXPOSITORY WRITING

Qi-Cheng-Zhuan-He

An approach to Asiatic writing qi-cheng-zhuan-he (ki-shoAring -ten-ketsu inJapanese and ki-sung-chon-kyual in Korean) is a prevalent though not theonly model used in analysis of Chinese texts It was originally employedby Fan-heng a scholar in the Yuen Dynasty to analyze poetry and waslater applied to structuring essays Though relevant for narrative and

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 515

poetry as well as public speaking it is most frequently applied inexpository writing (Kojima 1972) Literally qi means beginningmdashtheopening of a topic Cheng means followingmdashelaborating the openingZhuan means turning or changing Here the writer expresses anotherpoint of view turns to an example or develops the idea further Hemeans wrapping up when the writer provides the highest level ofgeneralization

Qi-cheng-zhuan-he is not a rigidly formalized structure the number ofwords and paragraphs as well as the amount of information devoted toeach of the four parts can vary disproportionately As long as the order ofpresentation remains the essence of the structure is there (Kojima1972) Hence qi-cheng-zhuan-he designates a conceptualreasoning se-quence in overall organization

Because qi-cheng-zhuan-he is simply one model for rhetorical sequenc-ing texts that use this model also exhibit other features commonly foundin Chinese writing These features differ markedly from those commonin Western writing (for examples see Appendix A) A tally of suchdifferences frequently starts with the Western preference for deductivestyle and the Chinese tendency toward inductive approaches particu-larly in expository writing The practice of putting the main thesis of atext before supporting ideas violates a Chinese readerrsquos expectation forwhat Kachru (1998) calls a ldquodelayed introduction of purposerdquo orldquodelayed topic statementrdquo (p 55)

Chen (1986) in comparing the expository discourse structure ofEnglish and Chinese passages found that Chinese paragraphs tended tointroduce more subtopics than did their Western counterparts Kirkpatrick(1995) quoting Wu (1988) suggests how inductive reasoning mightcondition Chinese rhetorical schemata Wu posits that inductive reason-ing is felt to be ldquonaturalrdquo whereas deductive reasoning requires aninformation sequence that goes against ldquoa natural sequencerdquo (p 281) asperceived by Chinese readers

A second difference researchers have noted between Western andChinese rhetorical conventions is closely related to the way main andsubordinate information is sequenced in each Discourse cues forsentential and intersentential cohesion tend not to occur as often inChinese as they do in English writing (Normant 1986) Further com-pounding this Chinese tendency toward indirect rather than directpresentation of authorial intent Jensen (1998) stated is that subtletyanalogy and metaphor highly valued writing techniques in classicalChinese tend to hide the writerrsquos intents He attributed this tendencytoward indirect allusion to the Chinese cultural heritage in rhetoric

As a result of these differences researchers have predicted thatChinese EFL readers because they have rhetorical expectations thatdiffer from those of Western readers might experience problems

516 TESOL QUARTERLY

comprehending a passage written with Western rhetorical conventionsMain ideas for example might not be recognized as such due to theirearly presentation and thus may fail to facilitate comprehension

Consequently Chinese EFL readers could be at a disadvantage whenreading English texts written by native speakers of English EFL readerswho predict a Chinese discourse structure may nd a mismatch betweentheir expectations and the macrostructures (the organization of contentand the sequence of ideas) presented in the Western text This gapbetween the Chinese readerrsquos anticipation of text development and thetextrsquos actual macrostrutures might result in lower comprehension andretention of an English text written in an English rhetorical structureConversely a text written in the rhetorical structure common in Chi-nesemdashthat is according to conventions such as those of qi-cheng-zhuan-he mdashmight facilitate recall for Chinese readers by providing them withfamiliar organizational macrostructures

The Role of Rhetorical Conventions inTaiwanese High Schools

In the past Taiwanese studentsrsquo exposure to English reading andwriting before they entered college focused more on the lexical andsyntactic features of a passage than on its rhetorical conventions Theirreading in Chinese on the other hand frequently dealt with theschemata of Chinese rhetorical conventions In terms of expositorywriting in their native language the classical Chinese rhetorical struc-ture qi-cheng-zhuan-he exerts a signi cant in uence as a macrostructurefor text analysis

For the students who participated in this study work with Englishlanguage texts did not emphasize the macrostructures of those passagesUntil 1999 when competitive marketing of textbooks was introducedthe traditional English textbooks commonly used in Taiwanese highschools offered little explicit training in the rhetorical organization oftexts Although reading formed the core component of these textbooksexercises for reading passages emphasized word- and sentence-levelmeanings rather than rhetorical organization Most of the readings werewritten originally by native speakers of English but were adapted oredited to control for length and to simplify vocabulary and syntax

The reading aids that commonly accompanied the reading passages inthe traditional textbooks were phonetic transcription and a glossary ofvocabulary translation and interpretation of idioms and phrases Cover-age of the rhetorical aspect of the reading was limited Reading andwriting exercises typically consisted of cloze vocabulary matching phrase

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 517

substitution translation questions on factual content and essay ques-tions on content exercises only rarely focused on the development ofideas and the discourse markers signaling the textual organization ofthose ideas Importantly for this study these textbooks did not explicitlycall studentsrsquo attention to rhetorical structures common in Englishwriting

In reading in Chinese on the other hand rhetorical structures play asigni cant role in secondary school instruction High school studentsrsquocourses in Chinese involve intensive reading of classical Chinese andmemorization of classical texts as part of the literary tradition A keypractice is the paraphrasing of classical texts into modern ChineseBecause text reproduction tasks require the construction of a text basestudents must inevitably attend to the global structure of the textconsciously or subconsciously in the retention and retrieval processesPossibly then students in this study developed robust text-structureschemata based on rhetorical conventions often found in Chinesewriting such as the above-mentioned qi-cheng-zhuan-he

The English Majorrsquos Literacy Experience at College in Taiwan

In Taiwan English majors such as those who participated in this studyenroll in content courses in linguistics and English literature in whichthey use collections such as The Norton Anthology of American Literature(Abrams 1996) that have been compiled for native-English-speakingstudents In the freshman year most courses focus on language skills afew are introductory courses in linguistics and literature As studentsmove on to the sophomore junior and senior years the proportion oflanguage skill courses decreases and content courses in linguistics andliterature increase Because this study was conducted while the freshmenwere in their rst semester and the seniors in their seventh semester thefreshmenrsquos exposure to authentic texts was much more limited than thatof the seniors

Regardless of grade level the texts in linguistics and literature thatEnglish majors read in college are very different from those they read inhigh school English classes The readings are considerably longer andmore demanding in content and style To increase their reading speedand their ability to cope with large quantities of EFL reading Englishmajors must either resort to translations or shift their reading style to amore macrotextual level

Whether reading English texts in British or US anthologies orChinese translations of those texts English majors in Taiwan are readinginformation presented in English rhetorical structure After 3 years ofsuch exposure students in their senior year may well be able to process

518 TESOL QUARTERLY

these texts as ef ciently as they can a text in Chinese rhetorical structureparticularly if the passage in question is expository or presents theauthorrsquos point of view typical for a classical essay written in the style of qi-cheng-zhuan-he No research has ever been conducted to test thatassumption

Chinese rhetorical structures found in the commonly used styles ofexpository texts contrast sharply with the rhetorical premises of Westernwriting Consequently the impact of culture-speci c rhetorical structureson the reading comprehension of Chinese students may be particularlyprofound for this type of writing Although researchers have looked atlanguage pro ciency (Carrell 1991) and text structure (Carrell 1992) asvariables to our knowledge no study has looked at both simultaneously1

This study examined the recall of readers at two different learning levelsmdashfreshmen and seniorsmdashwho read texts having identical content butrepresenting two distinct culture-speci c rhetorical conventions Thestudy poses the following questions (a) Does reading an L2 text thatfollows L1 rhetorical conventions affect EFL studentsrsquo recall and percep-tions of the text differently from reading the same text in L2 conventions(b) Do effects of rhetorical convention depend on the readerrsquos grade level

METHOD

Participants

To establish text selection criteria and choose suitable measurementsfor the main study we conducted a pilot study in June 1998 withfreshman English majors at Providence University a private university inTaiwan A total of 240 English majors at the same university participatedin the main study on a voluntary basis in December 1998 Half of theparticipants (98 females and 22 males) were recruited from fourfreshman sections and half (102 females and 18 males) were from foursenior classes The freshmen had had at least 6 years of formal EFLlanguage education in Taiwan which as noted above focused more onlanguage features than on the rhetorical aspects of English and includedrelatively limited access to extensive reading of naturally occurringEnglish prose written by Western authors for Western readers

In contrast the seniors had had 3 years of extensive exposure toEnglish texts written for native speakers of English in linguistics and

1 Work on the effects of rhetorical convention on reading comprehension has been done inother Asian languages notably Korean (Eggington 1987) and Japanese (Hinds 1987 Ricento1989)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 519

literature courses Presumably then the seniors had not only greater L2language pro ciency than did the freshmen but also considerably moreexposure to English rhetorical style Both ESL groups had had a formalMandarin Chinese education with a focus on classical Chinese while inhigh school and college and thus had been exposed to a rich L1environment with texts featuring traditional rhetorical conventions

Materials

Reading Passages

Four passages in English that followed Chinese rhetorical conventions(hereafter C versions) were selected from a Taiwanese bilingual magazineand modi ed to conform to English rhetorical conventions (hereafter Eversions) All the passages (Hsiao 1993a 1993b 1993c 1994) wereeditorials originally written in Chinese by a Chinese author and trans-lated into English for the magazine Editorials were chosen because thisgenre tends to use conventions that re ect features of qi-cheng-zhuan-heThe passages were judged as representative of qi-cheng-zhuan-he by sevenprofessors in the Department of Chinese Literature at a private universityin Taiwan

The four passages dealt with sociological issues of contemporaryTaiwanese society the unintended consequences of childcare for moth-ers with careers (ldquoChildcarerdquo Hsiao 1993c) the crisis one charityhospital in Taiwan faced because nonindigent patients abused its re-sources (ldquoCharityrdquo Hsiao 1993b) gender issues for male nurses inTaiwan (ldquoMale Nursesrdquo Hsiao 1994) and the legal problems facingparents who wanted to start alternative schools in Taiwan (ldquoSchoolingrdquoHsiao 1993a)

The E versions were developed in consultation with native-English-speaking faculty at a US university The second author revised thepassages to cue the organization of information in line with Westernrhetorical conventions Revisions involved both the sequencing of tex-tual chunks and the cueing of the ideas with discourse markers (fordetailed descriptions of this process and the resultant texts see Chu1999 pp 137ndash159) The sequencing principles were1 putting the topic or commentthesis and argument at the beginning

of the essay2 putting background information about the topic at the beginning of

the essay3 modifying general statements so that the argument changed from

inductive to deductive and was explicitly marked as such

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 515

poetry as well as public speaking it is most frequently applied inexpository writing (Kojima 1972) Literally qi means beginningmdashtheopening of a topic Cheng means followingmdashelaborating the openingZhuan means turning or changing Here the writer expresses anotherpoint of view turns to an example or develops the idea further Hemeans wrapping up when the writer provides the highest level ofgeneralization

Qi-cheng-zhuan-he is not a rigidly formalized structure the number ofwords and paragraphs as well as the amount of information devoted toeach of the four parts can vary disproportionately As long as the order ofpresentation remains the essence of the structure is there (Kojima1972) Hence qi-cheng-zhuan-he designates a conceptualreasoning se-quence in overall organization

Because qi-cheng-zhuan-he is simply one model for rhetorical sequenc-ing texts that use this model also exhibit other features commonly foundin Chinese writing These features differ markedly from those commonin Western writing (for examples see Appendix A) A tally of suchdifferences frequently starts with the Western preference for deductivestyle and the Chinese tendency toward inductive approaches particu-larly in expository writing The practice of putting the main thesis of atext before supporting ideas violates a Chinese readerrsquos expectation forwhat Kachru (1998) calls a ldquodelayed introduction of purposerdquo orldquodelayed topic statementrdquo (p 55)

Chen (1986) in comparing the expository discourse structure ofEnglish and Chinese passages found that Chinese paragraphs tended tointroduce more subtopics than did their Western counterparts Kirkpatrick(1995) quoting Wu (1988) suggests how inductive reasoning mightcondition Chinese rhetorical schemata Wu posits that inductive reason-ing is felt to be ldquonaturalrdquo whereas deductive reasoning requires aninformation sequence that goes against ldquoa natural sequencerdquo (p 281) asperceived by Chinese readers

A second difference researchers have noted between Western andChinese rhetorical conventions is closely related to the way main andsubordinate information is sequenced in each Discourse cues forsentential and intersentential cohesion tend not to occur as often inChinese as they do in English writing (Normant 1986) Further com-pounding this Chinese tendency toward indirect rather than directpresentation of authorial intent Jensen (1998) stated is that subtletyanalogy and metaphor highly valued writing techniques in classicalChinese tend to hide the writerrsquos intents He attributed this tendencytoward indirect allusion to the Chinese cultural heritage in rhetoric

As a result of these differences researchers have predicted thatChinese EFL readers because they have rhetorical expectations thatdiffer from those of Western readers might experience problems

516 TESOL QUARTERLY

comprehending a passage written with Western rhetorical conventionsMain ideas for example might not be recognized as such due to theirearly presentation and thus may fail to facilitate comprehension

Consequently Chinese EFL readers could be at a disadvantage whenreading English texts written by native speakers of English EFL readerswho predict a Chinese discourse structure may nd a mismatch betweentheir expectations and the macrostructures (the organization of contentand the sequence of ideas) presented in the Western text This gapbetween the Chinese readerrsquos anticipation of text development and thetextrsquos actual macrostrutures might result in lower comprehension andretention of an English text written in an English rhetorical structureConversely a text written in the rhetorical structure common in Chi-nesemdashthat is according to conventions such as those of qi-cheng-zhuan-he mdashmight facilitate recall for Chinese readers by providing them withfamiliar organizational macrostructures

The Role of Rhetorical Conventions inTaiwanese High Schools

In the past Taiwanese studentsrsquo exposure to English reading andwriting before they entered college focused more on the lexical andsyntactic features of a passage than on its rhetorical conventions Theirreading in Chinese on the other hand frequently dealt with theschemata of Chinese rhetorical conventions In terms of expositorywriting in their native language the classical Chinese rhetorical struc-ture qi-cheng-zhuan-he exerts a signi cant in uence as a macrostructurefor text analysis

For the students who participated in this study work with Englishlanguage texts did not emphasize the macrostructures of those passagesUntil 1999 when competitive marketing of textbooks was introducedthe traditional English textbooks commonly used in Taiwanese highschools offered little explicit training in the rhetorical organization oftexts Although reading formed the core component of these textbooksexercises for reading passages emphasized word- and sentence-levelmeanings rather than rhetorical organization Most of the readings werewritten originally by native speakers of English but were adapted oredited to control for length and to simplify vocabulary and syntax

The reading aids that commonly accompanied the reading passages inthe traditional textbooks were phonetic transcription and a glossary ofvocabulary translation and interpretation of idioms and phrases Cover-age of the rhetorical aspect of the reading was limited Reading andwriting exercises typically consisted of cloze vocabulary matching phrase

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 517

substitution translation questions on factual content and essay ques-tions on content exercises only rarely focused on the development ofideas and the discourse markers signaling the textual organization ofthose ideas Importantly for this study these textbooks did not explicitlycall studentsrsquo attention to rhetorical structures common in Englishwriting

In reading in Chinese on the other hand rhetorical structures play asigni cant role in secondary school instruction High school studentsrsquocourses in Chinese involve intensive reading of classical Chinese andmemorization of classical texts as part of the literary tradition A keypractice is the paraphrasing of classical texts into modern ChineseBecause text reproduction tasks require the construction of a text basestudents must inevitably attend to the global structure of the textconsciously or subconsciously in the retention and retrieval processesPossibly then students in this study developed robust text-structureschemata based on rhetorical conventions often found in Chinesewriting such as the above-mentioned qi-cheng-zhuan-he

The English Majorrsquos Literacy Experience at College in Taiwan

In Taiwan English majors such as those who participated in this studyenroll in content courses in linguistics and English literature in whichthey use collections such as The Norton Anthology of American Literature(Abrams 1996) that have been compiled for native-English-speakingstudents In the freshman year most courses focus on language skills afew are introductory courses in linguistics and literature As studentsmove on to the sophomore junior and senior years the proportion oflanguage skill courses decreases and content courses in linguistics andliterature increase Because this study was conducted while the freshmenwere in their rst semester and the seniors in their seventh semester thefreshmenrsquos exposure to authentic texts was much more limited than thatof the seniors

Regardless of grade level the texts in linguistics and literature thatEnglish majors read in college are very different from those they read inhigh school English classes The readings are considerably longer andmore demanding in content and style To increase their reading speedand their ability to cope with large quantities of EFL reading Englishmajors must either resort to translations or shift their reading style to amore macrotextual level

Whether reading English texts in British or US anthologies orChinese translations of those texts English majors in Taiwan are readinginformation presented in English rhetorical structure After 3 years ofsuch exposure students in their senior year may well be able to process

518 TESOL QUARTERLY

these texts as ef ciently as they can a text in Chinese rhetorical structureparticularly if the passage in question is expository or presents theauthorrsquos point of view typical for a classical essay written in the style of qi-cheng-zhuan-he No research has ever been conducted to test thatassumption

Chinese rhetorical structures found in the commonly used styles ofexpository texts contrast sharply with the rhetorical premises of Westernwriting Consequently the impact of culture-speci c rhetorical structureson the reading comprehension of Chinese students may be particularlyprofound for this type of writing Although researchers have looked atlanguage pro ciency (Carrell 1991) and text structure (Carrell 1992) asvariables to our knowledge no study has looked at both simultaneously1

This study examined the recall of readers at two different learning levelsmdashfreshmen and seniorsmdashwho read texts having identical content butrepresenting two distinct culture-speci c rhetorical conventions Thestudy poses the following questions (a) Does reading an L2 text thatfollows L1 rhetorical conventions affect EFL studentsrsquo recall and percep-tions of the text differently from reading the same text in L2 conventions(b) Do effects of rhetorical convention depend on the readerrsquos grade level

METHOD

Participants

To establish text selection criteria and choose suitable measurementsfor the main study we conducted a pilot study in June 1998 withfreshman English majors at Providence University a private university inTaiwan A total of 240 English majors at the same university participatedin the main study on a voluntary basis in December 1998 Half of theparticipants (98 females and 22 males) were recruited from fourfreshman sections and half (102 females and 18 males) were from foursenior classes The freshmen had had at least 6 years of formal EFLlanguage education in Taiwan which as noted above focused more onlanguage features than on the rhetorical aspects of English and includedrelatively limited access to extensive reading of naturally occurringEnglish prose written by Western authors for Western readers

In contrast the seniors had had 3 years of extensive exposure toEnglish texts written for native speakers of English in linguistics and

1 Work on the effects of rhetorical convention on reading comprehension has been done inother Asian languages notably Korean (Eggington 1987) and Japanese (Hinds 1987 Ricento1989)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 519

literature courses Presumably then the seniors had not only greater L2language pro ciency than did the freshmen but also considerably moreexposure to English rhetorical style Both ESL groups had had a formalMandarin Chinese education with a focus on classical Chinese while inhigh school and college and thus had been exposed to a rich L1environment with texts featuring traditional rhetorical conventions

Materials

Reading Passages

Four passages in English that followed Chinese rhetorical conventions(hereafter C versions) were selected from a Taiwanese bilingual magazineand modi ed to conform to English rhetorical conventions (hereafter Eversions) All the passages (Hsiao 1993a 1993b 1993c 1994) wereeditorials originally written in Chinese by a Chinese author and trans-lated into English for the magazine Editorials were chosen because thisgenre tends to use conventions that re ect features of qi-cheng-zhuan-heThe passages were judged as representative of qi-cheng-zhuan-he by sevenprofessors in the Department of Chinese Literature at a private universityin Taiwan

The four passages dealt with sociological issues of contemporaryTaiwanese society the unintended consequences of childcare for moth-ers with careers (ldquoChildcarerdquo Hsiao 1993c) the crisis one charityhospital in Taiwan faced because nonindigent patients abused its re-sources (ldquoCharityrdquo Hsiao 1993b) gender issues for male nurses inTaiwan (ldquoMale Nursesrdquo Hsiao 1994) and the legal problems facingparents who wanted to start alternative schools in Taiwan (ldquoSchoolingrdquoHsiao 1993a)

The E versions were developed in consultation with native-English-speaking faculty at a US university The second author revised thepassages to cue the organization of information in line with Westernrhetorical conventions Revisions involved both the sequencing of tex-tual chunks and the cueing of the ideas with discourse markers (fordetailed descriptions of this process and the resultant texts see Chu1999 pp 137ndash159) The sequencing principles were1 putting the topic or commentthesis and argument at the beginning

of the essay2 putting background information about the topic at the beginning of

the essay3 modifying general statements so that the argument changed from

inductive to deductive and was explicitly marked as such

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

516 TESOL QUARTERLY

comprehending a passage written with Western rhetorical conventionsMain ideas for example might not be recognized as such due to theirearly presentation and thus may fail to facilitate comprehension

Consequently Chinese EFL readers could be at a disadvantage whenreading English texts written by native speakers of English EFL readerswho predict a Chinese discourse structure may nd a mismatch betweentheir expectations and the macrostructures (the organization of contentand the sequence of ideas) presented in the Western text This gapbetween the Chinese readerrsquos anticipation of text development and thetextrsquos actual macrostrutures might result in lower comprehension andretention of an English text written in an English rhetorical structureConversely a text written in the rhetorical structure common in Chi-nesemdashthat is according to conventions such as those of qi-cheng-zhuan-he mdashmight facilitate recall for Chinese readers by providing them withfamiliar organizational macrostructures

The Role of Rhetorical Conventions inTaiwanese High Schools

In the past Taiwanese studentsrsquo exposure to English reading andwriting before they entered college focused more on the lexical andsyntactic features of a passage than on its rhetorical conventions Theirreading in Chinese on the other hand frequently dealt with theschemata of Chinese rhetorical conventions In terms of expositorywriting in their native language the classical Chinese rhetorical struc-ture qi-cheng-zhuan-he exerts a signi cant in uence as a macrostructurefor text analysis

For the students who participated in this study work with Englishlanguage texts did not emphasize the macrostructures of those passagesUntil 1999 when competitive marketing of textbooks was introducedthe traditional English textbooks commonly used in Taiwanese highschools offered little explicit training in the rhetorical organization oftexts Although reading formed the core component of these textbooksexercises for reading passages emphasized word- and sentence-levelmeanings rather than rhetorical organization Most of the readings werewritten originally by native speakers of English but were adapted oredited to control for length and to simplify vocabulary and syntax

The reading aids that commonly accompanied the reading passages inthe traditional textbooks were phonetic transcription and a glossary ofvocabulary translation and interpretation of idioms and phrases Cover-age of the rhetorical aspect of the reading was limited Reading andwriting exercises typically consisted of cloze vocabulary matching phrase

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 517

substitution translation questions on factual content and essay ques-tions on content exercises only rarely focused on the development ofideas and the discourse markers signaling the textual organization ofthose ideas Importantly for this study these textbooks did not explicitlycall studentsrsquo attention to rhetorical structures common in Englishwriting

In reading in Chinese on the other hand rhetorical structures play asigni cant role in secondary school instruction High school studentsrsquocourses in Chinese involve intensive reading of classical Chinese andmemorization of classical texts as part of the literary tradition A keypractice is the paraphrasing of classical texts into modern ChineseBecause text reproduction tasks require the construction of a text basestudents must inevitably attend to the global structure of the textconsciously or subconsciously in the retention and retrieval processesPossibly then students in this study developed robust text-structureschemata based on rhetorical conventions often found in Chinesewriting such as the above-mentioned qi-cheng-zhuan-he

The English Majorrsquos Literacy Experience at College in Taiwan

In Taiwan English majors such as those who participated in this studyenroll in content courses in linguistics and English literature in whichthey use collections such as The Norton Anthology of American Literature(Abrams 1996) that have been compiled for native-English-speakingstudents In the freshman year most courses focus on language skills afew are introductory courses in linguistics and literature As studentsmove on to the sophomore junior and senior years the proportion oflanguage skill courses decreases and content courses in linguistics andliterature increase Because this study was conducted while the freshmenwere in their rst semester and the seniors in their seventh semester thefreshmenrsquos exposure to authentic texts was much more limited than thatof the seniors

Regardless of grade level the texts in linguistics and literature thatEnglish majors read in college are very different from those they read inhigh school English classes The readings are considerably longer andmore demanding in content and style To increase their reading speedand their ability to cope with large quantities of EFL reading Englishmajors must either resort to translations or shift their reading style to amore macrotextual level

Whether reading English texts in British or US anthologies orChinese translations of those texts English majors in Taiwan are readinginformation presented in English rhetorical structure After 3 years ofsuch exposure students in their senior year may well be able to process

518 TESOL QUARTERLY

these texts as ef ciently as they can a text in Chinese rhetorical structureparticularly if the passage in question is expository or presents theauthorrsquos point of view typical for a classical essay written in the style of qi-cheng-zhuan-he No research has ever been conducted to test thatassumption

Chinese rhetorical structures found in the commonly used styles ofexpository texts contrast sharply with the rhetorical premises of Westernwriting Consequently the impact of culture-speci c rhetorical structureson the reading comprehension of Chinese students may be particularlyprofound for this type of writing Although researchers have looked atlanguage pro ciency (Carrell 1991) and text structure (Carrell 1992) asvariables to our knowledge no study has looked at both simultaneously1

This study examined the recall of readers at two different learning levelsmdashfreshmen and seniorsmdashwho read texts having identical content butrepresenting two distinct culture-speci c rhetorical conventions Thestudy poses the following questions (a) Does reading an L2 text thatfollows L1 rhetorical conventions affect EFL studentsrsquo recall and percep-tions of the text differently from reading the same text in L2 conventions(b) Do effects of rhetorical convention depend on the readerrsquos grade level

METHOD

Participants

To establish text selection criteria and choose suitable measurementsfor the main study we conducted a pilot study in June 1998 withfreshman English majors at Providence University a private university inTaiwan A total of 240 English majors at the same university participatedin the main study on a voluntary basis in December 1998 Half of theparticipants (98 females and 22 males) were recruited from fourfreshman sections and half (102 females and 18 males) were from foursenior classes The freshmen had had at least 6 years of formal EFLlanguage education in Taiwan which as noted above focused more onlanguage features than on the rhetorical aspects of English and includedrelatively limited access to extensive reading of naturally occurringEnglish prose written by Western authors for Western readers

In contrast the seniors had had 3 years of extensive exposure toEnglish texts written for native speakers of English in linguistics and

1 Work on the effects of rhetorical convention on reading comprehension has been done inother Asian languages notably Korean (Eggington 1987) and Japanese (Hinds 1987 Ricento1989)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 519

literature courses Presumably then the seniors had not only greater L2language pro ciency than did the freshmen but also considerably moreexposure to English rhetorical style Both ESL groups had had a formalMandarin Chinese education with a focus on classical Chinese while inhigh school and college and thus had been exposed to a rich L1environment with texts featuring traditional rhetorical conventions

Materials

Reading Passages

Four passages in English that followed Chinese rhetorical conventions(hereafter C versions) were selected from a Taiwanese bilingual magazineand modi ed to conform to English rhetorical conventions (hereafter Eversions) All the passages (Hsiao 1993a 1993b 1993c 1994) wereeditorials originally written in Chinese by a Chinese author and trans-lated into English for the magazine Editorials were chosen because thisgenre tends to use conventions that re ect features of qi-cheng-zhuan-heThe passages were judged as representative of qi-cheng-zhuan-he by sevenprofessors in the Department of Chinese Literature at a private universityin Taiwan

The four passages dealt with sociological issues of contemporaryTaiwanese society the unintended consequences of childcare for moth-ers with careers (ldquoChildcarerdquo Hsiao 1993c) the crisis one charityhospital in Taiwan faced because nonindigent patients abused its re-sources (ldquoCharityrdquo Hsiao 1993b) gender issues for male nurses inTaiwan (ldquoMale Nursesrdquo Hsiao 1994) and the legal problems facingparents who wanted to start alternative schools in Taiwan (ldquoSchoolingrdquoHsiao 1993a)

The E versions were developed in consultation with native-English-speaking faculty at a US university The second author revised thepassages to cue the organization of information in line with Westernrhetorical conventions Revisions involved both the sequencing of tex-tual chunks and the cueing of the ideas with discourse markers (fordetailed descriptions of this process and the resultant texts see Chu1999 pp 137ndash159) The sequencing principles were1 putting the topic or commentthesis and argument at the beginning

of the essay2 putting background information about the topic at the beginning of

the essay3 modifying general statements so that the argument changed from

inductive to deductive and was explicitly marked as such

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 517

substitution translation questions on factual content and essay ques-tions on content exercises only rarely focused on the development ofideas and the discourse markers signaling the textual organization ofthose ideas Importantly for this study these textbooks did not explicitlycall studentsrsquo attention to rhetorical structures common in Englishwriting

In reading in Chinese on the other hand rhetorical structures play asigni cant role in secondary school instruction High school studentsrsquocourses in Chinese involve intensive reading of classical Chinese andmemorization of classical texts as part of the literary tradition A keypractice is the paraphrasing of classical texts into modern ChineseBecause text reproduction tasks require the construction of a text basestudents must inevitably attend to the global structure of the textconsciously or subconsciously in the retention and retrieval processesPossibly then students in this study developed robust text-structureschemata based on rhetorical conventions often found in Chinesewriting such as the above-mentioned qi-cheng-zhuan-he

The English Majorrsquos Literacy Experience at College in Taiwan

In Taiwan English majors such as those who participated in this studyenroll in content courses in linguistics and English literature in whichthey use collections such as The Norton Anthology of American Literature(Abrams 1996) that have been compiled for native-English-speakingstudents In the freshman year most courses focus on language skills afew are introductory courses in linguistics and literature As studentsmove on to the sophomore junior and senior years the proportion oflanguage skill courses decreases and content courses in linguistics andliterature increase Because this study was conducted while the freshmenwere in their rst semester and the seniors in their seventh semester thefreshmenrsquos exposure to authentic texts was much more limited than thatof the seniors

Regardless of grade level the texts in linguistics and literature thatEnglish majors read in college are very different from those they read inhigh school English classes The readings are considerably longer andmore demanding in content and style To increase their reading speedand their ability to cope with large quantities of EFL reading Englishmajors must either resort to translations or shift their reading style to amore macrotextual level

Whether reading English texts in British or US anthologies orChinese translations of those texts English majors in Taiwan are readinginformation presented in English rhetorical structure After 3 years ofsuch exposure students in their senior year may well be able to process

518 TESOL QUARTERLY

these texts as ef ciently as they can a text in Chinese rhetorical structureparticularly if the passage in question is expository or presents theauthorrsquos point of view typical for a classical essay written in the style of qi-cheng-zhuan-he No research has ever been conducted to test thatassumption

Chinese rhetorical structures found in the commonly used styles ofexpository texts contrast sharply with the rhetorical premises of Westernwriting Consequently the impact of culture-speci c rhetorical structureson the reading comprehension of Chinese students may be particularlyprofound for this type of writing Although researchers have looked atlanguage pro ciency (Carrell 1991) and text structure (Carrell 1992) asvariables to our knowledge no study has looked at both simultaneously1

This study examined the recall of readers at two different learning levelsmdashfreshmen and seniorsmdashwho read texts having identical content butrepresenting two distinct culture-speci c rhetorical conventions Thestudy poses the following questions (a) Does reading an L2 text thatfollows L1 rhetorical conventions affect EFL studentsrsquo recall and percep-tions of the text differently from reading the same text in L2 conventions(b) Do effects of rhetorical convention depend on the readerrsquos grade level

METHOD

Participants

To establish text selection criteria and choose suitable measurementsfor the main study we conducted a pilot study in June 1998 withfreshman English majors at Providence University a private university inTaiwan A total of 240 English majors at the same university participatedin the main study on a voluntary basis in December 1998 Half of theparticipants (98 females and 22 males) were recruited from fourfreshman sections and half (102 females and 18 males) were from foursenior classes The freshmen had had at least 6 years of formal EFLlanguage education in Taiwan which as noted above focused more onlanguage features than on the rhetorical aspects of English and includedrelatively limited access to extensive reading of naturally occurringEnglish prose written by Western authors for Western readers

In contrast the seniors had had 3 years of extensive exposure toEnglish texts written for native speakers of English in linguistics and

1 Work on the effects of rhetorical convention on reading comprehension has been done inother Asian languages notably Korean (Eggington 1987) and Japanese (Hinds 1987 Ricento1989)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 519

literature courses Presumably then the seniors had not only greater L2language pro ciency than did the freshmen but also considerably moreexposure to English rhetorical style Both ESL groups had had a formalMandarin Chinese education with a focus on classical Chinese while inhigh school and college and thus had been exposed to a rich L1environment with texts featuring traditional rhetorical conventions

Materials

Reading Passages

Four passages in English that followed Chinese rhetorical conventions(hereafter C versions) were selected from a Taiwanese bilingual magazineand modi ed to conform to English rhetorical conventions (hereafter Eversions) All the passages (Hsiao 1993a 1993b 1993c 1994) wereeditorials originally written in Chinese by a Chinese author and trans-lated into English for the magazine Editorials were chosen because thisgenre tends to use conventions that re ect features of qi-cheng-zhuan-heThe passages were judged as representative of qi-cheng-zhuan-he by sevenprofessors in the Department of Chinese Literature at a private universityin Taiwan

The four passages dealt with sociological issues of contemporaryTaiwanese society the unintended consequences of childcare for moth-ers with careers (ldquoChildcarerdquo Hsiao 1993c) the crisis one charityhospital in Taiwan faced because nonindigent patients abused its re-sources (ldquoCharityrdquo Hsiao 1993b) gender issues for male nurses inTaiwan (ldquoMale Nursesrdquo Hsiao 1994) and the legal problems facingparents who wanted to start alternative schools in Taiwan (ldquoSchoolingrdquoHsiao 1993a)

The E versions were developed in consultation with native-English-speaking faculty at a US university The second author revised thepassages to cue the organization of information in line with Westernrhetorical conventions Revisions involved both the sequencing of tex-tual chunks and the cueing of the ideas with discourse markers (fordetailed descriptions of this process and the resultant texts see Chu1999 pp 137ndash159) The sequencing principles were1 putting the topic or commentthesis and argument at the beginning

of the essay2 putting background information about the topic at the beginning of

the essay3 modifying general statements so that the argument changed from

inductive to deductive and was explicitly marked as such

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

518 TESOL QUARTERLY

these texts as ef ciently as they can a text in Chinese rhetorical structureparticularly if the passage in question is expository or presents theauthorrsquos point of view typical for a classical essay written in the style of qi-cheng-zhuan-he No research has ever been conducted to test thatassumption

Chinese rhetorical structures found in the commonly used styles ofexpository texts contrast sharply with the rhetorical premises of Westernwriting Consequently the impact of culture-speci c rhetorical structureson the reading comprehension of Chinese students may be particularlyprofound for this type of writing Although researchers have looked atlanguage pro ciency (Carrell 1991) and text structure (Carrell 1992) asvariables to our knowledge no study has looked at both simultaneously1

This study examined the recall of readers at two different learning levelsmdashfreshmen and seniorsmdashwho read texts having identical content butrepresenting two distinct culture-speci c rhetorical conventions Thestudy poses the following questions (a) Does reading an L2 text thatfollows L1 rhetorical conventions affect EFL studentsrsquo recall and percep-tions of the text differently from reading the same text in L2 conventions(b) Do effects of rhetorical convention depend on the readerrsquos grade level

METHOD

Participants

To establish text selection criteria and choose suitable measurementsfor the main study we conducted a pilot study in June 1998 withfreshman English majors at Providence University a private university inTaiwan A total of 240 English majors at the same university participatedin the main study on a voluntary basis in December 1998 Half of theparticipants (98 females and 22 males) were recruited from fourfreshman sections and half (102 females and 18 males) were from foursenior classes The freshmen had had at least 6 years of formal EFLlanguage education in Taiwan which as noted above focused more onlanguage features than on the rhetorical aspects of English and includedrelatively limited access to extensive reading of naturally occurringEnglish prose written by Western authors for Western readers

In contrast the seniors had had 3 years of extensive exposure toEnglish texts written for native speakers of English in linguistics and

1 Work on the effects of rhetorical convention on reading comprehension has been done inother Asian languages notably Korean (Eggington 1987) and Japanese (Hinds 1987 Ricento1989)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 519

literature courses Presumably then the seniors had not only greater L2language pro ciency than did the freshmen but also considerably moreexposure to English rhetorical style Both ESL groups had had a formalMandarin Chinese education with a focus on classical Chinese while inhigh school and college and thus had been exposed to a rich L1environment with texts featuring traditional rhetorical conventions

Materials

Reading Passages

Four passages in English that followed Chinese rhetorical conventions(hereafter C versions) were selected from a Taiwanese bilingual magazineand modi ed to conform to English rhetorical conventions (hereafter Eversions) All the passages (Hsiao 1993a 1993b 1993c 1994) wereeditorials originally written in Chinese by a Chinese author and trans-lated into English for the magazine Editorials were chosen because thisgenre tends to use conventions that re ect features of qi-cheng-zhuan-heThe passages were judged as representative of qi-cheng-zhuan-he by sevenprofessors in the Department of Chinese Literature at a private universityin Taiwan

The four passages dealt with sociological issues of contemporaryTaiwanese society the unintended consequences of childcare for moth-ers with careers (ldquoChildcarerdquo Hsiao 1993c) the crisis one charityhospital in Taiwan faced because nonindigent patients abused its re-sources (ldquoCharityrdquo Hsiao 1993b) gender issues for male nurses inTaiwan (ldquoMale Nursesrdquo Hsiao 1994) and the legal problems facingparents who wanted to start alternative schools in Taiwan (ldquoSchoolingrdquoHsiao 1993a)

The E versions were developed in consultation with native-English-speaking faculty at a US university The second author revised thepassages to cue the organization of information in line with Westernrhetorical conventions Revisions involved both the sequencing of tex-tual chunks and the cueing of the ideas with discourse markers (fordetailed descriptions of this process and the resultant texts see Chu1999 pp 137ndash159) The sequencing principles were1 putting the topic or commentthesis and argument at the beginning

of the essay2 putting background information about the topic at the beginning of

the essay3 modifying general statements so that the argument changed from

inductive to deductive and was explicitly marked as such

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 519

literature courses Presumably then the seniors had not only greater L2language pro ciency than did the freshmen but also considerably moreexposure to English rhetorical style Both ESL groups had had a formalMandarin Chinese education with a focus on classical Chinese while inhigh school and college and thus had been exposed to a rich L1environment with texts featuring traditional rhetorical conventions

Materials

Reading Passages

Four passages in English that followed Chinese rhetorical conventions(hereafter C versions) were selected from a Taiwanese bilingual magazineand modi ed to conform to English rhetorical conventions (hereafter Eversions) All the passages (Hsiao 1993a 1993b 1993c 1994) wereeditorials originally written in Chinese by a Chinese author and trans-lated into English for the magazine Editorials were chosen because thisgenre tends to use conventions that re ect features of qi-cheng-zhuan-heThe passages were judged as representative of qi-cheng-zhuan-he by sevenprofessors in the Department of Chinese Literature at a private universityin Taiwan

The four passages dealt with sociological issues of contemporaryTaiwanese society the unintended consequences of childcare for moth-ers with careers (ldquoChildcarerdquo Hsiao 1993c) the crisis one charityhospital in Taiwan faced because nonindigent patients abused its re-sources (ldquoCharityrdquo Hsiao 1993b) gender issues for male nurses inTaiwan (ldquoMale Nursesrdquo Hsiao 1994) and the legal problems facingparents who wanted to start alternative schools in Taiwan (ldquoSchoolingrdquoHsiao 1993a)

The E versions were developed in consultation with native-English-speaking faculty at a US university The second author revised thepassages to cue the organization of information in line with Westernrhetorical conventions Revisions involved both the sequencing of tex-tual chunks and the cueing of the ideas with discourse markers (fordetailed descriptions of this process and the resultant texts see Chu1999 pp 137ndash159) The sequencing principles were1 putting the topic or commentthesis and argument at the beginning

of the essay2 putting background information about the topic at the beginning of

the essay3 modifying general statements so that the argument changed from

inductive to deductive and was explicitly marked as such

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

520 TESOL QUARTERLY

In addition to textual chunks being rearranged discourse markersexplicitly linking local structures were added as follows

1 Time markers were inserted to show contrast between macroprop-ositions

2 Key terms were elaborated or reiterated in context for semanticcohesion

3 Pronouns and demonstratives were inserted to increase syntacticcohesion

Not surprisingly these naturally occurring passages required differentamounts of modi cation to conform to English rhetorical conventionsIn ldquoChildcarerdquo and ldquoCharityrdquo discourse connectors were added betweenmost paragraphs and several sentences In ldquoMale Nursesrdquo and ldquoSchool-ingrdquo changes were restricted largely to topical rearrangement As iscommon in studies involving textual passages we included a Passagefactor in the data analysis to check for consistency of effects acrosspassages This check also allowed us to examine informally whether Eversions that required more modi cations produced effects similar tothe others

Altogether the resultant eight passages ranged between 486 and 558words in length and received scores on the Bormuth Grade Level(Bormuth 1969) ranging from 100 to 115 The scores of the C and Eversions of the passages differed only slightly (30) on this measure sothey can be considered equivalent in their readability (see Table 1)

Questionnaires and Recall Measures

Two questionnaires (see Appendix B) and two recall measures wereemployed in this study The Passage Perception Questionnaire assessedparticipantsrsquo judgments of eight features of the passages (contentinterest familiarity concreteness comprehensibility memorability clar-ity of argument organization and rhetorical identity characteristic ofChinese or English usage) each on a 5-point Likert scale with 1 the mostnegative and 5 the most positive ranking The Topic Assessment Ques-tionnaire was designed as a text-independent assessment of the Chinesestudent readersrsquo interest in and familiarity with the topics of the fourpassages With the passage titles serving as prompts the participantsrated their interest and familiarity on separate 5-point Likert scalesagain with 1 the lowest and 5 the highest ranking

Free-recall tests were administered immediately after the participantsread a passage and at a 1-week delay Participants wrote their recallprotocols in Chinese Although the passages were presented in English

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 521

the questionnaires and recall prompts were presented in Chinese tominimize possible confounding effects of language deciencies (Lee 1986)

Design

This study employed a 2 2 4 mixed factorial design withRhetorical Convention (Chinese vs English versions) as a within-subjectfactor and Grade Level (freshmen vs seniors) and Passage (the fourpassages ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo and ldquoSchoolingrdquo) asbetween-subject factors

The four passages yielded eight texts because each appeared in twoversions re ecting the two rhetorical styles Chinese and English Eachstudent read and responded to only two passages one in each rhetoricalconvention Passage pairings were counterbalanced with rhetorical con-vention versions and passages thus producing 24 passage-version pairswhich were randomly assigned to students within intact classes Eachpassage-version pair was read by 5 freshmen and 5 seniors Overall thenwith a total of 240 participants 30 participants at each grade level readeach passage in either its Chinese or its English version

The Passage variable compared results for the four passages to see iftheir content or other features made them harder or easier to read

TABLE 1

Readability of the Four Passages in Two Versions

Passage

ldquoChildcarerdquo ldquoCharityrdquo ldquoMale Nursesrdquo ldquoSchoolingrdquo

Measure C E C E C E C E

Words 530 542 486 586 531 516 555 558Characters 2616 2759 2433 2971 2733 2663 2945 2984Paragraphs 10 8 13 14 8 6 13 10Sentences 26 28 27 29 18 18 31 29Mean sentencesper paragraph 26 35 20 20 22 3 23 29

Mean wordsper sentence 203 193 178 20 295 286 179 185

Mean charactersper word 47 49 48 49 50 50 51 51

Passivesentences () 11 7 11 13 22 22 16 13

BormuthGrade Level 103 100 101 103 115 115 103 106

Note C = Chinese rhetorical version E = English rhetorical version

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

522 TESOL QUARTERLY

Passage was designated as a four-level between-subjects factor Thisdecision re ects the fact that although individual participants read onlytwo of the four passages they did so in 24 groupings that had 24 differentpairings of the four passages all of which were counterbalanced with theorder of passage topics and the order of rhetorical conventions Weadjusted the Passage variable statistically to account for a within-subjectinterclass correlation between any two of the four passages using the SASMixed Procedure so that the Passage could be assessed as a between-subject factor The within-subject intraclass correlations between the twopassages were 6 for immediate recall and 7 for delayed recall

The dependent variables were (a) scores on the immediate- anddelayed-recall tests (b) responses to the Passage Perception Question-naire and (c) responses to the Topic Assessment Questionnaire Thesigni cance level for all statistical analyses was set at p 05

Rather than pretesting for interest and familiarity we administeredthe Topic Assessment Questionnaire at the end of the experiment assuggested by Spyridakis and Wenger (1991) In this way we eliminatedthe chance that the questionnaire would prime content knowledge or atextual schema and thus enhance comprehension or recall Further toavoid having the interest and familiarity scores themselves confoundedby participation in the experiment we excluded responses for thepassages that the participants had read only responses for the twopassages that the participants had not read were subjected to statisticalanalysis

Procedure

The study was conducted in two sessions either during regular classperiods or at times scheduled by the researcher in 2 consecutive weeksIn the rst session each participant read the rst passage for 15 minutesthen with the passage removed lled out a Passage Perception Ques-tionnaire for 3 minutes and immediately wrote a free-recall response for25 minutes The participants were instructed to write their free-recallresponses in Chinese because writing in a foreign language may obscureevidence of comprehension or inhibit recall (Lee 1986) They werereminded orally to write as much as they could remember and to adhereas closely as possible to information in the passage This procedure wasrepeated for the second passage One week later the participants had amaximum of 40 minutes to complete a delayed-recall test for each of thepassages and then took 5 minutes to ll out the Topic AssessmentQuestionnaire for each

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 523

Scoring of Recall Protocols

The recall protocols were scored using a pausal unit system developedby Johnson (1970) and validated by Bernhardt (1991) First two nativeEnglish speakers read the eight passages aloud to themselves andmarked the boundaries of semantic units Two native Chinese speakers uent in English followed Johnsonrsquos procedures for resolving discrepan-cies in unit identi cation They chose the narrower unit in all casesexcept when a larger unit corresponded naturally to a common four-character Chinese idiom or when speci c syntactic or morphologicallanguage differences (eg treatment of possessives) made a larger unitmore natural in Chinese Care was taken to ensure consistency betweenthe corresponding units in the C and E versions of each passage Thenumber of resulting pausal units for the eight passages ranged from 110to 129 units (ldquoChildcarerdquo C version 122 E version 129 ldquoCharityrdquo Cversion 110 E version 126 ldquoMale Nursesrdquo C version 121 E version 120ldquoSchoolingrdquo C version 117 E version 119)

The rst author of this study a native speaker of Mandarin Chinesescored student recall protocols for the presence or absence of each unitno partial credit was given To establish reliability a second nativespeaker of Chinese scored 20 randomly selected recall protocols Themean Pearson product-moment coef cient was 96 for the immediate-recall test and 95 for the delayed-recall test Recall scores were reportedas the proportion of total pausal units recalled for each passage

Assumptions and Analyses

Analysis of variance (ANOVA) and analysis of covariance (ANCOVA)were employed in this study We expected that the three assumptions ofANOVA and ANCOVA analysesmdashnormality homogeneity of varianceand independencemdashwere met based on the following observationsStudents at each grade level were from a homogeneous background andthe sample size was large Hence we assumed that the dependentvariables eight passage perceptions and two types of recall had a normaldistribution within each group and consequently homogeneity ofvariance For the third assumption independence the following designfeatures and analysis procedures were observed First the participantswere randomly sampled and randomly assigned to groups and thequestionnaire and recall data were taken independently from eachparticipant to ensure independent sampling Second the dependentvariables were analyzed separately in other words ANOVA was performedon each of the eight passage perceptions and ANCOVA was performedon immediate recall and delayed recall respectively

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

524 TESOL QUARTERLY

A three-way ANOVA mixed procedure was rst performed on theresponses to each of the eight items on the Passage Perception Question-naire to test the effect of Passage Rhetorical Convention and GradeLevel on each of the eight perceptions We did not adjust the signi -cance level because these were all planned comparisons and subsequentcorrelational analyses served as a check on reliability Five perceptionsmdashinterest familiarity comprehensibility memorability and organizationmdashshowed signi cant effects of Passage (see Appendix C)

To control for variance in recalls due to difference in passageperceptions therefore we entered these ve perception scores ascovariates in subsequent analyses of the recall data A three-way ANCOVAmixed procedure was then performed on immediate- and delayed-recallscores with the ve perception scores entered as covariates (see Appen-dix D) The correlation matrix for ve perception variables was checkedto avoid the inclusion of redundant variables in ANCOVAs The resultsshow correlations among variables ranging from 13 to 57 indicatingweak correlations among covariate variables Therefore all ve percep-tion variables were judged relevant

RESULTS

Recall Measures

The likelihood ratio chi-square values are 4122 for the immediate-recall model and 5853 for the delayed-recall model Both are statisticallysigni cant at the 0001 level Both Rhetorical Convention and GradeLevel signi cantly in uenced the participantsrsquo recall (see Table 2)Overall students recalled signi cantly more of the passages in the Cversions than in the E versions (adjusted mean percentage = 324 and282 respectively) in the immediate-recall test F(1221) = 3111 p 0001 (see Appendix D) On the delayed-recall test students alsorecalled a signi cantly higher percentage of pausal units in the Cversions than in the E versions (adjusted mean percentage = 231 and196 respectively) F(1 221) = 3994 p 0001 (see Appendix D) Inother words at both test intervals students remembered more of anEnglish passage when it followed Chinese rhetorical conventions

As expected grade level also in uenced recall Seniors recalledsigni cantly more than freshmen did (adjusted mean percentage = 352and 254 respectively) on the immediate-recall test F(1 238) = 7224p 0001 On the delayed test seniors again recalled a much higheraverage score than freshmen (adjusted mean percentage = 251 and 176respectively) across rhetorical conventions F(1238) = 6271 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 525

No signi cant interaction between Rhetorical Convention and GradeLevel emerged for immediate recall Both freshmen and seniors scoredsubstantially higher in recall of C versions than of E versions (adjustedmean percentage = 269 and 239 respectively for freshmen 378 and325 respectively for seniors see Table 2 and Appendix D) Accordinglythe effect size is a mean percentage of 30 for freshmen and 53 forseniors in immediate recall

For the delayed-recall measure however a signi cant interactionbetween Rhetorical Convention and Grade Level emerged F(1 221) =724 p 01 (see Appendix D) Although the superiority of C-versionscores held for both freshmen and seniors the effect size for freshmendropped from 30 percentage points in immediate recall to 19 as shownin Table 2 The senior effect size 50 remained largely the same as thatfor immediate recall

Overall these results suggest that English rhetorical conventions wererelatively unfamiliar to both freshmen and seniors As for the diminishedeffect size on freshmen after a delay the recall test may have proved sodif cult for freshmen that any advantage of reading a text with familiarChinese rhetorical conventions was lost For seniors higher languagepro ciency in the L2 did not result in a proportionately higher level oflong-term retention of texts in the L2 rhetorical convention Apparentlythe seniors who had initially recalled more detail (presumably on the

TABLE 2

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Grade Level

Passage version

Chinese English

Adjusted AdjustedGrade level mean SE mean SE F(1 221) p

Immediate recallFreshmana 269 96 239 97 807 0049Seniora 378 96 325 96 2545 0001Both levelsb 324 68 282 68 3111 0001

Delayed recallFreshmana 186 78 166 78 562 0186Seniora 276 78 226 78 3884 0001Both levelsb 231 55 196 55 3694 0001

Note Scores are mean percentages of total pausal units recalled for each passage adjusted bywithin-subject correlation and by effects of ve covariates perceived interest perceivedfamiliarity perceived comprehensibility perceived memorability and perceived organizationaN = 120 bN = 240

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

526 TESOL QUARTERLY

basis of their greater language pro ciency) lost an equivalent amount ofdetail in the delayed recalls for both the C and the E versions

Although Passage did not produce a main effect on recall at either testinterval an interaction of Passage and Rhetorical Convention wassigni cant for immediate recall F(3 221) = 406 p 01 (see Appen-dix D) Further comparisons on Rhetorical Convention for the fourpassages show that only the passages ldquoCharityrdquo and ldquoMale Nursesrdquoproduced signi cantly greater differences between the C and E versions(see Table 3) For the ldquoChildcarerdquo and ldquoSchoolingrdquo passages althoughstudents appeared to recall more units for the C than for the E versionsthese differences do not reach statistical signi cance These resultsindicate that familiar rhetorical conventions offer no guarantee thatcomprehension will improve a passage may be so easy or so hard that thefamiliar rhetorical conventions can have no additional effect

In the effort to modify texts so that they would seem natural for nativespeakers of English the passages ldquoChildcarerdquo and ldquoCharityrdquo needed moreextensive insertion of cohesive devices (discourse markers) and reorga-nization than did ldquoMale Nursesrdquo and ldquoSchoolingrdquo Predictably thenldquoChildcarerdquo and ldquoCharityrdquo might be expected to show greater effects forRhetorical Convention However further comparisons on RhetoricalConvention for the four passages on immediate and delayed recalls andthe three text perceptions concreteness comprehensibility and memo-rability as shown in the following section did not bear out thesepredictions Modi cation had an impact on ldquoCharityrdquo but not on

TABLE 3

Immediate-Recall and Delayed-Recall Scores by Rhetorical Convention and Passage

Version

Chinese English

Adjusted AdjustedPassagea mean SE mean SE F(1 221) p

Immediate recallldquoChildcarerdquo 314 129 286 130 248 1165ldquoCharityrdquo 333 128 249 131 2359 0001ldquoMale Nursesrdquo 343 131 285 130 1034 0015ldquoSchoolingrdquo 304 129 309 129 008 7840

Delayed recallldquoChildcarerdquo 215 101 192 102 272 1008ldquoCharityrdquo 241 101 187 103 1441 0002ldquoMale Nursesrdquo 244 104 203 102 856 0038ldquoSchoolingrdquo 223 101 204 101 169 1946

Note See note to Table 2aFor each passage N = 120

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 527

ldquoChildcarerdquo These ndings suggest that modi cation for local cohesiondid not have an effect on the recall and text perceptions of Chinese EFLreaders The changed location of topics and subtopicsmdashmacrostruc-turesmdashnot the insertion of local cohesion features were the key read-ability factors

In other words students recalled more about texts in which thethematic focus appeared midway or later in the text and in which logicalrelationships were linked implicitly rather than expressed explicitly inorganizational patterns (eg comparison-contrast or cause-effect) typi-cally found more readable by both L1 (Meyer amp Freedle 1984) and L2students (Carrell 1984) from Western countries

Passage Perception Questionnaire

The likelihood ratio chi-square values for the eight perception vari-ables were signi cant except for the value for Rhetorical ConventionNone of the passage perception scores even for items that relateddirectly to text organizationmdashclarity of argument text organization andrhetorical identitymdashshowed main effects of Rhetorical Convention or ofGrade Level Apparently then the students none of whom had anopportunity to compare two versions of a single passage did not perceiverhetorical differences between the C and the E versions Students gaveidentical scores (average = 31) for both versions when asked if thestructure or organization of the passage they had read was more typicalof Chinese writing or English writing style2

Only one perception rating memorability re ected an interactiveeffect for Rhetorical Convention and Grade Level F(1 226) = 528 p 05 (Appendix C) Further comparisons on rhetorical convention for thetwo grade levels showed that freshmen found texts re ecting Chineserhetorical conventions more memorable than passages modi ed torepresent English rhetorical conventions (M = 34 and 31 respectively)whereas senior subjects did not perceive a signi cant difference (M = 33and 34 respectively)

Three passage perception items did show an interactive effect ofRhetorical Convention and Passage concreteness F(3 226) = 419 p 01 comprehensibility F(3 226) = 396 p 01 and memorability F(3226) = 379 p 01 (Appendix C) As in the immediate-recall results

2 For the Passage Perception Questionnaire within-subject intraclass correlations accountedfor in the eight ANOVAs were interest 6 familiarity 5 concreteness 5 comprehensibility 5memorability 6 clarity of argument 5 organization 4 and rhetorical identity 6 As noted inthe Assumptions and Analyses section ve of the eight perceptions re ected an in uence fromPassage and were used as covariates in analyses of recall (see Appendix C)

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

528 TESOL QUARTERLY

further comparisons on Rhetorical Convention for the four passagesindicate that the passages ldquoCharityrdquo and ldquoMale Nursesrdquo produced themost reliable effects due to Rhetorical Convention For the ldquoCharityrdquopassage studentsrsquo ratings were signi cantly higher for the C version thanfor the E version on concreteness (M = 37 and 34 respectively F = 378p 05) comprehensibility (M = 38 and 34 respectively F = 820 p 05) and memorability (M = 33 and 29 respectively F = 605 p 05)For the ldquoMale Nursesrdquo passages students rated the C version signi cantlyhigher for concreteness (M = 40 and 35 respectively F = 686 p 05)and memorability (M = 37 and 33 respectively F = 624 p 05)Overall the preferences for the C versions of these passages correspondto the signi cant effects of rhetorical convention on immediate recall(For a discussion of topic effects as measured by readersrsquo perceptions ofa passage see Chu 1999 pp 89ndash108)

Topic Assessment Questionnaire

The results of the recall measures and the passage perception analysesare consistent Students recalled more from the C versions than from theE versions of two passages ldquoCharityrdquo and ldquoMale Nursesrdquo and they ratedthese passages more highly on concreteness and memorability than theother two passages ldquoChildcarerdquo and ldquoSchoolingrdquo The results of theTopic Assessment Questionnaire were used in a follow-up analysis of thedifferences among the passages The Topic Assessment Questionnaireasked participants to assess the interest and familiarity of the topics ofthe passages based on the title of the passage alone As such the topicitems are independent of the content of the passage and indicate thereadersrsquo baseline reactions to the general topics

The ldquoChildcarerdquo and ldquoSchoolingrdquo topics were rated as signi cantlymore interesting and more familiar than the ldquoCharityrdquo and ldquoMaleNursesrdquo topics For this analysis we combined scores for the topicsldquoCharityrdquo and ldquoMale Nursesrdquo and for the topics ldquoChildcarerdquo and ldquoSchool-ingrdquo ldquoChildcarerdquo and ldquoSchoolingrdquo scored signi cantly higher thanldquoCharityrdquo and ldquoMale Nursesrdquo on interestmdashcombined M = 39 versus 35F(1 226) = 2550 p 0001mdashand on familiaritymdashcombined M = 29versus 24 F(1 226) = 4250 p 0001

These results suggest that when students perceive topics as interestingand familiar as the students in this study perceived ldquoChildcarerdquo andldquoSchoolingrdquo they do not gain an advantage from familiar Chineserhetorical conventions In other words students seem to have moredif culty coping with English rhetorical conventions when the topic ofthe passage is less interesting and familiar to them

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 529

DISCUSSION

Effects of Rhetorical Conventions

The EFL students in this study recalled a signi cantly larger percent-age of text units from the four English texts written in Chinese rhetoricalconvention than they did from the four parallel texts written in Englishrhetorical convention in both immediate recall and delayed recall Thisresult suggests a robust in uence from an unfamiliar rhetorical conven-tion on foreign language reading comprehension If as argued in thisstudy rhetorical convention as a characteristic cultural artifact is deeplyrooted in many Chinese readersrsquo schemata of how a text is structuredthese readersrsquo own familiar text structure exerts an in uence when theyread an English text written according to Western conventions Thatin uence apparently obscures macro- and microfeatures of texts writtenin the style of alternative unfamiliar rhetorical conventions inhibitingtheir comprehension by Chinese speakers The ndings therefore corre-spond to those of Hinds (1984) and Eggington (1987) both of whichrevealed that for Oriental readers texts written in Oriental rhetoricalstructure yielded signi cantly greater reading recall than did textswritten in Western rhetorical structure

Indirectly the results of this study also correspond to Youngrsquos (1982) ndings that Chinese oral discourse structure posed a problem for thelistening comprehension of native English speakers Parallel ndings forspoken as well as written conventions suggest that as a result of years ofexposure to the conventions of Chinese rhetoric on the one hand anda pedagogical focus on the microfeatures of English texts for freshmenon the other macrostructures may also affect the way these readersprocess and reconstruct texts especially if those texts are writtenaccording to the rhetorical norms of different cultures

In a similarly indirect way the ndings of this study discon rm Mohanand Lorsquos (1985) speculation that EFL studentsrsquo problems in organizingwriting in both the L1 and the L2 are attributable to cognitive develop-ment rather than to interference from L1 practice or cultural expecta-tions If such conjecture were the case the seniors in this study shouldhave developed more skills than freshmen in recognizing differences inrhetorical practice Further as cognitively more mature students theywould presumably have been more adept in using both L1 and L2rhetorical conventions than would cognitively less mature freshmenparticipantsmdashan L1 nding of Meyer et al (1980) However the ndingsin this study are inconsistent with that assumption Although thecomprehension of seniors was higher overall negative transfer of L1rhetorical convention seemed to affect reading recall in the L2 to a

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

530 TESOL QUARTERLY

similarly signi cant degree regardless of the participantsrsquo distinctlydifferent developmental stages

Given the facilitating effects of familiar rhetorical conventions in therecall and retention of texts about topics deemed unfamiliar by theirreaders the results of this study support and extend the conclusionsmade in L1 reading studies (Meyer et al 1980 Meyer amp Freedle 1984van Dijk amp Kintsch 1983) as well as in L2 reading studies (Carrell 1984Tian 1990) that readers capable of using text-based logical structurescomprehended texts better than those who did not On the other handdata on readersrsquo perceptions about the textsrsquo content (in terms ofinterest familiarity concreteness memorability and comprehensibility)and textsrsquo form (in terms of thematic clarity organization and rhetoricalidentity) revealed that readers were not conscious of these strategies

The EFL participants in this study did not perceive an overalldifference between texts using two distinctly different rhetorical conven-tions This nding poses several intriguing questions Do rhetoricalstructures in uence comprehension at a level of automaticity Andwould training in metalinguistic strategies improve that comprehensionIf so Carrellrsquos (1984) suggestion that most ESL students particularlynon-Europeans may not possess the appropriate formal schemata toidentify rhetorical organization of an English text should be consideredin a pedagogical light If as results from recall data in this studydemonstrate a familiar rhetorical convention aids in comprehendingtexts then it follows that students might pro t from practice in identify-ing rhetorical structures unfamiliar to them

A reasonable inference from these ndings is that unfamiliar rhetori-cal conventions interfere with an L2 readerrsquos comprehension just as theydo with an L1 readerrsquos comprehension In this sense the ndings alsosubstantiate the conclusion that preferred rhetorical patterns of nativelanguages seem to interfere with ESL readersrsquo retention of English textsFurther although student recall re ected the impact of rhetoricalconvention readers could not distinguish between the rhetorical orien-tations of the texts That nding suggests that perception and cognitionmay be working at two different levels

The fact that effects of rhetorical convention held over time forseniors whereas the effects for freshmen declined slightly in delayedrecall suggests that the sustaining of effects of rhetorical convention overtime may correspond positively to language pro ciency (Carrell 1991)Such a conclusion is supported by the work of L2 researchers who havefound that the higher the language pro ciency of readers the more theyactivate higher level processing (Cooper 1984 Cziko 1978 1980Devine 1987 Hudson 1988) Senior students may be able to create amore robust macrostructure during the reading process and retain amore powerful retrieval structure thereafter than freshmen readers can

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 531

Whereas the freshmen perceived texts written in Chinese rhetoricalconvention as more memorable than those written in English rhetoricalconvention the seniors rated texts in both conventions as equallymemorable This null result for the more pro cient readers suggests thatwhen readers understand much of the textual message they perceive theC and E versions as equally memorable no matter what rhetoricalconvention they are written in On the other hand when comprehen-sion is less successful as was the case for freshman readers informationconveyed in a familiar rhetorical convention apparently promotes theperception of memorability in readers

Reader Factors Mediating the Effect of Rhetorical Convention

In conjunction with the signi cant impact for rhetorical conventionsthat re ect different cultural traditions this studyrsquos ndings also suggestthat rhetorical convention may have a more signi cant effect only whenreaders report signi cantly lower interest in and familiarity with thetopic Such indices were corroborated by the recall data because theycorrespond to the dimensions of familiarity and interest The fact thattopics rated lower in familiarity and interest show a rhetorical conventioneffect supports the proposals by Carrell (1987) and Roller (1990) thatprior knowledge may interact with the effect of text structure Furtherthese results echo studies assessing the relative effect of familiarityknowledge and text features on comprehension (Goldman Saul amp Coteacute1995 Salager-Meyer 1994 Schnotz 1984) In these studies text variableswere found to in uence comprehension only when readers read texts forwhich their familiarity and prior knowledge were low Stanovichrsquos (1980)interactive-compensatory model which proposes that de cits in one skillmay interfere with or lead to greater dependence on other processingskills might explain why we found no signi cant effects for students whoread materials with more familiar content but less familiar rhetoricalorganization Conceivably text structure became more important whenreaders lacked appropriate content schemata and as a result theybecame more dependent on rhetorical cues to construct meaning

IMPLICATIONS

Future Research

The caveats for this study suggest several avenues for replication andadditional research With regard to linguistic and cultural differencesthe present study was a one-way street comparing the effect of Chinese

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

532 TESOL QUARTERLY

rhetorical conventions on speakers of Chinese A study of these sameeffects on speakers of English would show whether these ndings holdacross cultures Because the subjects in this study were all English majorsand predominantly women future work might explore effects on stu-dents in other subject areas or on male readers Further Chinesespeakers not from Taiwan might respond differently to contrastingrhetorical use The responses of other language groups reading Englishas an L2 would con rm or discon rm the conclusion presented herethat mental representations of textual rhetoric seem to be culturallyconditioned

Because the reading passages and their counterpart revisions arecentral to the ndings in this study several possibilities remain to becon rmed or discon rmed Possibly rhetorical changes in other Chi-nese texts made along the lines suggested here would yield differentresults Experts in discourse analysis might make other more reliablerevisions or characterize differences in rhetorical conventions in differ-ent ways (eg Beck McKeown Sinatra amp Loxterman 1991 Britton VanDusen Gulgoz amp Glynn 1989) Further it would be useful to knowwhether similar effects hold for different genres such as poetry ctionalprose and drama

Rhetorical Structure and Cultural ThoughtProcesses in the Classroom

Phenomenological thinkers have long held that readersrsquo preexistentculturally conditioned ideas about the content and the presentationalstructures of a text in uence their capacity to understand that text(Ingarden 1973) These theorists suggest that what many in literarycriticism now refer to as a readerrsquos horizon of expectation (Jauss 1982) is acultural phenomenon based on human cognitive capabilities that areregionally conditioned rather than ontologically given As Kintsch (1998)elucidates ldquoCultural needs drive the unfolding of mental representa-tionsrdquo (p 29) Our ndings support philosophical and psycholinguisticclaims that readability is anchored in cultural expectations rather thanuniversally normed cognitive ones

This study in conjunction with those cited earlier in this article pointsto the need to teach cultural expectations related to the rhetoricalstructures that seem to in uence reading recall In that sense the ndings support recent initiatives in Taiwan and ESLEFL educationelsewhere that stress recognition of main ideas and discursive features oftexts If Western rhetorical conventions are viewed as cultural phenom-ena rather than as absolute norms for readability that are hardwired in

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 533

the brain this study adds to a growing body of evidence for teachingthose conventions

To teach culturally unfamiliar rhetorical practices however morethan training in recognition of main ideas and discursive features may benecessary Our study suggests that an essential rst step may be to spendclassroom time having students learn to distinguish between differentrhetorical styles Rather than identifying the main idea of only a singlepassage such work would involve locating the main idea in both nativeand target language passages in texts chosen for their differences in thisregard Students might then compare the placement of chief argumentsor examples in the two texts and distinguish between direct and lessdirect modes of expression Awareness of such differences would helpreaders develop metalinguistic strategies to adjust their expectationsabout textual messages If recall is a measure readersrsquo expectations inthis study were in uenced by rhetorical convention whether theyrealized it or not

Once ESLEFL readers recognize differences in rhetorical structuretechniques for teaching them to apply those macrostructures in holistictext base reconstruction (eg Kern 2000 Swaffar Arens amp Byrnes1991) might prove useful in facilitating retention of information Theseskills help students not only re-create but also articulate the messages oftexts Especially in an age of global communication the ability to use therhetorical conventions of another language may be a signi cant factor insuccessful exchange of ideas Over time the pedagogical implications ofongoing work in contrastive discourse analysis represented in journalssuch as Discourse and Society will help ESL researchers and teachersdevelop more de nitive models for characterizing differences in rhetori-cal conventions by culture and genre It behooves us in the TESOL eldto keep abreast of these developments and to recognize the signi canceof their implications for reading recall

ACKNOWLEDGMENTS

This article is drawn from the PhD dissertation of the rst author which wassupervised by the second author The latter contributed primarily to this paperrsquostheoretical design and its discourse and the third author primarily to its statisticaldesign and analyses We thank Providence University colleagues for their supportand assistance in facilitating this study For comments and suggestions on an earlierdraft of this article we thank Chun-chi Chen Pei-chi Chen Chiou-lan Chern Su-yueh Huang Diane Schallert Marilla Svinicki David Wright and two anonymousreaders

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

534 TESOL QUARTERLY

THE AUTHORS

Hsi-chin Janet Chu is an associate professor in the Department of English LanguageLiterature and Linguistics at Providence University Her teaching and researchinterests include EFL reading TEFL methodology and teacher education

Janet Swaffar teaches German language and literature with special attention to 20th-century German culture and the reading of texts in all media Her research emphasishas been on reading as an integrated component of language learning She writesabout ways to anchor language acquisition by having students apply the discourseand intentionalities they nd expressed in a variety of texts to address culturaldifferences and similarities

Davida Charney is a professor in the Division of Rhetoric and Composition at theUniversity of Texas at Austin Her research has appeared in Written CommunicationCollege Composition and Communication Journal of Business and Technical CommunicationTechnical Comunication Quarterly and Research in the Teaching of English

REFERENCES

Abrams M H (Ed) (1996) The Norton anthology of English literature (6th ed) NewYork W W Norton

Barnitz J G (1986) Toward understanding the effects of cross-cultural schemataand discourse structure on second language reading comprehension Journal ofReading Behavior 18 95ndash116

Beck I L McKeown M G Sinatra G M amp Loxterman J A (1991) Revisingsocial studies text from a text-processing perspective Evidence of improvedcomprehensibility Reading Research Quarterly 26 251ndash276

Bernhardt E B (1991) Reading development in a second language Norwood NJ AblexBormuth J R (1969) Factor validity of cloze tests as measures of reading compre-

hension ability Reading Research Quarterly 4 358ndash365Brennan A D Connie A B amp Winograd P N (1986) The effects of structural

variation on childrenrsquos recall of basal reader stories Reading Research Quarterly 2191ndash101

Britton B K Van Dusen L Gulgoz S amp Glynn S M (1989) Instructional textsrewritten by ve expert teams Revisions and retention improvements Journal ofEducational Psychology 81 226ndash239

Carrell P L (1984) The effects of rhetorical organization on ESL readers TESOLQuarterly 18 441ndash469

Carrell P L (1987) Content and formal schema in ESL reading TESOL Quarterly21 461ndash481

Carrell P L (1991) Second language reading Reading ability or languagepro ciency Applied Linguistics 12 159ndash179

Carrell P L (1992) Awareness of text structure Effects on recall Language Learning42 1ndash20

Carrell P L amp Wise T E (1998) The relationship between prior knowledge andtopic interest in second language reading Studies in Second Language Acquisition20 285ndash309

Chen S (1986) Argumentative discourse structure in Chinese and English writing Acomparative analysis Unpublished masterrsquos thesis Simon Fraser University BurnabyBritish Columbia Canada

Chou T-Y (1998) On the love of lotus In Chinese Textbook Compilation amp Review

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 535

Committee (Eds) Standard high school Chinese textbook II (pp 52ndash53) TaipeiTaiwan National Institute for Compilation and Translation

Chu H J (1999) The effects of culture-specic rhetorical conventions on the L2 reading recallof Chinese students Unpublished doctoral dissertation University of Texas atAustin

Clarke M (1980) The short circuit hypothesis of ESL readingmdashor when languagecompetence interferes with reading performance Modern Language Journal 64203ndash209

Cooper M (1984) Linguistic competence of practiced and unpracticed non-nativereaders of English In J C Alderson amp A H Urquhart (Eds) Reading in a foreignlanguage (pp 122ndash135) Essex England Longman

Cummins J (1979) Cognitiveacademic language pro ciency linguistic interde-pendence the optimum age question and some other matters Working Papers onBilingualism 19 197ndash205

Cziko G A (1978) Difference in rst and second language reading The use ofsyntactic semantic and discourse constraints The Canadian Modern LanguageReview 34 473ndash489

Cziko G A (1980) Language competence and reading strategies A comparison of rst and second language oral reading errors Language Learning 30 101ndash116

Devine J (1987) General language competence and adult second language readingIn J Devine P L Carrell amp D E Eskey (Eds) Research in reading in English as asecond language (pp 75ndash87) Washington DC TESOL

Eggington W G (1987) Written academic discourse in Korean Implications foreffective communication In U Connor amp R Kaplan (Eds) Writing acrosslanguage Analysis of L2 text (pp 141ndash152) Menlo Park CA Addison-Wesley

Goetz E T amp Armbruster B B (1980) Psychological correlates of text structure InJ S Rand C B Bertram amp W F Brewer (Eds) Theoretical issues in readingcomprehension Perspectives from cognitive psychology linguistics articial intelligenceand education (pp 201ndash220) Hillsdale NJ Erlbaum

Goldman S R Saul E U amp Coteacute N (1995) Paragraphing reader and task effectson discourse comprehension Discourse Processes 20 273ndash305

Hinds J (1984) Retention of information using a Japanese style of presentationStudies in Language 8 45ndash69

Hinds J (1987) Reader versus writer responsibility A new typology In U Connor ampR Kaplan (Eds) Writing across language Analysis of L2 text (pp 141ndash152) MenloPark CA Addison-Wesley

Hsiao S (1993a March 1) Bring back the tourists (P Newell Trans) SinoramaMagazine 2ndash3

Hsiao S (1993b June 1) Compassion nd joy in giving (P Newell Trans) SinoramaMagazine 2ndash3

Hsiao S (1993c July 1) Formosa today (P Newell Trans) Sinorama Magazine 3Hsiao S (1994 October 1) Dif culties and hard work (R Taylor Trans) Sinorama

Magazine 3Hudson T (1988) The effects of induce schemata on the ldquoshort circuitrdquo in L2

reading Non-decoding factors in L2 reading performance Language Learning 321ndash31

Ingarden R (1973) The cognition of the literary work of art (R A Crowley amp K R OlsonTrans) Evanston IL Northwestern University Press

Jauss H R (1982) Toward an aesthetic of reception Minneapolis University ofMinnesota Press

Jensen J V (1998) Teaching East Asian rhetoric The Rhetoric Society Quarterly 18136ndash149

536 TESOL QUARTERLY

Johnson R E (1970) Recall of prose as a function of the structural importance ofthe linguistic units Journal of Verbal Learning and Behavior 9 12ndash20

Kachru Y (1998) Culture and argumentative writing in world Englishes In L Smithamp R Forman (Eds) World English 2000 (pp 48ndash67) Honolulu University ofHawaii Press

Kaplan R B (1966) Cultural thought patterns in intercultural education LanguageLearning 16 1ndash20

Kern R (2000) Literacy and language teaching Oxford Oxford University PressKintsch W (1998) Comprehension A paradigm for cognition New York Cambridge

University PressKintsch W amp Yarbrough J C (1982) Role of rhetorical structure in text

comprehension Journal of Educational Psychology 74 823ndash834Kirkpatrick A (1995) Chinese rhetoric Methods of argument Multilingua 149

271ndash295Kojima K (1972) Chung kuo wen shuei tung lun [General introduction to Chinese

literature] (L K Sun Trans) Taipei Taiwan Shan-wuLee J F (1986) Background knowledge and L2 reading The Modern Language

Journal 54 350ndash354Mead M (1979) Are men more creative than women In M C Bateson amp R Met-

raux (Eds) Margaret Mead Some personal views (pp 167ndash170) New York WalkerMeyer B F J Brandt D M amp Bluth G J (1980) Use of top-level structure in text

Key for reading comprehension of ninth-grade students Reading Research Quar-terly 16 72ndash103

Meyer B F J amp Freedle R O (1984) Effects of discourse type on recall AmericanEducational Research Journal 21 121ndash143

Mohan B amp Lo W A (1985) Academic writing and Chinese student Transfer anddevelopmental factors TESOL Quarterly 19 515ndash532

Normant N (1986) Organizational structure of Chinese subjects writing in Chineseand in ESL Journal of the Chinese Language Teachers Association 21(3) 49ndash72

Raymond P M (1993) The effects of structure strategy training on the recall ofexpository prose for university students reading French as a second language TheModern Language Journal 77 445ndash458

Ricento T (1989) An analysis of the rhetorical structure of English and Japaneseeditorials Issues and Developments in English and Applied Linguistics 4 51ndash67

Roller M R (1990) The interaction between knowledge and structure variables inthe processing of expository prose Reading Research Quarterly 25 79ndash89

Rumelhart D E (1977) Towards an interactive model of reading In S Dornic(Ed) Attention and performance (Vol 6 pp 537ndash603) Hillsdale NJ Erlbaum

Salager-Meyer F (1994) Reading medical English abstracts A genre study of theinteraction between structural variables and the readerrsquos linguistico-conceptualcompetence (L2) Journal of Research in Reading 17 120ndash146

Schnotz W (1984) Comparative instructional text organization In H MandleN L Sein amp T Trabasso (Eds) Learning and comprehension of text (pp 53ndash75)Hillsdale NJ Erlbaum

Spyridakis J H amp Wenger M J (1991) An empirical method of assessing topicfamiliarity in reading comprehension research British Educational Research Journal17 353ndash360

Stanovich K E (1980) Toward an interactive-compensatory model of individualdifferences in the development of reading uency Reading Research Quarterly 1632ndash71

Swaffar J Arens K amp Byrnes H (1991) Reading for meaning An integrated approachto language learning Englewood Cliffs NJ Prentice Hall

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 537

Tian G S (1990) The effects of rhetorical organization in expository prose on ESLreaders in Singapore RELC Journal 21(2) 1ndash13

van Dijk T amp Kintsch W (1983) Strategies of discourse comprehension New YorkAcademic Press

Weaver C A III amp Kintsch W (1991) Expository text In R Barr M L Kamil P Mos-enthal amp P D Pearson (Eds) Handbook of reading research (Vol 2 pp 230ndash245)White Plains NY Longman

Wu Y (1988) Wenzhang jiegouxue [The structure of text] Beijing Peoplersquos Republicof China Zhongguo Renmin Daxue Chubanshe

Young L (1982) Inscrutability revisited In J Gumperz (Ed) Language and socialidentity (pp 72ndash85) Cambridge Cambridge University Press

APPENDIX A

Rhetorical Conventions in Western

Expository Style and Qi-Cheng-Zhuan-HeTexts A and B illustrate respectively the characteristics of Western expository style and of qi-cheng-zhuan-he with regard to (a) where the topic sentence is located (b) how subsets ofinformation are ordered (c) whether or not discourse cues render explicit the hierarchy andrelationships of ideas and (d) whether explicit conclusions are drawn based on the foregoing

Text AThere are three possible positions one can take about male and female creativity The rst isthat males are inherently more creative in all elds The second is that if it were not for thegreater appeal of creating and cherishing young human beings females would be as creative asmales If this were the case then if men were permitted the enjoyment women have always hadin rearing young children male creativity might be reduced also (There is some indication inthe United States today that this is so) The third possible position is that certain forms ofcreativity are more congenial to one sex than to the other and that the great creative acts willtherefore come from only one sex in a given eld (Mead 1979 pp 167ndash170)

Text BOf all kinds of owers in the grass and on the trees in the world many are very lovely Tao Yuan-ming in Chin Dynasty favors chrysanthemum Ever since the Lee and Tang Dynasties worldlypeople favor the peony I favor the lotus because it grows out of mud but is not stained by mudand it is washed by the water and does not appear sensual It is straight going without branchesThe farther the fragrance spreads the more refreshing it is Every single bud stands out straightIt is best to appreciate it from a distance but not to lay a hand on it

I would say this Chrysanthemum the hermit of the owers peony the opulent of the owerslotus the gentleman of the owers Aye the love of chrysanthemum is scarcely heard since TaoYuan-ming the love of lotus who else but me the love of peony countless people (Chou1998 pp 52ndash53 translated by the rst author)

AnalysisReaders unfamiliar with the rhetorical conventions of qi-cheng-zhuan-he may not have concludedwhat for most Chinese readers will be self-evidentmdashnamely that Text B is a meditation on therelative merits of aesthetic concerns and material wealth In Western writing an explicitargument in a thesis statement is valued as good writing style (Meadrsquos ldquothree possible positionsrdquoin the initial sentence of Text A) in traditional Chinese writing suggesting a thesis in anoblique reference is favored in order to leave the reader room for re ection (Choursquos ldquoI favorthe lotusrdquomdasha metaphoric reference to that which is unstained and pure) In Text A the writerstates three options for thinking about gender and creativity In Text B an option is implied inthe last sentence Instead of pursuing fortune people should re ect on the spirit of a hermitand a gentleman

538 TESOL QUARTERLY

APPENDIX B

Questionnaires

Passage Perception Questionnaire (Translated From Chinese)Directions We are interested in knowing how you perceive the article you have just read Pleaserate your perception of it based on your subjective feelings Circle the number that best answersthe question Circle only one number for each question

1 How interesting did you nd this text to readVery Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

2 How familiar did you nd the topic and content to beVery Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

3 How concrete did you nd the content to be That is how easy or hard is it for you toform a mental image

Very Somewhat Somewhat Veryabstract abstract Neutral concrete concrete

1 2 3 4 5

4 How dif cult was this article for you to read and understandVery Somewhat Not too

dif cult dif cult Neutral dif cult Very easy1 2 3 4 5

5 How dif cult do you think this text will be for you to rememberVery Somewhat Not too

dif cult dif cult Neutral dif cult Very easy1 2 3 4 5

6 How clear was the main line of thought or the main argument of the textVery Somewhat Mostly

unclear unclear Neutral clear Very clear1 2 3 4 5

7 How organized do you think the text was that you readVery Somewhat Somewhat Very

unorganized unorganized Neutral organized organized1 2 3 4 5

8 Is the structureorganization of this text more similar to that of Chinese writing or Englishwriting

Very much like Somewhat like Somewhat like Very much likethe structure of the structure of the structure of the structure ofChinese writing Chinese writing Not sure English writing English writing

1 2 3 4 5

Items 1 4 5 6 and 8 adapted from Reader Assessment of Text (Raymond 1993)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 539

Topic Assessment Questionnaire (Translated From Chinese)Age_____________ Gender____________

Have you ever gone to schools abroad _____________ If yes in which country __________

When ______________ For how long __________________

We are interested in knowing how you think about the following four topics Childcareproblems for working mothers abuse of charitable institutions male nurses and alternativeschools For each of the four topics please rate your perception based on your subjectivefeelings Circle the number that best answers the question Circle only one number for eachquestion

1 Interesta Childcare Problems for Working Mothers

Very Somewhat Somewhat Veryuninteresting uninteresting Neutral interesting interesting

1 2 3 4 5

b Abuse of Charitable InstitutionsVery Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

c Male NursesVery Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

d Alternative Schools (eg Forest School)Very Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

2 Familiaritya Childcare Problems for Working Mothers

Very Somewhat Somewhat Veryunfamiliar unfamiliar Neutral familiar familiar

1 2 3 4 5

b Abuse of Charitable InstitutionsVery Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

c Male NursesVery Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

d Alternative Schools (eg Forest School)Very Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

540 TESOL QUARTERLY

AP

PE

ND

IX C

Ana

lysi

s of

Var

ianc

e fo

r E

ight

Pas

sage

Per

cept

ions

Sour

ce a

nd

df

RG

PG

R

P

GP

R

P

G

Rc2

Per

cept

ion

(1 2

26)

(1 2

38)

(3 2

26)

(1 2

26)

(3 2

26)

(3 2

26)

(3 2

26)

(1)

Inte

rest

175

277

652

466

034

066

046

232

3

Fam

iliar

ity

229

010

834

015

068

250

125

120

2

Con

cret

enes

s1

140

262

560

010

134

19

022

107

4

Com

pre

hen

sibi

lity

058

079

332

1

041

193

96

093

149

6

M

emor

abili

ty2

810

354

91

528

2

383

79

167

212

6

C

lari

ty0

141

172

080

031

401

111

7411

52

O

rgan

izat

ion

181

001

336

0

002

021

040

243

35

Rh

etor

ical

ori

enta

tion

000

030

021

014

388

0

320

100

08

Not

e R

esu

lts

are

F ra

tios

R =

Rh

etor

ical

Con

ven

tion

G =

Gra

de L

evel

P =

Pas

sage

p

0

5

p

01

p 0

01

p

0

001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 541

APPENDIX D

Analysis of Covariance for Immediate and Delayed Recall

Effect df F p

Immediate recallPerceived interest 1 221 404 0456Perceived familiarity 1 221 097 3267Perceived comprehensibility 1 221 602 0149Perceived memorability 1 221 001 9398Perceived organization 1 221 329 0710R 1 221 3111 lt 0001G 1 238 7224 lt 0001P 3 221 136 2557G R 1 221 233 1284P G 3 221 004 7520P R 3 221 406 0078P G R 3 221 037 7753

Delayed recallPerceived interest 1 221 611 0142Perceived familiarity 1 221 018 6698Perceived comprehensibility 1 221 996 0018Perceived memorability 1 221 016 6860Perceived organization 1 221 113 2884R 1 221 3694 lt 0001G 1 238 6271 lt 0001P 3 221 154 2041G R 1 221 724 0077P G 3 221 054 6545P R 3 221 120 3120P G R 3 221 061 6079

Note R = Rhetorical Convention G = Grade Level P = Passage c2(1) = 4122 for immediaterecall and 5853 for delayed recall Perceived interest perceived familiarity perceivedcomprehensibility perceived memorability and perceived organization are ve covariatesp 05 p 01 p 001 p 0001

538 TESOL QUARTERLY

APPENDIX B

Questionnaires

Passage Perception Questionnaire (Translated From Chinese)Directions We are interested in knowing how you perceive the article you have just read Pleaserate your perception of it based on your subjective feelings Circle the number that best answersthe question Circle only one number for each question

1 How interesting did you nd this text to readVery Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

2 How familiar did you nd the topic and content to beVery Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

3 How concrete did you nd the content to be That is how easy or hard is it for you toform a mental image

Very Somewhat Somewhat Veryabstract abstract Neutral concrete concrete

1 2 3 4 5

4 How dif cult was this article for you to read and understandVery Somewhat Not too

dif cult dif cult Neutral dif cult Very easy1 2 3 4 5

5 How dif cult do you think this text will be for you to rememberVery Somewhat Not too

dif cult dif cult Neutral dif cult Very easy1 2 3 4 5

6 How clear was the main line of thought or the main argument of the textVery Somewhat Mostly

unclear unclear Neutral clear Very clear1 2 3 4 5

7 How organized do you think the text was that you readVery Somewhat Somewhat Very

unorganized unorganized Neutral organized organized1 2 3 4 5

8 Is the structureorganization of this text more similar to that of Chinese writing or Englishwriting

Very much like Somewhat like Somewhat like Very much likethe structure of the structure of the structure of the structure ofChinese writing Chinese writing Not sure English writing English writing

1 2 3 4 5

Items 1 4 5 6 and 8 adapted from Reader Assessment of Text (Raymond 1993)

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 539

Topic Assessment Questionnaire (Translated From Chinese)Age_____________ Gender____________

Have you ever gone to schools abroad _____________ If yes in which country __________

When ______________ For how long __________________

We are interested in knowing how you think about the following four topics Childcareproblems for working mothers abuse of charitable institutions male nurses and alternativeschools For each of the four topics please rate your perception based on your subjectivefeelings Circle the number that best answers the question Circle only one number for eachquestion

1 Interesta Childcare Problems for Working Mothers

Very Somewhat Somewhat Veryuninteresting uninteresting Neutral interesting interesting

1 2 3 4 5

b Abuse of Charitable InstitutionsVery Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

c Male NursesVery Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

d Alternative Schools (eg Forest School)Very Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

2 Familiaritya Childcare Problems for Working Mothers

Very Somewhat Somewhat Veryunfamiliar unfamiliar Neutral familiar familiar

1 2 3 4 5

b Abuse of Charitable InstitutionsVery Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

c Male NursesVery Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

d Alternative Schools (eg Forest School)Very Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

540 TESOL QUARTERLY

AP

PE

ND

IX C

Ana

lysi

s of

Var

ianc

e fo

r E

ight

Pas

sage

Per

cept

ions

Sour

ce a

nd

df

RG

PG

R

P

GP

R

P

G

Rc2

Per

cept

ion

(1 2

26)

(1 2

38)

(3 2

26)

(1 2

26)

(3 2

26)

(3 2

26)

(3 2

26)

(1)

Inte

rest

175

277

652

466

034

066

046

232

3

Fam

iliar

ity

229

010

834

015

068

250

125

120

2

Con

cret

enes

s1

140

262

560

010

134

19

022

107

4

Com

pre

hen

sibi

lity

058

079

332

1

041

193

96

093

149

6

M

emor

abili

ty2

810

354

91

528

2

383

79

167

212

6

C

lari

ty0

141

172

080

031

401

111

7411

52

O

rgan

izat

ion

181

001

336

0

002

021

040

243

35

Rh

etor

ical

ori

enta

tion

000

030

021

014

388

0

320

100

08

Not

e R

esu

lts

are

F ra

tios

R =

Rh

etor

ical

Con

ven

tion

G =

Gra

de L

evel

P =

Pas

sage

p

0

5

p

01

p 0

01

p

0

001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 541

APPENDIX D

Analysis of Covariance for Immediate and Delayed Recall

Effect df F p

Immediate recallPerceived interest 1 221 404 0456Perceived familiarity 1 221 097 3267Perceived comprehensibility 1 221 602 0149Perceived memorability 1 221 001 9398Perceived organization 1 221 329 0710R 1 221 3111 lt 0001G 1 238 7224 lt 0001P 3 221 136 2557G R 1 221 233 1284P G 3 221 004 7520P R 3 221 406 0078P G R 3 221 037 7753

Delayed recallPerceived interest 1 221 611 0142Perceived familiarity 1 221 018 6698Perceived comprehensibility 1 221 996 0018Perceived memorability 1 221 016 6860Perceived organization 1 221 113 2884R 1 221 3694 lt 0001G 1 238 6271 lt 0001P 3 221 154 2041G R 1 221 724 0077P G 3 221 054 6545P R 3 221 120 3120P G R 3 221 061 6079

Note R = Rhetorical Convention G = Grade Level P = Passage c2(1) = 4122 for immediaterecall and 5853 for delayed recall Perceived interest perceived familiarity perceivedcomprehensibility perceived memorability and perceived organization are ve covariatesp 05 p 01 p 001 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 539

Topic Assessment Questionnaire (Translated From Chinese)Age_____________ Gender____________

Have you ever gone to schools abroad _____________ If yes in which country __________

When ______________ For how long __________________

We are interested in knowing how you think about the following four topics Childcareproblems for working mothers abuse of charitable institutions male nurses and alternativeschools For each of the four topics please rate your perception based on your subjectivefeelings Circle the number that best answers the question Circle only one number for eachquestion

1 Interesta Childcare Problems for Working Mothers

Very Somewhat Somewhat Veryuninteresting uninteresting Neutral interesting interesting

1 2 3 4 5

b Abuse of Charitable InstitutionsVery Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

c Male NursesVery Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

d Alternative Schools (eg Forest School)Very Somewhat Somewhat Very

uninteresting uninteresting Neutral interesting interesting1 2 3 4 5

2 Familiaritya Childcare Problems for Working Mothers

Very Somewhat Somewhat Veryunfamiliar unfamiliar Neutral familiar familiar

1 2 3 4 5

b Abuse of Charitable InstitutionsVery Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

c Male NursesVery Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

d Alternative Schools (eg Forest School)Very Somewhat Somewhat Very

unfamiliar unfamiliar Neutral familiar familiar1 2 3 4 5

540 TESOL QUARTERLY

AP

PE

ND

IX C

Ana

lysi

s of

Var

ianc

e fo

r E

ight

Pas

sage

Per

cept

ions

Sour

ce a

nd

df

RG

PG

R

P

GP

R

P

G

Rc2

Per

cept

ion

(1 2

26)

(1 2

38)

(3 2

26)

(1 2

26)

(3 2

26)

(3 2

26)

(3 2

26)

(1)

Inte

rest

175

277

652

466

034

066

046

232

3

Fam

iliar

ity

229

010

834

015

068

250

125

120

2

Con

cret

enes

s1

140

262

560

010

134

19

022

107

4

Com

pre

hen

sibi

lity

058

079

332

1

041

193

96

093

149

6

M

emor

abili

ty2

810

354

91

528

2

383

79

167

212

6

C

lari

ty0

141

172

080

031

401

111

7411

52

O

rgan

izat

ion

181

001

336

0

002

021

040

243

35

Rh

etor

ical

ori

enta

tion

000

030

021

014

388

0

320

100

08

Not

e R

esu

lts

are

F ra

tios

R =

Rh

etor

ical

Con

ven

tion

G =

Gra

de L

evel

P =

Pas

sage

p

0

5

p

01

p 0

01

p

0

001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 541

APPENDIX D

Analysis of Covariance for Immediate and Delayed Recall

Effect df F p

Immediate recallPerceived interest 1 221 404 0456Perceived familiarity 1 221 097 3267Perceived comprehensibility 1 221 602 0149Perceived memorability 1 221 001 9398Perceived organization 1 221 329 0710R 1 221 3111 lt 0001G 1 238 7224 lt 0001P 3 221 136 2557G R 1 221 233 1284P G 3 221 004 7520P R 3 221 406 0078P G R 3 221 037 7753

Delayed recallPerceived interest 1 221 611 0142Perceived familiarity 1 221 018 6698Perceived comprehensibility 1 221 996 0018Perceived memorability 1 221 016 6860Perceived organization 1 221 113 2884R 1 221 3694 lt 0001G 1 238 6271 lt 0001P 3 221 154 2041G R 1 221 724 0077P G 3 221 054 6545P R 3 221 120 3120P G R 3 221 061 6079

Note R = Rhetorical Convention G = Grade Level P = Passage c2(1) = 4122 for immediaterecall and 5853 for delayed recall Perceived interest perceived familiarity perceivedcomprehensibility perceived memorability and perceived organization are ve covariatesp 05 p 01 p 001 p 0001

540 TESOL QUARTERLY

AP

PE

ND

IX C

Ana

lysi

s of

Var

ianc

e fo

r E

ight

Pas

sage

Per

cept

ions

Sour

ce a

nd

df

RG

PG

R

P

GP

R

P

G

Rc2

Per

cept

ion

(1 2

26)

(1 2

38)

(3 2

26)

(1 2

26)

(3 2

26)

(3 2

26)

(3 2

26)

(1)

Inte

rest

175

277

652

466

034

066

046

232

3

Fam

iliar

ity

229

010

834

015

068

250

125

120

2

Con

cret

enes

s1

140

262

560

010

134

19

022

107

4

Com

pre

hen

sibi

lity

058

079

332

1

041

193

96

093

149

6

M

emor

abili

ty2

810

354

91

528

2

383

79

167

212

6

C

lari

ty0

141

172

080

031

401

111

7411

52

O

rgan

izat

ion

181

001

336

0

002

021

040

243

35

Rh

etor

ical

ori

enta

tion

000

030

021

014

388

0

320

100

08

Not

e R

esu

lts

are

F ra

tios

R =

Rh

etor

ical

Con

ven

tion

G =

Gra

de L

evel

P =

Pas

sage

p

0

5

p

01

p 0

01

p

0

001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 541

APPENDIX D

Analysis of Covariance for Immediate and Delayed Recall

Effect df F p

Immediate recallPerceived interest 1 221 404 0456Perceived familiarity 1 221 097 3267Perceived comprehensibility 1 221 602 0149Perceived memorability 1 221 001 9398Perceived organization 1 221 329 0710R 1 221 3111 lt 0001G 1 238 7224 lt 0001P 3 221 136 2557G R 1 221 233 1284P G 3 221 004 7520P R 3 221 406 0078P G R 3 221 037 7753

Delayed recallPerceived interest 1 221 611 0142Perceived familiarity 1 221 018 6698Perceived comprehensibility 1 221 996 0018Perceived memorability 1 221 016 6860Perceived organization 1 221 113 2884R 1 221 3694 lt 0001G 1 238 6271 lt 0001P 3 221 154 2041G R 1 221 724 0077P G 3 221 054 6545P R 3 221 120 3120P G R 3 221 061 6079

Note R = Rhetorical Convention G = Grade Level P = Passage c2(1) = 4122 for immediaterecall and 5853 for delayed recall Perceived interest perceived familiarity perceivedcomprehensibility perceived memorability and perceived organization are ve covariatesp 05 p 01 p 001 p 0001

CULTURAL REPRESENTATIONS OF RHETORICAL CONVENTIONS 541

APPENDIX D

Analysis of Covariance for Immediate and Delayed Recall

Effect df F p

Immediate recallPerceived interest 1 221 404 0456Perceived familiarity 1 221 097 3267Perceived comprehensibility 1 221 602 0149Perceived memorability 1 221 001 9398Perceived organization 1 221 329 0710R 1 221 3111 lt 0001G 1 238 7224 lt 0001P 3 221 136 2557G R 1 221 233 1284P G 3 221 004 7520P R 3 221 406 0078P G R 3 221 037 7753

Delayed recallPerceived interest 1 221 611 0142Perceived familiarity 1 221 018 6698Perceived comprehensibility 1 221 996 0018Perceived memorability 1 221 016 6860Perceived organization 1 221 113 2884R 1 221 3694 lt 0001G 1 238 6271 lt 0001P 3 221 154 2041G R 1 221 724 0077P G 3 221 054 6545P R 3 221 120 3120P G R 3 221 061 6079

Note R = Rhetorical Convention G = Grade Level P = Passage c2(1) = 4122 for immediaterecall and 5853 for delayed recall Perceived interest perceived familiarity perceivedcomprehensibility perceived memorability and perceived organization are ve covariatesp 05 p 01 p 001 p 0001