Submitted by Ariel Marcelo Fernández Quiroz (rel.fernandezq@gmail.com) on 2018-07-11T15:45:48Z No. of bitstreams: 1 Dissertação Ariel Fernández.pdf: 5074427 bytes, checksum: 67e509c12de12d41d7b1c49b9590090d (MD5) / Approved for entry into archive by Paula Torres Monteiro da Torres (paulatms@sjrp.unesp.br) on 2018-07-12T12:19:59Z (GMT) No. of bitstreams: 1 quiroz_amf_me_sjrp.pdf: 5074427 bytes, checksum: 67e509c12de12d41d7b1c49b9590090d (MD5) / Made available in DSpace on 2018-07-12T12:19:59Z (GMT). No. of bitstreams: 1 quiroz_amf_me_sjrp.pdf: 5074427 bytes, checksum: 67e509c12de12d41d7b1c49b9590090d (MD5) Previous issue date: 2018-04-20 / Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) / O principal problema da maioria das dublagens de produtos audiovisuais humorísticos são os laugh tracks (sons artificiais de um público rindo), já que toda vez que há trilha sonora de risadas, estas devem coincidir com uma piada para não causar estranheza no público-alvo. Neste trabalho analisaremos, por meio de um corpus paralelo, os problemas de tradução presentes na dublagem de um desses produtos: o seriado “El Chavo del 8” (“Chaves” no Brasil) do espanhol do México para o português do Brasil, com base nas teorias de dublagem fundamentadas por Hurtado Albir (1996); de humor, fundamentadas por Raskin (1987), Bergson (1983), Posada (1995) entre outros; e de técnicas de tradução propostas por Hurtado Albir (2001). Apresentamos uma análise realizada em três etapas: na primeira, criamos um quadro com as minutagens das piadas para cada um dos 18 episódios analisados e uma seção “houve/não houve piada”; na segunda, 12 participantes responderam se houve piada ou não em cada trecho selecionado; finalmente, na terceira etapa, criamos quadros para cada piada nas quais os participantes determinaram que não houve piada e explicamos o motivo dessa perda. Com base na definição dos problemas e nas técnicas de tradução, pretende-se apresentar as possíveis soluções que os tradutores audiovisuais teriam para traduzir as piadas em caso de perda de comicidade. / The main problem with dubbing translation in most humorous audiovisual products is the laugh track, since every time there is a laugh track it must match a joke not to cause any strangeness in the target audience. In this research, we will analyze, through a parallel corpus, the translation problems in the dubbing of the series "El Chavo del 8" ("Chaves" in Brazil) from Mexican Spanish to Brazilian Portuguese, based on theories of audiovisual translation by Hurtado Albir (1996), Humor by Raskin (1987), Bergson (1983) and Posada (1995), and translation strategies by Hurtado Albir (2001).We show an analysis performed in three stages: in the first one, we created tables with the minutes of the jokes in 21 episodes and a “yes / no” joke section; in the second one, 14 participants answered whether or not there was a joke in each selected section; finally, in the third one, we created tables for each joke in which participants determined if there was no joke. Based on the definition of the problems and translation strategies, we intended to offer possible solutions for the audiovisual translators when dealing with jokes. / CNPq:190394/2015-3

Les équivalences traductionnelles extraites à partir d’un corpus parallèle deviendraient une ressource précieuse permettant d’étudier différents contextes traductionnels envisagés entre les deux langues distinctes. L’utilisation des textes traductionnels constitue aujourd’hui un thème essentiel en traductologie et en études contrastives des langues. Les méthodes textométriques opèrent une série de calculs statistiques portant sur les unités textuelles dans un corpus parallèle segmenté en occurrences. Elles fournissent les indices quantitatifs permettant de mettre en évidence le lien traductionnel de ces unités. En examinant des formes bilingues issues des corpus parallèles français-coréens, nous avons vérifié l’utilité de cette méthodologie appliquée aux textes traductionnels en français-coréen. Elles ont effectivement donné un résultat positif, d’une part, et un résultat négatif, d’autre part, tout au long de nos travaux. Pourtant, grâce à ces méthodes, nous avons pu étudier divers liens traductionnels entre unités textuelles du français et du coréen. La plupart de méthodes automatisées consacrées au corpus parallèle en langues hétérogènes n’ont pas produit de résultat acceptable. À ce titre, la textométrie, qui vise à l’observation quantitative des éléments lexicaux d’un corpus, serait très intéressante lorsqu’il s’agit notamment d’un corpus parallèle en langues sans parenté. / The translational equivalences extracted from a parallel corpus become a valuable resource enable to study the various translational contexts between the two distinct languages. The use of translational texts is now a principal subject in the translation studies and the contrastive studies of languages. The textometry operate a set of statistical calculations on the textual units in a parallel corpus divided into the tokens. They provide the quantitative evidence that verify the translational relation of the linguistic units. In exploring bilingual words in the French-Korean parallel corpora, we verified the usefulness of this methodology applied to the French-Korean translational texts. They produced actually a positive result, on the one hand, and a negative result, on the other hand, throughout our work. Yet, these methods made also observe the various translational relations of textual units between French and Korean. The most automated methods devoted to the parallel corpora of heterogeneous language pairs have not produced the approvable result. For the reason, the textometry, which aims to observe the lexical elements of a corpus from a statistical point of view, would be very practical method when we deal with a parallel corpus that consists of different language pairs.

This Master thesis deals with machine translation from Czech to Slovak. The first chapter motivates the work, the second discusses various approaches to machine translation and the third details evaluation of the methods. Chapter 4 introduces the design and implementation of my system, paying a special attention to a new parallel corpus that has been created. Chapter 5 summarizes testing and evaluation of the developed system.

Šio darbo tyrimo objektas yra lietuvių ir latvių parodomieji įvardžiai, o tikslas – nustatyti jų santykius lietuvių – latvių kalbų vertimuose ir jų keliamas problemas vertėjams. Iškeliama tokia hipotezė: parodomieji įvardžiai vertimuose iš lietuvių kalbos į latvių kalbą dažniausiai, bet ne visada, yra keičiami latvių kalbos parodomaisiais įvardžiais. Darbo medžiaga yra dabar (nuo 2011 metų sausio iki 2012 metų gruodžio) Vytauto Didžiojo universiteto ir Latvijos universiteto bendrai rengiamas lietuvių – latvių lygiagretusis tekstynas. Į šį tekstyną įtraukiami lietuviškų tekstų vertimai į latvių kalbą. Iš šio tekstyno lygiagrečiųjų tekstynų konkordavimo programa ParaConc buvo sudaryti konkordansai pagal lietuvių parodomuosius įvardžius ir tirta, kaip jie verčiami į latvių kalba. Darbas susideda iš penkių dalių. Pirmoji dalis – įvadas, kuriame trumpai aprašytas tyrimo tikslas ir uždaviniai. Antrajame skyriuje nustatoma, kas yra parodomieji įvardžiai ir tarpusavyje lyginami šios grupės lietuviški ir latviški žodžiai. Trečias skyrius skirtas tyrimo metodikai ir panaudojamam tekstynui aprašyti. Ketvirtame skyriuje atliekama iš tekstyno išgautų duomenų analizė. Atskirai apžvelgtas lietuviškų parodomųjų įvardžių vertimas latviškais parodomaisiais įvardžiais; lietuviškų parodomųjų įvardžių vertimas kitais įvardžiais ir kitoms kalbos dalims priskiriamais žodžiais; lietuviškų parodomųjų įvardžių nevertimas. Penktoje dalyje pateikiami darbo rezultatai ir kelios rekomendacijos... [toliau žr. visą tekstą] / Lithuanian and Latvian demonstrative pronouns are the object of this paper. The aim is to identify their relations in Lithuanian – Latvian translations and the problems they may cause to the translators. The hypothesis is as follows: Lithuanian demonstrative pronouns in Lithuanian – Latvian translations are mostly but not always replaced by Latvian demonstrative pronouns. This research is based on the Lithuanian – Latvian parallel corpus which is now being prepared in two partner universities: Vytautas Magnus University and University of Latvia. In this corpus translations of Lithuanian texts into Latvian are being collected. Concordances of Lithuanian demonstrative pronouns are extracted from this corpus using concordancer ParaConc. It is studied how these pronouns are translated into Latvian. The paper consists of five chapters. The first one is introduction where the aim and tasks are shortly described. The second chapter presents demonstrative pronouns, the definitions of this group and the words that belong to it in Lithuanian and Latvian languages. The third chapter describes methodology and the corpus used in this research. In the fourth chapter analysis of the corpus data is performed. Translation of Lithuanian demonstrative pronouns as Latvian demonstrative pronouns; translation of Lithuanian demonstrative pronouns as other Latvian pronouns or other word classes; and discarding of Lithuanian demonstrative pronouns are discussed. Conclusions and some recommendations... [to full text]

This thesis deals with the French Supplementations (étoffement), which when converted to Czech are usually resolved with a simple prepositions because word by word translation in this case was flawed. Its objective is based on the phenomena to be investigated to establish similarities and deduce a general theory of (general model) that would characterize the formation and use of French Supplementations and also ascertain to what extent is Supplementations translation and linguistic problem. For the purposes of the objectives was used parallel corpus Intercorp. The work is divided into five chapters, during which it is committed to the most comprehensive source of description and characterization of the French Supplementations and not least also to reveal the reasons why these constructions in French translation into Czech discharged.

This study investigates how the semi-modal need to is translated into German/Swedish and which German/Swedish correspondences are translated into need to. To this end, the Linnaeus University English–German–Swedish Corpus (LEGS) is used. Nida’s (1964: 159-162) concept of formal and dynamic equivalence is used to perform the qualitative analysis and to discuss the results from the quantitative part of the study. The use of semi-modals such as be going to, have to and want to have increased during the second half of the 20th century (Leech et al.: 2009: 99). need to represents the obligation as being in the best interest of the subject and is associated with objectivity (Kastrone 2008: 829; Aijmer 2017: 28) Thus, need to is used to distance the speaker to avoid an authoritarian stance. This trend is a sign of an ongoing democratization (Leech et al. 2009: 270). The results showed that the preferred German translation is müssen (‘must’) (55%) and  the preferred Swedish translation is behöva (‘need’) (47%). ‘Other’ is the second preferred German translation and the third preferred Swedish translation. These results are reflected in the structures translated from German and Swedish. The results indicate that the semantic category of the co-occurring main verb and the co-occurring subject affect translation. Based on these results, it could be said that English, followed by Swedish, is leading the process of democratization. However, further studies are needed to confirm this hypothesis.

The goal of diploma project was to implement parallel corpus manager, which can align parallel texts in different languages and insert them into corpus, where several more processing functions are provided. Program provides possibilities of automatic text alignment and its interactive editing. These aligned texts are then inserted into corpus. Program can work with multiple corpora, parallel corpus is allways identified by a couple of languages. In corpus, there are possibilities to search by many categories, view and edit particular selections, lemmatize and morphologically tag given texts, sort selections, import and export data, in many ways edit corpus for further easy navigation and add new expressions to managed dictionaries. Particular chapters describe introduction to corpus problematics, theory of aligning parallel texts, morphological text tagging and lemmatization, external tools used in program, most common subtitle formats and implementation solution of particular problems.

L'objectif de cette thèse est de lever les verrous que constituent le manque de disponibilité de ressources ou d'outils TAL pour la langue arabe dans les domaines de spécialité en proposant des méthodes permettant l'extraction de termes à partir de textes en arabe standard moderne. Dans ce contexte, nous avons d'abord construit un corpus parallèle anglais-arabe dans un domaine de spécialité. Il s'agit d'un ensemble de textes médicaux produits par la bibliothèque nationale de médecine américaine (NLM). Par la suite, nous avons proposé des méthodes d'acquisition terminologique, permettant d'extraire des termes ou d'acquérir des relations entre ces termes, pour la langue arabe en se basant sur: i)adaptation d'un extracteur terminologique existant pour la languefrançaise ou anglaise, ii) l'exploitation de la translittération des termes anglais en caractères arabes et iii) l'application de la la notion de transfert translingue. Appliqué au niveau terminologique, le transfert consiste à mettre en œuvre un processus d'extraction de termes ou d'acquisition de relations entre termes sur des textes d'une langue source (ici, le français ou l'anglais) puis à transférer les informations extraites sur des textes d'une langue cible (ici, l’arabe standard moderne) pour ainsi identifier le même type d'informations terminologiques. Nous avons évalué les listes de termes monolingues et bilingues obtenues lors des différentes expériences que nous avons réalisées, suivant une méthode transparente, directe et semi-automatique: les termes candidats extraits sont confrontés à une terminologie de référence avant d'être vérifiés manuellement. Cette évaluation suit un protocole que nous avons proposé. / The goal of this thesis is to reduce the lack of available resources and NLP tools for Arabic language in specialised domains by proposing methods allowing the extraction of terms from texts in Modern Standard Arabic. In this context, we first constructed an English-Arabic parallel corous in a specific domain.It is a set of medical texts produced by the US National Library of Medicine (NLM). Thereafter, we have proposed terminological acquisition methods, toextract terms or acquire relations between these terms, for Arabic based on: i) the adaptation of an existing terminology extractor for French or English, ii) the transliteration of English terms in Arabic characters and iii) cross-lingual transfer. Applied at the terminological level, transfer aims to implement a process of term extraction or relationship acquisition between terms in the texts of a source language (here, French or English) and then to transfer the extracted information to target language texts (in this case, Modern Standard Arabic), thereby identifying the same type of terminologicalinformation. We have evaluated the monolingual and bilingual term lists that we have obtained by the experiments we carried out, according to a transparent, direct and semi-automatic method: the extracted term candidates are confronted with a reference terminology before being validated manually. This evaluation follows a protocol that we proposed.

This paper investigates variation of lexical and analytic causatives in 15 European languages from the Germanic, Romance, and Slavic genera based on a multilingual parallel corpus of film subtitles. Using typological parameters of variation of causatives from the literature, this study tests which parameters are relevant for the choice between analytic and lexical causatives in the sample of languages. The main research question is whether the variation is constrained by one semantic dimension, namely, the conceptual integration of the causing and caused events, as suggested by previous research on iconicity in language, or whether several different semantic and syntactic factors are at play. To answer this question, I use an exploratory multivariate technique for categorical data (Multiple Correspondence Analysis with supplementary points) and conditional random forests, a nonparametric regression and classification method. The study demonstrates the importance of corpus data in testing typological hypotheses.

Since the publication of the seminal work by Lakoff and Johnson (1980a), Metaphors we live by, countless research has been done on metaphor. This research was conducted because, in the past, metaphor was considered a deviant and poetic device that could be used only by those who were skilful. These scholars offered another view: metaphor is a matter of thought. They showed that linguistic metaphor is the manifestation of conceptual metaphors that are in our subconscious mind and are found in every day language. In other words, metaphors are a revelation of how we think. Linguists and translation scholars claim that it is a challenge to translate metaphor. The main objective of this study is to determine how conceptual metaphor theory can contribute towards the development of translation in African languages. The study seeks to identify conceptual metaphors in Nelson Mandela’s autobiography, Long Walk to Freedom, which was written in English, and then analyse how these were translated into isiXhosa and isiZulu. This implies that this study involves a corpus. In identifying metaphors from the source text a Metaphor Identification Procedure Vrije Universiteit (MIPVU), which was conceived by Steen and his colleagues at the Vrije Universiteit Amsterdam, was used (Steen et al 2010). ParaConc concordancer was used to investigate and compare how the metaphors were translated. The results showed that most metaphors were translated the same way in isiXhosa and isiZulu, which implies that the translators conceptualised the metaphors in the same way. These results revealed that the translators’ styles were similar. This confirms the fact that the two languages are related. As scholars in earlier research indicated, metaphors in translation pose a problem, yet the translators of Mandela’s book successfully met this challenge. They were able to render the metaphors in their respective languages in an acceptable manner. They tried to adhere to the style of the source text writer, but traces of their own style are evident in the texts. / Linguistics and Modern Languages