Academic literature on the topic 'Translation model'

This thesis studies the effectiveness of existing translation models in the context of everyday translation and proposes a new translation model. The thesis reviews a number of approaches to the process of translation from the Roman times to the present before focusing on contemporary translation theory and the representation of the translation process by means o f translation models. The thesis introduces - and comments on - a number of existing translation models and then proceeds to develop a new model of the process, which aims to present a more holistic view of the process than the models discussed. The second part of the thesis concentrates on the testing of the model. Two very practical tests are applied to the model in order to assess the accuracy of the representation and the usability of the model in the context o f everyday translation. The first test applied to the model has, however, another function. It aims to provide a contemporary readership with a readable English translation of a Renaissance Latin text, the first book of the Instructiones historico-theologicae de doctrina Christiana et vario rerum statua temporibus Apostolici, ad tempora usque seculi decimi septime prior a (1645) by John Forbes o f Corse. This enables a wide audience with very little or no knowledge o f Latin to gain access to the complex theological argument contained in the specimen text. The commentary on the English translation, and on extracts of the German and French translations of this work serves to test the applicability of the model in the context of translation into more than one language. The second test concentrates on the translation from English into German and German into English. For this test, two groups of students from the Universities of Trier and Rostock in Germany were asked to carry out the same translation exercise. The study o f the work received from these students allows me to assess the usability o f the model as a guideline for translators. The thesis concludes by saying that the model has proved successful on both occasions, and by offering suggestions for further study.

Thesis (Ph. D.)--Ohio State University, 2002.<br>Title from first page of PDF file. Document formatted into pages; contains xvi, 306 p.; includes graphics. Includes abstract and vita. Advisor: Chris Brew, Dept. of Linguistics. Includes indexes. Includes bibliographical references (p. 284-293).

The term equivalence describes the relationship between a translation and the text from which it is translated. Translation is generally viewed as indeterminate insofar as there is no single acceptable translation - but many. Despite this, the rationalist metaphor of translation equivalence prevails. Rationalist approaches view translation as a process in which an original text is analysed to a level of abstraction, then transferred into a second representation from which a translation is generated. At the deepest level of abstraction, representations for analysis and generation are identical and transfer becomes redundant, while at the surface level it is said that surface textual features are transferred directly. Such approaches do not provide a principled explanation of how or why abstraction takes place in translation. They also fail to resolve the dilemma of specifying the depth of transfer appropriate for a given translation task. By focusing on the translator�s role as mediator of communication, equivalence can be understood as the coordination of information about situations and states of mind. A fundamental opposition is posited between the transfer of rule-like or codifiable aspects of equivalence and those non-codifiable aspects in which salient information is coordinated. The Translation Salience model proposes that Transfer and Salience constitute bipolar extremes of a continuum. The model offers a principled account of the translator�s interlingual attunement to multi-placed coordination, proposing that salient information can be accounted for with three primary notions: markedness, implicitness and localness. Chapter Two develops the Translation Salience model. The model is supported with empirical evidence from published translations of Arabic and English texts. Salience is illustrated in Chapter Three through contextualized interpretations associated with various Arabic communication resources (repetition, code switching, agreement, address in relative clauses, and the disambiguation of presentative structures). Measurability of the model is addressed in Chapter Four with reference to emerging computational techniques. Further research is suggested in connection with theme and focus, text type, cohesion and collocation relations.

Nowadays several indicators suggest that the statistical approach to machinetranslation is the most promising. It allows fast development of systems for anylanguage pair provided that sufficient training data is available.Statistical Machine Translation (SMT) systems use parallel texts ‐ also called bitexts ‐ astraining material for creation of the translation model and monolingual corpora fortarget language modeling.The performance of an SMT system heavily depends upon the quality and quantity ofavailable data. In order to train the translation model, the parallel texts is collected fromvarious sources and domains. These corpora are usually concatenated, word alignmentsare calculated and phrases are extracted.However, parallel data is quite inhomogeneous in many practical applications withrespect to several factors like data source, alignment quality, appropriateness to thetask, etc. This means that the corpora are not weighted according to their importance tothe domain of the translation task. Therefore, it is the domain of the training resourcesthat influences the translations that are selected among several choices. This is incontrast to the training of the language model for which well‐known techniques areused to weight the various sources of texts.We have proposed novel methods to automatically weight the heterogeneous data toadapt the translation model.In a first approach, this is achieved with a resampling technique. A weight to eachbitexts is assigned to select the proportion of data from that corpus. The alignmentscoming from each bitexts are resampled based on these weights. The weights of thecorpora are directly optimized on the development data using a numerical method.Moreover, an alignment score of each aligned sentence pair is used as confidencemeasurement.In an extended work, we obtain such a weighting by resampling alignments usingweights that decrease with the temporal distance of bitexts to the test set. By thesemeans, we can use all the available bitexts and still put an emphasis on the most recentone. The main idea of our approach is to use a parametric form or meta‐weights for theweighting of the different parts of the bitexts. This ensures that our approach has onlyfew parameters to optimize.In another work, we have proposed a generic framework which takes into account thecorpus and sentence level "goodness scores" during the calculation of the phrase‐tablewhich results into better distribution of probability mass of the individual phrase pairs.

This thesis tests whether a translation training programme, combining the theoretical principles adopted in Hervey and Higgins' Thinking Translation series and in Bolanos' dynamic translation model, helps to improve Arab MA translation students' performance in English-to-Arabic translation. The course materials used were taken from Thinking Arabic Translation and Supplement to Thinking Arabic Translation. The experimental group were MA translation students at the Higher Institute of Translation and Interpretation, University of Damascus, while MA translation students at the University of Petra (Amman) were used as a control group. Both groups of students were given an initial translation exam. The University of Damascus students (but not the University of Petra) students were then given a 20-hour one-month translation training programme. At the end of this period, both the Damascus students and the University of Petra students were given a final translation exam. All errors made by both groups of students in the initial exam were precisely analysed, using Hervey and Higgins' matrices-level-rank model, as were all errors made by both groups of students in the final exam. The results showed that while University of Petra students had not improved between the initial exam and the final exam, University of Damascus students had made a significant improvement. These results corroborate the view that the theoretical principles developed by Hervey and Higgins, as operationalised for Arabic/English translation in Thinking Arabic Translation and Supplement to Thinking Arabic Translation, help to improve students' English/Arabic translation performance.

Translation evaluation is one of the main concerns of translation theorists, members of translation revision boards, and most importantly it is the concern of translator trainers. Translation quality has often been associated with the correctness of the grammatical structure and the appropriateness of the lexical item. Little concern has empirically been given to units larger than the sentence, i.e. text. This seems to be the result of the prevailing linguistic trend that has put more emphasis on a -context-free' sentence, rather than on text in context. This study proposes to investigate, discuss and develop a translation quality assessment model that takes text, not a sentence as the ultimate aim of analysis. The study will also attempt to explore the theoretical and practical implications of the model to be developed for the training of translators in the Arab world. The model to be developed should be based on the definition that translation is the replacement of a text in the source language by a semantically, pragmatically and textually equivalent text in the target language. Text, then, is the focus of interest in this study. Therefore, the model will be developed within the framework of text lingui4Vics for which text is regarded as a communicative occurrence. The developed model will serve as a means to evaluating the quality of Arabic-English translations of a particular type of texts, argumentative text type. Therefore, two argumentative texts in the form of newspaper editorials, selected from two Moroccan quality newspapers will be analyzed along the dimensions of what will be known in this study as a Situational/Textual model. The resultant "textual profile" will, then, be taken as a "yardstick" against which will be measured 81 translations collected from Fand School of Advanced Translation (FST) and 5 from the department of modern languages, Salford University (SU). The first introductory chapter lays out the main arguments of the thesis. Chapters two and three present and discuss sentence-oriented translation models, and text-oriented translation models respectively. Chapter four presents and discusses the following: a) the three aspects of meaning: semantic, pragmatic, and textual, b) language function vs. text function, and finally C) House's (1981) model of translation quality assessment. Chapter five presents the method of operation, discusses the decision criteria needed to deal with the dimensions linguistic correlates, and finally illustrates the extended situational/textual model for translation quality assessment. Chapter six is the application of the model on the two Arabic argumentative texts. In addition, argumentative text structure will be discussed and the difference between Arabic and English argumentative texts will be explained. Finally, chapter seven includes the source language text (SLT), and the target language text (TLT) statement of comparison and statement of quality, and a discussion of the theoretical implication of the model for the training of translators in the Arab world.

When J. R. R. Tolkien’s trilogy The Lord of the Rings was published in Swedish 1959-1961, the translation by Åke Ohlmarks was considered by most critics to be excellent. According to Ohlmarks, even J. R. R. Tolkien himself and his son Christopher were very pleased with it, which Ohlmarks was told by Christopher when he met him in 1975. This is, however, contradicted in the authorised biography of Tolkien by Carpenter (1978), in which Tolkien is said to have been most negative towards the way Ohlmarks handled the text. Before the biography was published, Christopher Tolkien and Ohlmarks had become bitter enemies, which might explain the re-evaluation. The schism has been described by Ohlmarks in his book Tolkiens arv (1978). But ever since The Lord of the Rings came out in paperback in 1971 there has been a discussion about the translation quality also in Sweden. When I first read the books in English I had the Swedish translation beside me. I soon discovered that Ohlmarks had taken great liberties with the text. I noticed that the descriptions were often more detailed in the Swedish translation than in the original and it was this fact that first roused my interest. Therefore, I decided to try to make a translation quality assessment of a part of the text, using a model by Juliane House.

We investigate a new approach for SMT system training within the streaming model of computation. We develop and test incrementally retrainable models which, given an incoming stream of new data, can efficiently incorporate the stream data online. A naive approach using a stream would use an unbounded amount of space. Instead, our online SMT system can incorporate information from unbounded incoming streams and maintain constant space and time. Crucially, we are able to match (or even exceed) translation performance of comparable systems which are batch retrained and use unbounded space. Our approach is particularly suited for situations when there is arbitrarily large amounts of new training material and we wish to incorporate it efficiently and in small space. The novel contributions of this thesis are: 1. An online, randomised language model that can model unbounded input streams in constant space and time. 2. An incrementally retrainable translationmodel for both phrase-based and grammarbased systems. The model presented is efficient enough to incorporate novel parallel text at the single sentence level. 3. Strategies for updating our stream-based language model and translation model which demonstrate how such components can be successfully used in a streaming translation setting. This operates both within a single streaming environment and also in the novel situation of having to translate multiple streams. 4. Demonstration that recent data from the stream is beneficial to translation performance. Our stream-based SMT system is efficient for tackling massive volumes of new training data and offers-up new ways of thinking about translating web data and dealing with other natural language streams.

Thesis (MPhil)--Stellenbosch University, 2001.<br>Full text to be digitised and attached to bibliographic record.<br>ENGLISH ABSTRACT: Bilingual dictionaries often do not satisfy the requirements of the dictionary users because they contain a relatively small percentage of articles displaying absolute equivalence between the source and the target language. This impedes the users from retrieving the required information. Communicative equivalence is a relation holding between source and target language entries in translation dictionaries. A translation equivalent should not be seen as a statement about the meaning of the lemma, but it should be regarded as an item that represents the target language and can be used to translate a specific occurrence of the source language item. If semantic and communicative equivalence do not hold between the source and the target language dictionary users will not be able to use the target language successfully. Instead of achieving communicative equivalence, the dictionary user is confused and ends up using language in an unacceptable way. Relying on translation dictionaries to improve their language skills and to learn more about the target language, dictionary users often come across haphazardly arranged articles in dictionaries that do not lead to communicative equivalence. In this case, the dictionary users are confused and end up using language in an unacceptable way. This thesis explores and investigates different ways to achieve communicative equivalence in translation dictionaries. The research formulates some of the needed guidelines for the lexicographers of translation dictionaries. It intends to assist lexicographers to compile useroriented translation dictionaries that can foster multilingualism amongst the dictionary users. As a result of this research, lexicographers will be able to compile translation dictionaries that will assist dictionary users to achieve communicative success. These dictionaries will establish semantic and communicative resemblance between the source and the target language forms. As a lexicographic problem, communicative equivalence is being addressed in the thesis and the researcher aims to make lexicographers aware of the mistakes in translation dictionaries. The objective of the thesis is to help lexicographers with the compilation of dictionaries where the users can achieve an optimal retrieval of information. The research focuses on metalexicographic issues in order to enhance the quality of practical lexicography. By doing so both the practical and theoretical lexicography can benefit from this research.<br>AFRIKAANSE OPSOMMING: Tweetalige woordeboeke voldoen dikwels nie aan die behoeftes van hulle gebruikers nie omdat hulle Onbeperkte aantal artikels bevat met absolute ekwivalensie tussen bron- en doeltaal. Dit weerhou die gebruikers daarvan om die verlangde inligting te onttrek. Kommunikatiewe ekwivalensie is onverhouding tussen die bron- en doeltaalinskrywings in vertalende woordeboeke. OnVertaalekwivalent moet nie beskou word as Onuitspraak oor die betekenis van die lemma nie maar dit moet gesien word as Onitem wat die doeltaal verteenwoordig en wat gebruik kan word om onbepaalde optrede van die brontaalvorm te vertaal. As semantiese en kommunikatiewe ekwivalensie nie tussen bron- en doeltaal bestaan nie sal woordeboekgebruikers nie daartoe in staat wees om die doeltaal suksesvol te gebruik nie. In stede daarvan om kommunikatiewe ekwivalensie te behaal, is die gebruiker verward en gebruik hy taal op Ononaanvaarbare manier. Wanneer woordeboekgebruikers op vertalende woordeboeke staatmaak om hulle taalgebruik te verbeter en meer oor die doeltaal te leer, vind hulle dikwels arbitrêr geordende artikels in woordeboeke wat nie tot kommunikatiewe ekwivalendie lei nie. Dit verwar gebruikers en lei tot foutiewe taalgebruik. Hierdie tesis ondersoek verskillende maniere waarop kommunikatiewe ekwivalensie in vertalende woordeboeke bereik kan word. Die navorsing lei tot die formulering van sommige riglyne wat nodig is vir die opstellers van vertalende woordeboeke. Dit poog om leksikograwe te help om gebruikersgerigte woordeboeke saam te stel wat veeltaligheid tussen die gebruikers kan bevorder. As gevolg van hierdie navorsing sal leksikograwe daartoe in staat wees om vertalende woordeboeke saam te stel wat gebruikers kan help om kommunikatiewe ekwivalensie te bereik. Hierdie woordeboeke salon semantiese en kommunikatiewe ooreenkoms tussen bron- en doeltaalvorme vestig. In hierdie tesis word kommunikatiewe ekwivalensie as Onleksikografiese probleem behandel en die navorser poog om leksikograwe bewus te maak van foute in vertalende woordeboeke. Die mikpunt van hierdie tesis is om leksikograwe te help met die samestelling van woordeboeke waar gebruikers onoptimale inligtingsherwinning kan behaal. Die navorsing fokus op metaleksikografiese aangeleenthede ter wille van Onverhoging in die gehalte van die praktiese leksikografie. Sodoende kan sowel die teoretiese as die praktiese leksikografie by hierdie navorsing baat vind.