ñ

cENTRo DE LlNoÜísncA APLIcADAMnriñruo DE crer¡oÁ,,nffi:%?i^" Meoro A^ BTENTE

ACTAS-I

slMPoSlo I NTEBNACIONALcoMuNlcAcloN soclAl

Edkjon: LÉ{}nelRuiz lvliyares, Alex Muñoz Alvarado y celiq Átvarez Moreno

C¡.tierta: Emtr" Pérez Yero

@ Sobe la Presente edición:- ÁCrnS - t, x S¡mposio Internacional de Comunicación Social

Santiago de Cuba, 22-26 de enero del 2007

Centro de Lingüística Apticada, 2007

ISBN: 959-7174-08-1

CENTRO DE LINGÜISTICA APLICADADelegación Territorial del CITMAAoartado Postal 4067. Vista Alegre'Sánt iago de Cuba. Cuba C'P' 90400óórreo'etectrónicq: leonel@lingapli'ciges'inf'cu

-08-1

ffilil4

4

rsBN 959-717

ffiIffiffifiil

BearÉ Arias ÁlvarezÚeatización de la consonante ¿implosiva? 184

Carla Montez Fernandesindexagáo vídeo e anotagáo tinguística de criaqÓes

multimedia 189

coreográficas contemporáneas: para uma ontologia

María Carmen AmaYa GalvánInformation and commui¡Jai¡on recnnology (lcr) in coltaborative Translation Teaching: Legal and Economic

Translation 194

Celia María Pérez Marqués Y otrosLa riqueza iéxica delos esco/ares guamenses tras ta aplicación de tos Programas de la RevotuciÓn en la escuela

primaria. Estudio Preliminar 198

CharalamPos MagoulasEpistemología de Ia ¡nterd¡sciplinaridad: de la filosofía aristotética a ta lingüística y la semiÓtica l1o4

Maria Cristina SeccitÁ Ángua de Jorge lbargüengoitia y su traducción l1o7

María Cristina Toledo Báez' Áii" ¡,áii I éitr"cti o n a n d d oc u m e ntatio.n process i n tr a n sl ati o n | 1 I 2

DanielaZizito" iróÁ^Or"s de "coftesía" y de "confianza" entre América, España y Cerdeña 1117

Daniel Zubía FernándezLos glrasoles se viran o la naturaleza en O Esplendor de Portugal /122

Dermeval Da HoraÁ-moitongagao na produgáo escrita: reflexo da fala? 1127

Diana Alexand ra Pérez MorenoMotivación léxica en

"t t¡itiiÚt a tavés del intercambio comercial novohispano 1132

Danilo De Luise Y Mara Morell iPuentes de Palabras 1137

Eloína Miyares Bermúdez y Vitelio Ruiz HernándezE! programa radiat "sanár Ándar", un esfuerzo páÁ me¡orar ta ortografía de los alumnos de secundaria básica e

Áói¿ií", b cultura idiomática de nuestro pueblo 1142

Andre Virginie Y Emmanuelle CanutEtudier les pratiques nióáói¿rii

"Ábs pour mieux appréhender les phénoménes de communication sociale:

!,exempledes repnses ei dles reformulations dans les interactions verbales 1145

Esmeralda Vailati NegráoLicenciamento por modificagáo: o papel das relativas l15O

Evelyn Breiteneder Y otrosÁei - nustr¡an Acaáemy Corpus and its Maior Applications 1155

Francisco G. Hernández HernándezAround fhe sounds /158

Francisco Javier Grande Alijatiriu"s u,t¡fi"i"t"r,a priorii, diccionarios y la clasificación del léxico 1162

F.X. Sligo y otrosPicturing words: tllustration-based research feedback to persons of low literacy 1168

Frey Vega VeranesLas estrategtas de composiciÓn en los fexfos escrlfos académicos de segundo nivel 1173

Gema Valdés AcostaE|estudiodelabantutdadtingüísticaenCuba.Problemasysoluciones|178

vl

',iMARÍA CRISTINA TOLEDO BAEZDepartment of Translat ion and InterpretingUniversity of MalagaMalaga, $paintoledo@uma.es

. Automatic extraction and documentation process in translation

l . lntroduct ionAs Lavid (2005) points out, information has become one of the basic elements in our current society, which maybe called the "Third Wave", paraphrasing Alvin Toffler's book (1996). First wave is the society after agrarianrevolution; Second wave is industrial. Third Wave represents information and knowledge revolution. Newmillennium's society is "information society", where Information and Communication(s) Technology (lCT) has aparamount importance. Therefore; the exchange of Ianguages and cultures plays an important role jn thisinformation society. Consequently, translators and interpreters may become fundamental mediators at globallevels.In this context, Internet appears as an essential tool since it offers new ways of communication and scientif icknowledge spreading. In addition, it facil i tates añd improves the documentation process. The translator, as aninformation user and an information producer, considers Internet to be a valuable documentation source and auseful cqmmunication system.According to Pinto Molina (2002: 2), the "informational revolution" makes possible to compile more information inless time and, consequently, to improve translator's efficiency. With the mushrooming of the quantity of on-linetext information, tr¡ggered in part by the groMh of the World Wide Web, it is especially useful to have tools whichcan help users digest information content.Nevertheless, translators have to be really skil l ful during documentation process since they need to be able todistinguish and choose only reliable information resources, This is because Internet, although it is a valuable andreally useful tool, contains much unrel¡able information.In that regard, an abstract may be quite useful for translators since it helps to choose the correct information in thedocumentation process. Given that translators normally have to meet tight deadlines, abstracting articles orelectron¡c resources is an advantageous solution and heavily facil i tates the translation process.

In this paper', we approximate to automatic text extraction in two languages (English and Spanish) by means oftwo elements:1. Terms and conditions of package tours -specifically cruises- extracted from multi l ingual macrocorpus Turicor,

which belongs to the project TURICOR: A multilingual corpus of tourism contracts (German, Spanish,English, ltalian) for automatic text generation and legal translation lTuricor: compilación de un corpus decontratos turisticos (alemán, español, inglés, italiano) para la generación textual multilingüe y Ia traducciónjurídical (Ref. no. BBF2003-04616 (2003-2006, Spanish Ministry of Science and Technology).

2. Six multilingual and free automatic extract¡on systems: Copernic Summarizer, Extractor, GisfSumm, MicrosoftWord Summarization. SweSum. and WebSumm.

Once the texts are automatically extracted, advantages and disadvantages of automatic extraction were analyzedin order tc check their impact and contribution to translators' documentation process.

2. AG: between NLG and AGInitially, Natural Language Generation (NLG) and Abstracts Generation (AG), both belonging to the fielf of NaturalLanguage Processing (NLP), could be considered to have the same characteristics since both are related toautomatic language generation. In fact, some'researchers think that AG is part of NLG (Lavid, 2005.216).However, other researchers have a different point of view. Regarding NLG, Reiter and Dale (1997:57) define it as:

Natural language generation is the subfield of artif icial intell igence and computational l inguistics that is concernedwith the construction of computer systems that can produce understandable texts in [,,.] human languages fromsome underlying non-linguistic representat¡on of information.

Concerning AG, Acero et al. (2001: 282) give the following definit ion:Por generación automát¡ca de resúmenes de texto entendemos el proceso por el cual se identif ica la informaciónsustancial proveniente de una fuente (o varias) para producir una versión abreviada destinada a un usuariopart¡cular (o grupo de usuarios) y una tarea (o tareas).

Therefore, NLG and AG share the same aims, i,e., text generation, but they use different information resourcesbecause NLG is based on non- l inguist ic representat ion of informat ion and AG is based on a l inguist icrepresentat¡on. However, we consider AG to be closely related to NLG and, consequently, they both belong tothe Translat ion Technologies (TT) f ie id. AG rs at r ts very peak due to the huge amount of informat ion on lnternetand the necessity of f i l tering the electronic resources and information.

'The research reported in this paper has been part¡al ly carr ied out in the framework of R & D projects Ref. nos. BBF2O03-04616(Fund¡ng source: Spanish Ministry of Science and Technology, National R&D Program, and European Union, EuropeanRegional Development Fund- ERDF. 2003-2006) and HUlt/-892 (Andalusian Ministry for Educalion, Spain. 2006-2009)

112

lnSpainthemostrepresentat ive' : : : l ' tT:Ltextautomattctext 'summarizat ionarebeingconduc:: i l : : : :

YIJÍUll,?1,"iJfl.,?:KJnÉTY,?t:""!'j:!ii:TüíI[:¡',"í"'ft itt':t"':filH;i*1tt**:Jiíi'Jíg'fffii,niíll,""=",i.lJJilJAi#i jui:fi':;IJ;;rñ'.Tiiil'""J:ÍJ'+;;;;;;1eBB,3. Automatic summarizatio v 2001: 239) point out that automatic.sumll:11:"t

of text is a

n'"TÍrur""":ñrtttlf ntgf ü#i#d¡ilüdi+d::i**''*'t#{rqi'!5:l*'*iltr"*itááioi

of informatio

::,T,[3'fifi:n5T,ll:iiJlf"':ni'*i%Et$J"Tliil'l¡;¿:*n:::l;] iJ'ffi;iJ"'lomiiar za"n s

*$i:j:*l'lü"';T=l.i; ill$il[i'#[t¿ir¡rr¡ll¿+,:**,**¿iru*:.'l'''''Énn' ¡;1:¡f ' :il:ffi',,'^ffi :S, L:',1 ;f li

"J : i $ii:ii","*""il ;' u t r o o o ¡

content of a text, even inouvr | ,1,: . , . : t : ." , t t i : : . . : , . : : , . ,

Anal)5is

".,o ¡:¡$i'.:"rrni*fi:i=#"üi*ffial 1999; Marcu, zuuu¡ 'iio',,'J;' i;o¡2, 32) uo't'u"t'ü: +fl:'n:n

.,;3li'":"lii5i"yiil'¡;r.;:t¿¡;iixi:i1fl"["'']lll 'o""u' on

parses senteng!1. j]l)3,""n,"offiffi ún-guug" understanding

a: ,.:

Synth.ttsf r¡nifolm¡tlon

á-#oi' aooo' sol

lnthlspaperweWil lana|yzeautomat icextract ionbecausei t is iheonlyavai lableopt ioninautomat lc'=lt , .n"tüt i iá" systems on Internet

,{:H'"ilJ"#ffi:}j;i:',"i;:¡'J|ll'liili;!:,ild:[:'#,l"iüü;, R;;;; McKeown and Hovv

De Cunha Fanego' zuuc

An¡tysl3

113

20021 Hahn and Mani ' 2000

4. Evaluation of automatic extract ion systems

Next, we wil l analyze and evaluate a group of automatic extract ion systems which can be found on Internet forfree. Most of these systems merely extract a text, but we have to bear in mind that there are some commerctalsystems'that sum up real abstracts - not extracts - such as Summarist (Hovy and Lin 1999), DlmSum (Aoneet al . , 1999) and SCISOR (Rau, Jacobs and Zernik, 19Bg)Categ ori zation of i n p u t

The input of this analysis wil l be the terms and condit ions of package tours -specif ical ly cruises- taken out frommult i l lngual rnacrocorpus Turicor (Ref. no. BBF2003-04616 (2003-2006, Spanish Ministry of Science andTechnology). The mult id iscipl inary, appl ied R&D TURICoR project belongs to the f ie ld of Informat ion andComrnunicat ion(s) TechnoloSy ( lCT) appl ied to tour ist and legal t ranslat ion.Given the input, we wil l work with legal texts. However, i t should be added that legal discourse is a sublanguagethat, in order to extremely precise, gives some problems to NLP. According to Somers (2003: 286), "an exlremeexample of a dist inct ive sublanguage which has numerous features which are qui te MT- unfr iendly is ' legalese," .Therefore, i t is pert inent to wonder i f these automatic systems wil l be more eff icient when working with control ledlanguages (Rico and Torrejón, 2002; Mitamura, 1999). However, th is subjectwi l l not be studied in thrs paper.Automattc extract¡on sysle,?s ava¡lable an lnternetIn order to analyze how automatic extract ion systems work, we needed to f ind free on-l ine systems that yieldsummaries in Engl ish and Spanish (and, i f possible, in other languages). Most of the systems we found werebased upon stat ist jcal methods, such as sentence locat ion (Edmundson, 1969), word añd sentence frequency(Luhn, 1958) or key words or sentences (e.g, , " i t is important to consider") (Edmundson, '1 969), Furthermore,these systems frequently use the tf . idf weighting ("Term-frequency t imes inverse document frequencf ') (Saltonand McGil l , '1983; Hovy and Lin, 1999; Mani and Bloedorn, '1 999), which is usual ly used in informai ion retr ieval .once presented the common features of these systems, we must specify their names:

1 Copernic Summarizer

2. Extractor

3. GistSumm

4. Resumldor de Microsoft Word

5. SweSum

6. Websumm

Analysis and evaluation of outcomesFor the analysis and assessment of the outcomes, we focused on the demo version of an automatic extract ionsystem: Copernic Summarizer. Next, we show a part of the extract in English:Concepts:reserve, members, travel, booking, contract, charge, request, suppliers, payment, departure, amounts, hol iday, costs,arrangements, agents.Summary:

- B "You, yoüi 'and related expressions refer to the lead member and al l other members of the party named in the booking andthose on whose behalf the booking is made.- A Where we are supplying you with lwo or more signif icant services e.g. f l ights and accommodation our contract incorporatesal l of the information on Our Site relevant to the services you wish to purchase, any writ ten materials we or any of our suppliersprovjde to you, our confirmation ¡nvoice and the copy ofthese booking condit¡ons which we send to you fol lowing your telephonereservation or which you accept onl ine.

Then, the extract of terms and condit ions in Spanish is the fol lowing:Concepts:v¡aJe, contrato, consumidor, organizador, viaje combinado, precio, detal l ista, oferta, condiciones, documenlación del viaie,fecha, transporte, Condiciones Generales, fo¡leto, Agencia.Summary:

- A los efectos de las presentes Condjcíones Generales, el programa / fol leto es el documento informativo al que estas seincorpo¡a n.- El programa / oferta es ia descripción del combinado contenida en el programa / fol leto que consti tuye el objeto del contrato deviaje combinado.- La ínformación sobre el programa / oferta contenida en el programa / fol leto es vinculante para el organizador o detal l¡sta,salvo que concurra alguna de las siguientes circunstancias: Que los cambios en dicha información se hayan comunicadoclaramente por escri to al consumidor antes de la celebración del contrato y tal posibi l idad haya s¡do objeto de expresa menciónen el programa oferta,.

'On lnternet there are many comercial automatic summarization systems, such as: Brevity Document Summar¡zer(http://www lextek.comibrevity4, lntelligent Miner for Iext (http://www-3O6,ibm.com/software/data/iminerl), lnxight Summarizer(http://www.inx¡ght.com), Sinope Summarizer (http://www.sinope. info/en/Download) and TextAnalyst(http:/ /www. meg a puter. com/prod ucts/tal index. ph p3).

444

Reoarding the extracts yierded by copernie summarizerboth in spanish and Engrish, it is worth mentioning that

weiind tTJfl:".-".',liy:::1s;:, which reaily herps to know in advance the content of the document'

- Sentences are complete' coherent and consistent'

- Most of the basic elements in a package tour are.on the summary, such as "booking"' "reservation'' '

"comPensation"' etc'

Nevertheless, thefol lowingshortcomingsmustbeadded:- Some words are often rePeated

- There is some trrelevant information that should not be on a summary

í'tl';jff"Jiff."J':11';;J"i"ffifrif;':ffiii:oncernins the utlitv or automatic extraction in transration process'

lpJjtüiiit'r;; iiiñtiut¡ng i"*s anJ conditions of packase tours:

- lt gives a list of kev concepts ii,!Y^",1:1s^'"9es and' ":5":i::tlt

one can know beforehand the content

ofadocument,Then, i f thetrans|atorneed-stoknowthemostusedterminologyinEng| ishcontractsaSwell as ¡n terms and conditions, he/sñe wil l f ind "via1e' conüáto' contuñiiOor organizador' viaje

combinado, precio, detall ista, . oferta,, "onái"ion"",

dócumentación del viaje' fecha' transporte'

Condiciones Generales, folleto, Agenciu". in gngli ' i ' , ihe most remarkable terms are "reserve' members'

travel, booking, contract, charge, ,"qr".i lr"ppli"tt payment' Jeparture' amounts' holiday' costs'

i."i: ,t"ffli;ilili;,"T,",ff,T:l'"i. topic or the text In addition the most important thing is that

abstracts often help translators oec¡oe wrrJtle|. io ""

the text in their documentation process'

- |n|ega|trans|at ion, lheextractgivesindicat ionsofthemostusedstructure, for instance,theEng| ishstructure may + ¡nfinit ive without t" te"#'Áúi' zOoo' 15)

'( 'our Site and to create binding legal

obl igat ions for any l iabi l i ty you may incur as a resul t of such use") '

-Humanpostedi t ionmayberequiredin 'ordertoimprovetheoutcomesbecausetheyieldingabstractsarenot very

"on"'"n-t unO''consequently' their effective ness decreases

In short, in spite of some shortcomings, automatic extraction systems contribute to the documentation process.

Thus, using automarrc áxtraction in the documentation proceós may, be really interesting since translators

familiarize themselves *i in ir]e terminology "td

t;;;;; i ;; i no*-t patarról texts and other information sources'

Proposal for improving automatic abstraction: MUSI prolect t sponsored by

Weproposeamult i l ingua|andautomat icsummarizat ionsystemthatwou|dbebaseduponaproJecrhe EU: MLtS-MUS' pái""i: inlrl i , l¡nguat sum,iariÁl¡oi 'ro' the tnternet) (Lenci et at ' 2002) Althoush a smarl

scate researcn p,oi""t, ü-uti'Árr'iti.iJ to tacr<reinei*náireng., 'et uv túrt]lnsyal 3umm15"1¡:i:v

adopting an

originat approach our"J"oñr,"?;;it i"; ot "

,nur-uioniároiy tno répresenta-tion language' and on the reuse oT

existing l inguistrc rurorr""r. MUSI combin", uiiuii. i i"-UáiáO n'toOüfe for relevant lentence extraction and a

concepfbased "orpon"ii io

éñr"tu rurt,r,nw"i ,u*taries Thus' our system will be based on this structure

and it wil l be useo ror professional and educat,'onut-prrpor"s both in proteislonat translation and in Translation

Studies.5. Conclusions

Murtiringual automailc summarization highly faciritates the use of Iniernet as an information resource, especialty

for translators, slnce rt makes possibte to .ompire more information i¡ less time and' consequently' to improve

transratof s efficiency. Thus, the implementatioñ "r it

" future multi 'nguar auiámatic summarization program will

constitute an interesting tooi for both translatro" i"r"f- ' i .Sl.Jprofes'iónal t iánslation Therefore' it tends to be an

important and new "p;Yi;i;;; iÁG

and NLGio the Translation studies'

HH:iT:.oJoR M DIAZ, A cóMEZ .lJ^,1^y1Ñl,Yoj,,'oiJ ,T:"'::;:;il'"#"i"i de resúmenes

personarizado s,, xvil congreso de Ia .sic¡ádad Españoru d" lár"támiento de Lenouaie Natural

'(sEpLN'2001),12-l septiembre ?091, l3liiÉróu¡ul iublicado.en doiásam¡ento det Lensiaie Naturat' no

27, septiembre 2001, 281-287 ..nttp://www sü']!tói"ulttusEPLNlt"u'ttal27 ,t>tjt.: ': j:l^"

AONE,c.M' 'OKUROWSKI,E.,GORLINSKY,JyLARSEN,B:1.9?9: ' ,ATrainableSummari .zerwithKnow|edgeAcquired r|."nl nJnlri 'ÑLF iá.r,n,qu".' t; ffiÑl,llv MÁYeunv H¡

-regg Advances in Automatic rext

Summarization' i"OtiOé"' Massachusetts Institute of Technotogy'

coRpAS pASToR, G (ed ) 2003. Recurso s documentales y tecnológicos para la traducción del drscurso iurídico

(e sp a ñ ot' u, ",

u'l ¡l¡ nv tz !'1zi ¡á n o á r ab e) G ra na da : Ed itori a I co m a res'

DAcUNHA'1.2005, ' .Haciaunmode|ol ingÜÍst icoderesumenautomátrcodear l ículosmédicosenespañol . 'universidad pompeu Fabra, Inst i tuto, ,uniu"Li tu ' io de Lingüíst ica"Ápr i t t¿t Doctorado'-en ciencias del

Lensuaje v r-ingui5;; ;;;";¡" 'rrttp:l¡wwwlupi "JwÑi'r '/ i ia dacunha/#0202> [07-04-06]

Dpto. de uenguajes y ,- ,?:t:TÍ:,

Informáticos' UNED Proyecto Hermes Disponible en

<htto:/inlp uned es/hermesi> L'l 5-u4-uol

!1,r*,"

Yr,í

& 115

EDMUNDSON, H.P. 1969. "New Methods in Automat ic Abstract ing". En MANI, L y MAYBURY, M. 1999.Advances in Automatic Text Summarizafion. Cambridge: Massachusetts Institute of Technology

ENDRES-NIGGEMEYER, B. 1998. Summarizing lnformation. Hannover: Springer- Verlag Berlin Heidelberg.

HAHN, U. y MANI l. 2000. lThe Challenges of Automatic Summarization". \EEE Computer, ¡¡ ZOOO (11), 29-36.

HOVY, E. y LlN, C. 1999. "Automated Text Summarization in SUMMARIST". En MANI, l. y MAYBURY M. 1999.Advances in Automatic Text Summarizaflon. Cambridge: Massachusetts Institute of Technology.

LAVID, J. 2005. Lenguajes y nuevas tecnologías. Nuevas perspectivas, métodos y herramientas para el lingüistade! siglo XXl. Madrid: Cátedra.

LENCI, A. , BARTOLINI, R., CALZOLARI, N., AGUA, A., BUSEMANN, S., CARTIER E , CHEVREAU, K, YCOCH J. 2002. "Multi l ingual summar¡zation by integrating l inguistic resources in the MLIS-MUSI project".Proceedings of the 3rd international conference on language resources and evaluation (LREC'12), LasPalmas, España. <http://www.coli.uni-saarland.de/publikationen/softcopies/Lenci:2002:MSl.pdf> [03-03-06]

LUHN, H.P. 1969. "The Automatic Creation of l i teratura Abstracts'r. En MANI, l. y MAYBURY, M. 1999. Advancesin Automatic Text Summarization. Cambridge: Massachusetts Institute of Technology.

MANI, l. y MAYBURY, M. 1999. Advances in Automatic Text Summarizalion. Cambridge: Massachusetts Instituteof Technology.

MANN, WC., y THOMPSON, S.A. 1988. "Rhetorical Structure Theory: Toward a functional theory of textorganization". Iexf, 8 (3), 243-281 .

MARCU, D. 2000. The theory and practice of discourse parsing and summarization. Cambridge: MassachusettsInstitute of Technology,

MITAMURA, T. 1999. "Controlled Language for Multi l ingual Machine Translation". Machine Translation SummitV//, Singapur. <http://www.lti.cs.cmu.edu/Research/Kant/PDF/MTSummit99.pdf> [10-02-06]

PINTO MOLINA, M. 2000. "Documentación para la Traducción en la sociedad de la información". XV ColoquioAssociation lnternationale de Bibliologie. Salamanca, AlB, 2000.<http://www.mariapinto.es/web/mainframe htm> [05-04-06]

RADEV, D. HOVY, E. y MCKEOWN, K. 2002. "lntroduction to the Special lssue on Summarization". EnComputational Linguistics, 28 (4), 339-408.

RAU, P., JACOBS, S. y ZERNIK, U. 1989. "lnformation Extraction and Text Summarization using LinguisticKnowledge Acquisition". lnformation Processing and Management, vol.25, no. 4, pp.419-428.

fiElTER, E. y DALE, R. 1997. "Building'applied natural-language generation systems". Journal of Natural-Lang uage Engineering, 3:57 - 87 .

RICO, C. y Enrique TORREJON. 2002. "Controlled Translation: A New Teaching Scenario Tailor-made for theTranslation Industry". Proceedings of the 6'n European Association for Machine Translation Workshop. UMIST,Manchester.

SALTON, G. y MCGILL, M. 1983. lntroduction to Modern lnformation retrieval. New York: McGraw Hill.

SOMERS, H. (ed.) 2003. Computers and Translation. A translator's guide. Philadelphia: John BenjaminsPubl ishing Company.

116