Instrumente DE Traducere Asistată DE Calculator (tac) ÎN Societatea Contemporană

UNIVERSITATEA „LUCIAN BLAGA” DIN SIBIU

FACULTATEA DE LITERE ȘI ARTE

CATEDRA DE STUDII BRITANICE ȘI AMERICANE

LUCRARE DE DIPLOMĂ

COORDONATOR ȘTIINȚIFIC: CANDIDAT:

Asist. Univ. Dr. Andreea Maria Teodorescu Ana-Maria Neamțu

SIBIU

2016

UNIVERSITATEA „LUCIAN BLAGA” DIN SIBIU

FACULTATEA DE LITERE ȘI ARTE

CATEDRA DE STUDII BRITANICE ȘI AMERICANE

INSTRUMENTE DE TRADUCERE ASISTATĂ DE CALCULATOR (TAC) ÎN SOCIETATEA CONTEMPORANĂ: AVANTAJELE ȘI DEZAVANTAJELE MEMORIILOR DE TRADUCERE (MT)

COORDONATOR ȘTIINȚIFIC: CANDIDAT:

Asist. Univ. Dr. Andreea Maria Teodorescu Ana-Maria Neamțu

SIBIU

2016

LUCIAN BLAGA UNIVERSITY OF SIBIU

FACULTY OF LETTERS AND ARTS

DEPARTMENT OF BRITISH AND AMERICAN STUDIES

COMPUTER ASSISTED TRANSLATION (CAT) TOOLS IN MODERN SOCIETY: ADVANTAGES AND DISADVANTAGES OF TRANSLATION MEMORIES (TM)

SCIENTIFIC ADVISER: CANDIDATE:

Andreea Maria Teodorescu, PhD Ana-Maria Neamțu

SIBIU

2016

1. Introduction

During the past ten to twenty years, technological developments have not ceased to emerge and to determine changes and new approaches in all fields of society, including the domain of translation. In light of these recent developments, this paper aims to research and to identify the main technologies that have appeared on the market and to study what are the advantages and the disadvantages of these new tendencies.

In order to adapt to the 21st century and to be ahead of their competition, modern translators have had to acquire new skills and qualities the traditional translator did not have in the past such as computer skills and a much faster pace of working in order to be able to deliver large volumes of translated documents, of a higher quality and in a far more limited amount of time.

In the second chapter of this paper, I have tried to explain what the new tendencies in the domain of translation are and which causes have led to the status it has today by exploring the effects of globalization and cultural turn on modern society. The second part of the first chapter takes a closer look into what computer assisted translation is. It is explained why these software programs have become so popular and what are the main tools on the market. By means of a survey, some of the advantages and disadvantages of using CAT tools are presented and analysed.

Translation Memories (TM) are explored in the third chapter of this paper. We are aiming to understand how the concept of TM has emerged and why modern society was in need of such information technology programs. The reader can learn about the definitions and application of a TM and identify the principles that guide the internal structure of this type of database. The management of the translation process is explained step by step in order to give the reader an insightful image of how a source document is processed through the use of a translation memory.

The fourth chapter is dedicated to the advantages and disadvantages of TMs and the effects this innovative programs have had on the translation market and on the world of translation in contemporary society. It contains an analysis of a few interesting empirical studies conducted by different researchers in the field of translation. These case studies have allowed those interested to better understand how the human element and TMs interact. Through their understanding and analysis, I have been able to comprehend and identify some of the major benefits and some of the inconveniences of employing Computer Assisted Translation Tools into professional translators’ activity.

The world is changing at a rapid pace and translators are beginning to familiarize themselves with the new technologies existing on the market. This is one of the most important reason for which I have chosen this topic. As a young translator, I am aware that I will conduct my activity in a more technical environment and the research conducted for this paper has helped me take a glimpse into what the future hold for the young generation of translators.

2. Technological advancements in translation

The impact of globalization on translation

The professional domain of translation has been evolving in a specific context. Globalization has led to the emergence of expanding new demands in terms of translation. Owing to their international and multilingual character, a high number of companies also seem to express increasing demands for localization ( translation that takes into account the cultural particularities of a country or region) and terminology (the multitude of technical terms specific to a certain profession, study, domain) in order to adapt to the effects of this process.

The translation is undoubtedly linked to the global movement and it may be perceived as its vector as well as its product (Guidère 7). In other words, the translation can be a physical quantitative representation of the size, force and direction of the international process, respectively it is perceived as the result of globalization, given the linguistic needs of the industrial and economic fields.

The linguistic and cultural diversity of the contemporary society is sustained by linguistic politics and ambitious translation tools, as the international community is more than ever aware of the civilization elements intertwined with translation. The cultural turn, defined as the shift to an understanding of meaning in relation to a culture’s language rather than an approach based on a straightforward reading of state papers or economic data, has aroused the interest for research in translation studies, namely the research of linguistic differences, equivalence, and fidelity of translation.

According to Canadian Sherry Simon (1996), translation is “not only a simple transfer, but an original piece of writing and a conveyance of meaning in an ensemble of texts and discourses in the middle of society” (Gile 250). Other scientists concerned with the cultural turn movement have come to the conclusion that translation not only is integrated in the social and political background, but is also an active member of this context. Translation, in a broader sense, is the tool for examining concept of history, politics, ideology and identity.

Perceiving the cultural elements of a community may prove to be crucial for global companies because it can be what sets them apart from the competition. Adapting to the international needs of the countries businesses come in contact with leads to the implementation of translation into the process of communication responsible for a successful business relation, therefore the requirement for professional translations, proper terminology and localization is crucial.

The demand for translation services nowadays has been met with through information technology, which proves to be invaluable for the diminishing of costs related to the understanding and transmission of data (Shiyab 7). As a result, the English language has become more popular, reaching a world language status and the international demand for specialized translation is tremendously high. Therefore, the connection between internationalization and translation cannot be denied.

The innovations in technology brought about the use of a huge number of technical and nontechnical terms that the translator has had no choice but to adopt in order to enrich the target language and to raise its comprehensiveness to the reader. According to Juliane House and Nico Wiersema, “people as well as translators need to come to terms with the fact that words adopted from the target language can be enlightening to the reader as they genuinely mirror other cultures and their traditions” (Shiyab 8). In other words, adopting foreign terminology is sometimes a necessity and the premise for writing a piece of translation that is accurate in conveying the meaning of the source text and that proves that the translator has taken into account the specialized terms as well as the cultural and linguistic implications of the source language.

Usually, globalization and technology are mentioned in the same context, meaning there is an unbreakable link between the two concepts. Moreover, “the relationship between globalization and technology is that of cause and effect” (Shiyab 9). Due to technological innovations, globalization has emerged and spread as a consequence and this technology is employed in daily translation services everywhere in the world. In turn, translators have had to adapt their work style and reinvent their services.

Contemporary translators have the role of mediators among civilizations because they have managed to provide an improved understanding of different cultures. They have adapted to society’s recent increased attention to other communities and groups of people who speak foreign languages, a process caused by globalization, by the shifting demands of the labor market and by the age of the World Wide Web.

It can be assumed that in the near future, “the translation market will be more overpoweringly affected than any other market…because of the disappearance of national boundaries and custom tariffs” (Shiyab 9). The final product of translation will become a good purchased at an international level and as the demand increases, translators will have to keep up with the huge volume of work by making their work process more efficient and less time-consuming, although not in the detriment of quality. As a consequence, they’ll require the aide of new software designed to improve their efficiency and at least maintain their usual level of quality, although global requirements usually come with higher quality requirements.

In conclusion, globalization “can bridge the gap between peoples and nations” (Shiyab 10). Translating the source language and interpreting a new culture may be the most efficient tool at its disposal because the understanding of foreign language speaking nations is essential for the development of efficient and successful business, trade, social and political relations.

Computer Assisted Translation (CAT)

Brief history of the translation process

The translator’s daily routine has been revolutionized by the arrival of computer science and information technology. During the last century, he would type his text on a typing machine, simply write it by hand or dictate his interpretation of the source text to an assistant. After revising it, he would start over. The process of changing three words in a sentence would usually mean retyping the entire page. In contemporary times, such work conditions are unthinkable and overwhelming without the use of a text processing tool.

Owing to the invention of the personal computer (PC), translation industry embraced an invaluable development. The PC is nowadays used by each professional translator for editing pieces of writing, for looking up specialised terms in electronic dictionaries, other electronic resources or in their previous translations saved on their hard disk. The translator is provided with software tools designed specifically to augment his efficiency and decrease the complexity of the whole process in terms of time and cost.

Another technological innovation was the Internet, which had a major influence on the translation process as seen today. Today, the World Wide Web is used for electronic mail, which facilitates the sending and receiving of documents, for consulting international documentation resources comprising of hundreds of articles on topics pertaining to specialised domains. Terminology research is also facilitated through millions of documents, through the possibility of verifying the validity of the choices by looking up the number of appearances and validating the sources. Due to the internet, the translators can collaborate on projects with colleagues all over the country or all over the world by sharing texts, glossaries and translation memories.

The need for Computer Assisted Translation

Computer Assisted Translation (CAT) has evolved at a rapid pace since its origins and today it represents an ensemble of tools designed for optimising the work of translators. It is the process through which a human translator uses information technology (IT) tools in order to simplify his work and to reduce the time resource, aiming to successfully and adequately finish his translation to meet the deadline set by the client.

At the moment, professional translators are required to deliver higher quality through their work. In order to meet these demands they are faced with the analysis of the best information tools on the market, especially created to help them. The translator of the 21st century has no choice but to familiarize with the new technology implemented for computer assisted translations, which is rapidly multiplying and diversifying, and to decide which tools are best suited or are adequate for specific contexts in order to achieve the maximum of efficiency, accuracy and quality in translation.

Daniel Gouadec defines specialized translation as the translation of documents which refer to specialised subjects (technical or other) and/or a translation requiring the use of tools, techniques or procedures which imply a high level of knowledge (qtd. in Baron 1). If the second part of the definition is taken into consideration, an essential element for the translator’s activity is indicated: contemporary translators are obliged to know and use machine translation (MT) (automatic translation defined as an array of information tools , such as Google Translate, that are used to produce a machine translation, that is a translation which does not interact with humans and which uses substitutions of words to translate a text or speech from one language to another) or computer assisted translation (CAT) tools to help them with their translation as well as with the information research activity (Baron 1).

Marianne Lederer observes that “information technology is gradually integrated into people’s daily routines and consequently into the lives of translators. Whether we are referring to HAMT (Human Assisted Machine Translation) or MAHT (Machine Assisted Human Translation) , the machine and the translation process are inseparable today” (Lederer 141). Evidently, without consulting any type of electronic resources, it would be very hard for professionals to produce high quality translations, given the increasing workload and the tendency to reduce the time resource allocated for the completion of a task in order to maximize the profit in a restless society.

Computer Assisted Translation Tools

The emergence of the “traductique” (“traduction+informatique”), a French term used for the translation technology industry, in the 1990s, contributed to the development of technical linguistic tools designed for translators, but the applied translation studies status remained at an embryonic level. That reality was even more astonishing given the fact that the practical scope was broadening to reach the point where, today, it comprises all the information and communication media: the television (tools for subtitling and dubbing), the internet (localization tools for websites), the mobile industry (word processing and interpretation tools) and video chat (simultaneous interpretation).

All these fields have in common is the language component represented by translation.

Scientists interested in technological progress are concentrating their efforts on the development of the broad and evolving language industry, but their work concerns in fact redefined branches of Computational Linguistics, defined as an interdisciplinary field concerned with the processing of natural/human language from an IT perspective; the field requires the collaboration of linguists, computer scientists, experts in artificial intelligence, mathematicians, cognitive psychologists, psycholinguists etc. and it emerged with the human desire for automatic translation done by computers.

Since the booming of information technologies and the development of office and IT tools in the 1990s, the necessity of translators to simplify the complexity level of their work by employing the support of computer resources has intensely materialized. Similarly, the research conducted on human language has benefited from the endless possibilities provided by computers in order to reach new dimensions and facilitate the spread of new applications designated by the language industry.

The language industry represents the economic sector which covers all the products and services for the processing of languages that exist on the market as well as all the activities which require an automatic processing of natural language. This industry is based on the research in linguistic engineering, which is the applied part of computer linguistics. This term indicates all theoretical models of language description for computer processing. This branch of linguistics is interested in the development of techno-linguistic tools for the processing, interpretation, generation and understanding of written or oral human language.

Computational linguistics emerged against this background: it is the result of introducing informatics into the practice of translation. This area handles the analysis and development of linguistic resources and software tools which support the translation of general or specialised language in a bilingual or multilingual context (Bouillon et Clas qtd. in Guidère 134). Its objective is to automate all or part of the tasks involved in the translation process. Noteworthy is the fact that this automation is not always consistent, thorough or complete for two main reasons. Firstly, because not all steps of the translation activities possess the ability to become automatic. The field of translation will never be taken over by machines because, despite their extraordinary evolution, computers will never have the level of comprehensiveness needed for a successful interpretation of a text written in a foreign language. Secondly, because some stages of translation are not meant to be automatic. For example, a computer program is able to generate a lot of meanings for a phrase by taking into consideration different contexts saved in the software’s back-up system for retrieval upon demand, but a machine is unable to choose the best option for that specific situation simply because it does not have the subjectivity and insight of the human brain. This is one of the motives for the fact that this profession in particular will never be replaced entirely by computers. The human element and the translation process represent an undividable ensemble.

Therefore, the software for computer assisted translation, which automates a certain number of tasks, belongs to the ambit of computational linguistics. This mechanization presents two advantages: not only has it eliminated the laborious repetitive stages by reusing source code (an array of specialised terms dealt with in previous technical texts) from older projects, but it has also offered translators the possibility to optimise the translation process by providing them with CAT tools that reduce the amount of time required for the documentation, formatting (editing) and processing of written sources.

Translators can certainly identify useful IT programs within the scope of this information technology sector, tools that might simplify their activity to a high degree. Nevertheless, professionals play an important role in the developments of these CAT tools. His training, knowledge and aptitudes are involved in the software’s development process because each text filtered through the program leads to the enrichment of its terminological database and future sessions will deliver more accurate translations and more choices in meanings for a certain word, phrase, sentence, paragraph and so on.

There is a multitude of complex and diverse types of computer aided translation tools: electronic dictionaries destined for automatic translation, computer assisted technical writing tools, spelling checkers, morphological analysers and parsers ( software program used for syntax analysis ), concordancers (pieces of software that enable the user to search and identify words or phrases in a certain corpus), automated term extraction systems, statistical processing pieces of software for language data, software for documentary review and automatic indexation and others.

Therefore, it can be observed that computational linguistics is the result of the intersection between multiple disciplines and that it raises several topics: firstly, the methods for the analysis and processing of translations; secondly, the tools designed for the structuring and manipulating of data; thirdly, the IT applications for assisting translations. However, the boundaries of this branch are not yet clearly defined: some scientists consider this field to be the applied part of translation studies while others connect it to applied linguistics or automatic natural language processing (NLP), or to the language industry. Consequently, today computational linguistics is a booming domain, but subject to significant tensions and divisions: on the one hand, practical applications versus theoretical research, on the other hand, the use of linguistic and information technology techniques versus traditional traductological approaches (Guidère 134). The tensions mainly derive from the diverse related disciplines that form the theoretical basis for it: linguistics, informatics, traductology, artificial intelligence or cognitive sciences. Researches usually fall into one of the abovementioned categories, but the subjects they analyze and their objectives differ according to the point of view chosen regarding translation.

“The tools developed by computational linguistics can be classified into two large categories: software (independent programs, modules and components and data management tools) and linguistic (automatic spelling checkers, electronic dictionaries, terminological engines and others) applications” (Guidère 135). This distinction has not ceased to develop throughout the last decades and belongs to the literacy of a domain constantly evolving: the need to translate incoming texts, the possibility to access significantly large volumes of translated texts, the constitution of formal patterns for automatic translation and the increasing demand for terminological extraction and multilingual documentary monitoring.

Translation tools (such as translation memories or bilingual concordancers) designate techno-linguistic modules integrated in computer assisted translation or automatic translation systems. They consist of IT programs based on translation data reusable in other contexts. The researcher of applied traductology must first start by understanding how the existing tools operate before attempting to enrich them or to design even more efficient ones.

CAT tools were designed to assist professional translators during their work in order to enhance their productivity and simplify their activity by moving at a faster pace. Amongst the tools for aiding translation inventoried by Daniel Gouadec in his paper “Le traducteur, la traduction et l’entreprise” written in 1989 the following can be identified: the construction of documentary resources (linguistic and technical resources) and their development. He underlines the fact that the terms “CAT tools” refer to computer integrated resources which include text processing software that enable the access to one or a few data banks or to text files or memory Help files. These resources allow the reviewing of older translation projects, the communication between translators working in the same company, the simultaneous access to a certain terminological resource file, automatic extraction of recurring units, the insertion of already encountered terminology into the translation in progress. Spelling and grammar checkers and parsers (syntax analysers), electronic terminology processors (with or without access after the processing of text is complete, with or without automatic replacement of terms from the source text with dictionary entries, software for “assisting translation” ( text analysers, translation hypothesis generator), software for automatic translation (AT software) for complete translations, assisted drafting programs and editing systems, they all fall into the same category of CAT tools ( Gouadec, “Le traducteur, la traduction et l’entreprise” 78).

Beginner translators often wonder how CAT tools can help them. There are several reasons that must be considered when deciding whether to purchase this type of software or not. To begin with, the ability to accept more work is a noteworthy advantage. The localization needs of big companies require the fast analysis of large volumes of data in a rigorously consistent and coherent manner from the point of view of terminology and appropriate comprehensiveness. With the help of CAT packages, these demands can be dealt with. Translators capable of importing translation memories, of updating them through their projects and to share them with their colleagues can become indispensable employees and play a vital role in the localization process.

Secondly, even in traditional environments, a CAT tool can help improve productivity. Given a pre-translation tool and a translation memory reasonably enriched, the amount of time allocated for the translation process can be halved. However, it must be taken into account the similarities and concordances between the new source files and the content of the translation memory. Moreover, even if the document has an entirely new subject, the CAT tool still simplifies the work by aligning the phrases on the screen and once a certain technical term is translated, the program identifies all the appearances of the word or phrase and replaces them with the choice provided by the user. In addition, the translation memory is extended by automatic insertion of the new terminology and it will be at the translator’s disposal for future processes.

Thirdly, technical translations (such as legal or financial) require consistent terminology. Readers of different nationalities and cultures must be able to easily understand the text; therefore, the employment of synonyms or technical terms specific only to a certain civilisation would make the understanding of the translated document difficult and time-consuming, which would not please the potential client or the hired professional.

CAT tools use alignment for the target and the source text (two side-by-side columns segmented by using words, phrases or paragraphs as translation units, depending on the entries in the project’s translation memories). This characteristic provides a higher rate of efficiency and extraordinary quality by evidently reducing the number of errors in the final document.

Finally, a thoroughly conducted analysis of the tasks, revision, glossary and translation memory updating involved in a new translation project enables the translator to accurately assess his workload and charge a fair price for his activity. CAT tools are equipped with word, segment and unit counters that will aid him to reach a final and accurate decision on remuneration.

A statistical analysis of the effectiveness of CAT tools

Older translators’ activity consisted mainly of interpreting the source language into the target one. However, in the last decade, an increasing accent has been placed on their computer abilities and their competence to use diverse software packages to aid them with their translations. The greatest disadvantage of this aspect is that, even today, there are no specific trainings designed to help them acquire the level of competence needed in order to access this kind of IT resources, which are highly complex and not quite easy to handle. Consequently, the answer of whether a CAT tool should be used or not remains very popular nowadays. Researchers have been struggling to find out the spreading of this technology by conducting national surveys among professional translators. So was the case of Mgr. J. Absolon who, in November 2008, was in charge of a survey carried out on a European sample group of translators in order to test the efficiency of CAT tools.

Firstly, he wanted to find out to what extent computer assisted technology was being used. After analyzing the results, he gathered the following information, which can be observed in Figure 1: 67.6% of translators who were questioned said they were aware of the plethora of CAT tools existing on the market, but only 54.1% actually used them regularly. With barely half of them actually using the software, it can be observed that the complexity or the price may prove to be too much for the ordinary translator, considering the fact that there are no special preparation workshops to help them gain insight into the intrinsic structure of the software.

When questioned about the name of the software packages they have purchased, the majority of the sample group (54%) responded to have installed and run SDL Trados or SDLX, owned by the same company, as seen in Figure 2. SDL Trados is closely followed by Wordfast with 19%. Other products utilized by translators are: Across, Star Transit, Idiom Worldserver, Deja VU, Logoport, MetaTexis, OmegaT, Catalyst Lite, LocStudio, MS Localization Tools and Passolo. The result show without a doubt the popularity of SDL Trados and it can be concluded that, out of the array of software packages available for purchase, it is the most user-friendly (which is a vital aspect when dealing with IT technology) and it most surely has the best cost/ benefits ratio.

One of the most important aspects of the survey was the conclusions drawn regarding the volume of translations. As seen in Figure 3, most translators use the CAT tools for half of their work. Though it may not seem important at first glance, it must be noted that the amount of time put into the translation is halved. Therefore, professionals are able to accept and deliver twice the volume of work they would have delivered otherwise if not for the software, which represents a huge advantage.

“Comparing the productivity of translators who use CAT tools with those who do not, the results are clearly in favour of CAT users” (Absolon 3). This shows that, despite the complex interface of these programs, their effectiveness and benefits cannot be contested. Translators in the sample group have acknowledged the fact that with computer assistance they are able to deliver an average of 10 pages per day in comparison with traditional translators, who measure their work at 2.5 pages per day. There is a huge gap between the numbers shown in Figure 4, which proves the fact that the tendency of the contemporary translation domain should be in favour of utilising computer aided translation tools.

“Comparing the monthly earnings of translators who use CAT tools with those who do not, the results are again clearly in favour of CAT users” (Absolon 3). The financial part is, apart from the passion of doing a certain type of work, a strong motivator for employees in all fields of society, especially in the world of translation. As seen in figure 5, translators with strong software skills are able to earn more than 1330€ a month, approximately 500 to 1000 € more than traditional translators. Therefore, a solution for encouraging young talented minds to choose a career in translation would be the advertising of such software packages, which improve efficiency and quality as well as profits.

In conclusion, the spread and need of CAT tools in the context of modern society is a reality which cannot be ignored. “These findings should be a signal to translators that the times when using CAT tools were a big marketing advantage for translators are changing to such tools being a necessity” (Absolon 4). If tens of years ago, the traditional translator was content with his earnings and was not bothered with the complexity of the translation process, during the last decade, mentalities have shifted and demands have increased in terms of quality as well as volume. Therefore, the modern translator has no alternative but to adapt to the rapidly changing and evolving environment in order to meet these requirements. Such surveys should be conducted in all parts of the world and an emphasis should be placed on motivating universities to insert CAT training in their curriculum with the purpose of preparing talented young people for the shift in the expectations of employers on the labour market or in order to be as successful as possible as independent translators.

Advantages of using CAT tools

One of the most important contributions brought by the Computer Assisted Translation is the automation of the translator’s work. In fact, CAT tools enable the enhancement of multilingual data by automatically recycling existing translations. The aim of this technology is to never translate the same sentence twice.

By eliminating repetitive tasks, the translator can minimise the time assigned for a certain translation and consequently is able to avoid not meeting the deadlines set by a potential client. Once the time resource is augmented, the translator allows himself to handle more work, thus enhancing his earnings. In other words, CAT programs increase the productivity of its users as well as their profitability. For example, SDL Trados has published a statistics which shows that each new source document created by using saved already translated ones contain at least 30% translated code. The updating of documents may lead to the conclusion that they present 90% or more repetitions, fact that benefits not only the translator, but also the companies that contract them. In terms of delaying a document, the time saving may amount to somewhere between 20 and 25%, which means approximately one day a week. SDL Trados provides its future users a cost simulator that can be used to measure the economical aspects that result from the employment of Computer Assisted Translation software packages.

From a financial point of view, translation memory systems allow at the same time a significant reduction of costs with Desktop Publishing in the post-translation stage by accelerating the process. The copy/paste operations of the translated text are eliminated and the target document may be reinserted into Desktop Publishing files in the right place, in the appropriate format and maintaining the correct fonts. Due to recent developments of CAT tools, they are able to process any language from European to Asian or Arabic and any document format: Word, Excel, HTML, but even more complex and less used ones such as FrameMaker (Peraldi).

CAT tools allow the creation of coherent translations by matching the adequate terminology, thus improving the quality of target documents that need to be delivered to the clients. Moreover, these tools enable the homogeneity of group translations by the option of sharing language resources and due to the memory network and to different available glossaries uploaded by other users. The programs ensure the right terminology due to the translation memories, but to terminological extraction applications as well. Finally, the ensemble of tools aids the complying with a company’s individual style.

Disadvantages of using CAT tools

Firstly, the purchase of CAT software represents a significant financial investment which has to rapidly improve profitability, especially if the buyer is an independent translator at the beginning of his career. Fortunately, the ensemble of companies developing this type of tools have come to an agreement on a standard format, TMX ( Translation Memory Exchange), which allows the use of any type of memory regardless of the official product that created it ( for example Trados or Déjà VuX).

Secondly, CAT largely depends on the type of document processed. CAT applications are numerous: user manuals, websites, financial reports and statements, press releases, virtually any type of document specific to businesses is a perfect choice for computer aided translation due to its traditional template. On the downside, if the updating of a document is superficial or if the source is literary, even the case when a translator generates target documents that he will never use again, in other words, if a client does not have returning customers, then the purchase of a CAT tool is not justified.

Regarding the time resource, it is safe to assume that if the terminological glossaries are not constantly improved and updated, then the program will not be able to translate an important percentage of the document. Moreover, the creation of translation memories wastes a lot of time because it consists in the alignment of the source and target text, which is in reality far time consuming. In reality, an alignment program is merely e mechanical system devoid of human logic. The software analyses the sources only as a string of symbols such as letters, spaces and numbers and it is due to these symbols that he is able to recognize and generate the correspondence between segments. Unfortunately, this process is not without fault and it may lead to quite a large number of errors. If the alignment process becomes complicated because the source contains tables or images or any element susceptible of perturbing the machine’s functioning it leads to errors. Thus, the translator will have the annoying task of verifying the accuracy of the computer process in order to correct it and rearrange the text for optimal results. But, in the case of Similis, a CAT product, which is equipped with a parser , the alignment is analysed based on the elements of syntax and not just the punctuation (Peraldi 1), which gives a better balance between the work of the translator and the efficiency of the product. It must be reminded the fact that the user is never paid for the alignment or for the updating of glossaries of terminology. Thus the time spent with the construction of translation memories and terminological dictionaries means less time spent translating, which ultimately leads to less money.

Finally, the principle of segmentation might affect the quality of translation. By immensely focusing on sequences, the general meaning of the document and the cohesion might pass into second place. In fact, a user never translates isolated sentences; the whole context must be taken into consideration. Therefore, the translator will have to be sure of the accuracy of the segments generated by the machine so that he might not miss the semantics or the syntax errors which ultimately affect the global meaning of the text. Moreover, language evolves over time. Thus the translator will have to permanently be aware of the shifts in meaning that may occur over time or the new vocabulary that might emerge, given the daily development of technology.

Conclusion

In conclusion, there are more than a few factors which influence the relevance of Computer Assisted Translation Tools: knowing the type of the source text, the quality of the memory translations that back up the system, the quality of translation and especially the time resource.

The ups and downs above mentioned are a possibility. Nevertheless, the translator must not ignore these tools or reject them; it would be wiser for him to learn to adapt to what is necessary. The user must have an insight into the strong points and the weaknesses of the software before using it so that he may take advantage of the full potential these programs have to offer. Most of all, the potential buyer must analyze his needs and priorities before deciding to make the purchase.

3. Translation Memory (TM)

Brief history of Translation Memories

Translation memory (TM) technology is the most widely spread and used computer assisted translation tool on the market. It appeared in the middle of the 1990s and its popularity has not ceased to grow since (Christensen Schjoldager 1). An attempt to define a translation memory would be: “A TM is basically a database of segmented and paired source and target texts that translators can access in order to re-use previous translations while translating new texts” (Christensen Schjoldager 1).

Given the fact that the TM is a database, it can be stated that such a database may be edited, updated, improved or deleted like any other database in the IT world.

This technology is used in modern society on a regular basis, in a large number of jobs and for more languages, clients and translators than ever before. The overwhelming argument for embracing it appears to be the benefits that the system can bring in terms of productivity, cost savings, time efficiency and quality of the translation output. Indeed, it is no exaggeration to say that the advent of TM systems in the translation profession has led to drastic changes in translators’ processes, work-style and workflow. Despite the fact that many professional translators nowadays depend on some form of TM system, neither CAT in general or TM-assisted translation in particular has been the object of much research.

“By pointing out that translators working for the European Commission were wasting valuable time by retranslating (parts of) texts that had already been translated, Arthern suggested the compilation of a computerised storage of source and target texts that could easily be accessed by translators for re-use in current translations” (Christensen Schjoldager 2). This shows us that Arthern may be considered the pioneer of modern translation memory technology. He referred to this mode of translation as “translation by text-retrieval” (Christensen Schjoldager 2).

Another extraordinary idea was conceived by Kay and described in a 1980 paper entitled “The proper place of Men and Machines in Language Translation”. Unlike many of his contemporaries, who still assumed that machine translation (MT), for example fully automated translation, would soon be a feasible alternative to human translation, Kay argued that human translators should stay in control of the translation process and that the way forward was to develop computerised tools that could assist human translators in their work. Kay’s main idea was to develop and add specific translation tools to existing text processing technology. These tools should include various means for translators to keep track of earlier decisions and to access previous translations of the same or similar source-text passages, for instance. Another fundamental idea for the development of TM technology was proposed by Melby (1981) when he suggested computer-generated bilingual concordances as a tool for translators and documented that such a tool would enable translators to identify text segments with potential translation equivalents in relevant contexts.

The realisation of commercial TM systems was first made possible when tools for text alignment made bilingual databases of translations possible. Based on these ideas and technical developments, four commercial TM systems appeared on the market in the early 1990s: The Translation Manager from IBM, the Transit system from Star, the Eurolang Optimizer and the Translator’s Workbench from Trados (Christensen Schjoldager 2). Since the appearance of the first TM systems on the market, the dissemination of this particular technology has kept on growing and continues to be innovated even today by computer programmers who search to create large predefined TMs for import and by improving the characteristics of the database design.

Definitions and application of TMs

Computer-assisted translation (CAT) covers human-aided machine translation (HAMT) and machine-aided human translation (MAHT). In HAMT, translation is essentially carried out by the program itself, but humans are required to resolve specific language problems arising from the source text or to correct the resulting target text. In MAHT, translation is carried out by a human translator, but computer assistance is seen as an integral part of the process.

TM technology is the computer tool that is most widely used by individual translators, translation agencies and other organisations involved in translation. While TM systems may differ in internal processes that govern segmentation, alignment, indexing, searching and match retrieval, they all share the basic function of deploying existing translation resources in a new translation project. In some cases, this is combined with other types of software such as word processors or terminology.

As already mentioned, a TM is basically a database of segmented and paired source and target texts that the translator can access in order to re-use previous segments while translating. The TM continuously provides the translator with so-called matches, which are translation proposals stored in its database. TM systems operate with three kinds of matches: exact, fuzzy and no matches. The identification of these matches relies on an automatic comparison of character chains. Exact matches are found if the character chain of a new source-text segment is identical to that of a stored source-text segment; fuzzy matches are found if source-text segments are merely identified as similar; and no matches are found if no source-text segment is identified as (sufficiently) similar. The thresholds between these matches tend to be preset by the TM programme, but may also be set by individual translators.

Most TM systems define an exact match as a 100% correspondence between source-text segments; a fuzzy match is defined as a 70-99% correspondence; and no matches are found if the correspondence is below a 70% threshold, in which case the target-text segment is left empty. It is worth pointing out that the use of character chains for match identification is an entirely form-based process, which does not consider semantic, pragmatic or contextual aspects.

TM segments tend to be sentences or sentence-like units such as titles, headings, list items and table cells. Some researchers discuss the negative consequences of this and argue that TM segmentation below the sentence level may be more useful. Thus, for instance, Schäler (2001) draws attention to a possible relation between TM segmentation and translator productivity, Dragsted (2004, 2006) studies the relation between sentence-based segmentation and the translator’s cognitive process, and Colominas (2008) experiments with chunk-based TM systems.

Translators may engage with TM technology in a variety of manners and in diverse domains of modern society which require technical translations. In most commercial systems, the translator works with an interface that is characterised as interactive. This means that source text segments are presented one at a time, giving the translator the option to translate this segment, or, if an exact or a fuzzy match is retrieved from the TM, to accept, revise or reject the previous translation of this segment (Christensen Schjoldager 3).

Translators may also engage with TM technology using a pre-translation mode. This means that TM technology is applied to the source text prior to its translation, resulting in a hybrid output of exact and fuzzy matches from the TM plus some empty segments. Using a pre-translation mode, the translator concentrates on translating the empty segments and editing the target-text segments that have been suggested by the TM. This is a very useful advantage of TM technology because it reduces drastically the amount of time spent by the user on a text which contains a large volume of terms previously encountered and translated.

Translation process management

Perhaps the most intriguing aspect of translation memory is its ability to aid the user in managing projects, coordinating team efforts and building glossaries and dictionaries. Following are some additional features of TM that allow the translator or other user to manage translation projects more efficiently.

Internal attributes of a Translation Memory

Most TM products not only store language pairs; they also store other information, called attributes, with the pairs. The most common attributes stored include the creation date, the name of the user or creator, the client, the project ID and the main domain or field (e.g., legal, technical, etc.) of the translation. Once this information is stored with the translated segments, the translator or other user can filter the text for the most important attributes.

For example, the user can look for similar text segments by project, client, etc. when performing fuzzy matching, or a project manager may have more control over accountability for translated texts by filtering for creation date or the name of the creator of the translated segments. The latter is particularly useful when a number of translators are working on one large project, especially when the translators are all working with the same language combination (for example English and Romanian).

Terminology databases

Most TM products come with a terminology database so that the translator can take full advantage of all of the features of TM. Using an integrated terminology database allows a translator to perform fuzzy matching for a specific term or to use a term in the database suggested by TM. Without a terminology database that is compatible with translation memory, the TM user cannot easily obtain suggested translations for individual words without opening a separate electronic dictionary or looking through a conventional dictionary. Naturally, the user must enter the terminology into the database before it can be useful. Once the terms are in the database, however, an individual translator or team of translators can work on a project and receive the suggested terms from the database, maintaining terminological consistency throughout the translation.

The analysis of source documents through TMs

The ability to estimate in advance approximately how much time a project will take is not always an easy task. If the translation memory system is a good one, it will have the capability of analyzing a document for similar sentences and text repetition. It will also provide raw word counts, ignoring elements like graphics, HTML tags, software code that could influence the count. This analysis makes it easier for the translator or project manager to assess whether or not translation memory will be useful for the project and also helps him or her determine how much time may be involved in translating the document, depending on the amount of repetition, the word count, etc.

The user may also use the analysis function to compare different documents for similarities. Analysis can reveal if one document that has been translated previously and a newer document are in any way similar. Depending on how similar the two documents are, the user can estimate the time required for translation, which is an important advantage that places him ahead of his non-user of CAT tools competition.

Texts that are conducive to using translation memory reusability

The most important characteristic of a text that is conducive to translation memory is that the text will be reused in one way or another. Following are examples of how texts can be reused and how translation memory becomes involved in the process.

Updates

A not uncommon occurrence during the translation process is when an update of the text being translated is suddenly made available to the translator. An update is a change in a source text that occurs while the translation is still in progress. Receiving an

updated text can cause major difficulties for the translator if the text is large and changes have been made throughout the entire document. Making updates using translation memory has the advantage over the conventional update process in that the translator does not have to physically search through the entire document for changes.

Instead, the translator only has to run the updated source text through the translation memory program to identify new or changed segments and any new terminology. New terminology can be entered into the terminology database by the translator for future use.

In order for translation memory to be effective, all work must be done in TM and saved in TM format. Anything done outside of TM will not be stored in the memory database and therefore will not be a translation that can be manipulated in the future, unless one has access to an alignment tool. The best way to approach TM is to think about it as being an integral part of the main word processor, just like the word processor’s spell checker. If the TM system is a stand-alone product, always keep a copy of the text file that retains the TM product’s file format.

A translator can even begin the translation process before the final original document is completed. If the translator is given drafts of the original document in its early stages of development, the text can be translated and stored in the TM database.

Then, as updated sections of the text are made available, the translator can perform fuzzy and exact matching, thus isolating the new parts from the parts that have already been translated or that are similar to the original. Section 6.5 is an example of this

process.

Revisions

Many translators find that they continually receive revisions from the same clients. A revision is a new project amending a prior translation, reflecting changes made to a prior source text. Often a translator is asked by a client to revise the translation of a manual for the current product model that will be released within a short period of time. The client wants the translated manual to be available at the same time

ADVANTAGES AND DISADVANTAGES OF TRANSLATION MEMORY

16

that the product is launched on the market. If the translator were to use the

conventional translation process, it could take months before a very large document

would be ready, and the client might not have that much patience or time. If, however,

the translator uses translation memory, he or she can analyze what has changed within

the document and can provide the revised translation of the manual within a shorter

period of time than if he or she had used the conventional process. Section 6.4

illustrates this process.

4.1.3 “RECYCLING” PRIOR WORK

At times, a translator may find that he or she is translating a text very similar to

one that had been translated in the past. The translator may run across words or

phrases that are almost identical to words or phrases in the older document. The odds

that a translator will ever translate the same sentence twice in two different texts is very

low; however, the odds are higher that a translator will run across similar phrases or

words in texts within the same field and/or for the same client. If the translator has an

electronic copy of the target and source texts from the previous translation, then he or

she can quickly access the files and perform fuzzy matching with the new source text

against the old source and target texts.

REPETITIVE CONTENT

Another important factor is whether or not there is repetitive content within a text.

The higher the percentage of repetitive content within a text, the more desirable it is to

use translation memory. Repetitive content may include words, phrases or entire

paragraphs. There are a number of different text types, but some tend to have more

Advantages and disadvantages of Translation Memories (TMs)

The domain of Translation Memories has aroused the interest of many research scientists, but the literacy that exists on TM systems is more theoretical rather than practical. In order to accurately determine which are the advantages and disadvantages of these databases, it is important to look for and comprehend the empirical case studies conducted by different scientist in the 21st century or more accurately during the past 10 to 15 years. Empirical papers, “by definition…analyse and discuss data. In empirical TM research, relevant data appear to be source texts, target texts… and translation processes (both internal and external) ” (Christensen Schjoldager 4). In other words, all the elements that a TM interacts with during a project created by the user. After thoroughly investigating a large number of studies conducted by fellow colleagues, Christensen and Schjoldager have identified nine accurate studies that have taken place during 2000 and 2008. The list of researchers the two have decided to further investigate contains the following names: Lange/Bennett, Christensen, Dragsted, Fulford/ Granell-Zafra, Dillon/Fraser, Lagoudaki, Mandreoli et al., O’Brien and Colominas (Christensen Schjoldager 4).

Lange and Bennett

Aiming to expand their knowledge of translation memories, Lange and Bennett decided to conduct a six-month long project in order to explore the downfalls and advantages of the interaction between Machine Translation (MT) and TM ( Translation Memory). The project intended to further investigate how information technology might successfully aid the translation of help texts in the virtual world in order to achieve superior quality result in a more limited amount of time. The purpose they set was to achieve a halving of the time invested in the process by employing automated translation. “ The project was carried out in four phases: 1. Analysis of MT, 2. Analysis of a combination of MT and TM, 3. Analysis and readjustment of the translation workflow and 4. Enhancement of the translation output.” (Christensen Schjoldager 5). Once the experiment reached its final stage, the two researchers concluded that “translators’ productivity may indeed increase if TM and MT are combined…productivity will only increase in this way if translators are comfortable with their new role as post-editors of machine-controlled translations”. It is true that the ancestor of CAT tools was the machine translation and it was a big success when it emerged on the market. Despite its success, MT lacks the human component of the process; without it, the machine will never be able to deliver a coherent translation of a certain quality for the simple fact that it lacks cognitive functions only the human brain possesses. Given the fact that the profession of translation is a highly interactive one, it is doubtful that professionals will ever be fully content with the diminished role they are starting to play during the process of translation.

Moreover, this career requires maximum focus on the job in order to deliver a coherent target source; if the independent translator is disgruntled with his activity, with the work-style and workflow, then that will reflect in his work. He may end up losing his most trusted client simply because he is unhappy with the fact that he may become, in the near future, indispensable.

On the other hand, if a translator hired by a big company is working in a frustrated manner, he may ultimately affect not only his work, but that of his colleagues, too. Big companies usually purchase and implement the most expensive and effective tools on the market for two main reasons. To begin with, they have the money and they think of the business advantages in the long run. In a few years or even months, the financial investment will prove to have been an extremely inspired choice. To end with, given the high number of employees the company might have, it cannot risk its future on the market by choosing price over quality. Even the slightest mistake might prove to be fatal and it could end up losing V.I.P. customers, lasting partnerships with other companies or even a competitive place on the market for failure to meet deadlines or the quality expectations. If such an enterprise should choose a less efficient tool, the amount of work in post-editing translations put in by the employee would be far larger; therefore the cheaper the product, the more unsatisfied the translator. Expensive CAT tools on the market, which are the environment for the design and updating of translation memories, enable all the employees to communicate, to share projects, ideas, glossaries, personal dictionaries, translation memories, terminological databases. If a project is a group translation and one of the members of the team fails to produce maximum quality, this will automatically affect his colleagues, his company and will be the prerequisite for the failure to deliver the target source on time.

In conclusion, translation memory and machine translation taken together may prove to be an inadequate interaction due to the negative aspects of machine controlled translation, lacking the insightfulness of the human mind.

Christensen (2003)

In 2003, Christensen was interested in researching the usefulness of a Translation Memory (TM) for legal documents. Her dissertation was constructed upon the observation that “because of the complexity and culture specificity of legal communication, TM technology will be less useful for legal translation than for technical translation, for which it was first designed” (Christensen Schjoldager 5). What Christensen did was to mix relevant theories and methods encountered within legal studies or jurisprudence, translation studies, computer science (informatics) and linguistics.

The main three aspects she was interested in during her research were to find out if TM technology is helpful in connection with translation problems specific to the legal domain, the degree of usefulness of TMs for legal translation in general and last but not least important, if it’s necessary to combine a TM with a reference corpus of authentic target-language texts that serve similar functions (Christensen Schjoldager 5).

Based on her empirical study, she concluded that TM presents a major disadvantage when it comes to legal translation as TMs appear to require the access to “functionally identical segments in an authentic target-language corpus” (Christensen Schjoldager 6).

Despite her negative or disappointing results of her research, the most valuable aspect that must be taken into consideration when speaking about the benefits and inconveniences of translation memories is the fact that TMs can always and permanently be improved through the interaction with human users. If a TM is not suited for translations from a certain domain, it can be adapted to another field or it can be updated and improved to meet the requirements of the user.

One solution for the problem raised during her experiment might be the concept of authoring memory. In other words, the TM may be provided with an authoring memory meaning one that would generalize the source-text segments by allowing the pre-editing before the computer proceeds with the alignment with target-text segments.

Dragsted (2004) and (2006)

Dragsted conducted her research with the aim of finding out how translators’ minds are affected by the rules of sentence-based segmentation imposed by the Translation Memory systems. Dragsted was interested in three major issues: “How do translators segment text naturally? 2. How does the integration of TM systems (with enforced sentence-level segmentation) affect the cognitive translation process? 3. How can TM systems be optimised to conform better to translators’ natural segmentation and to render higher match values?” (Christensen Schjoldager 6).

The sample group of the experiment was represented by six professional translators which had worked professionally in the field of translation for minimum two years and six final year master students in the branch of specialised translation at the Copenhagen Business School.

Dragsted led two experiments. The first experiment was focused on human translation recorded with the help of a Translog (a key-stroke logging). The other experiment she researched Translation Memory-assisted translation with the help of the most popular CAT tool on the market, SDL Trados.

“Dragsted drew on…the translators’ own understanding of their choices and the text that they were given to translate…She focuses on the translators’’ revision time, the extent to which the source-text sentence structure is changes and the way in which translators appear to segment the source text” (Christensen Schjoldager 7). The translators’ style of segmentation was identified by recording the amount of time between key-strokes.

Through her experiment, Dragsted found out that in the cognitive process of human translators, the sentence-based segmentation imposed by the TM is not the central unit, with emphasis on the case of professionals rather than the younger minds of the students. The experiment concluded that experienced translators were much more affected by the rules imposed than students; to such a degree was their level of discomfort that their revision time was far longer than the time taken during human translation.

A disadvantage of TM that can be drawn from this research is the fact that sentence-fragmentation might not be such a good idea. It is understandable why the professionals had a hard time using the Translation Memory process because, given their experience, their work-style was far more elaborate than a sentence parsing type. They had trained their mind to take into consideration all aspects of the context, the syntax as well as the grammar, the punctuation. Furthermore, given the globalization tendency of modern society, the accurate translation of a certain document written in a foreign language will have to take into consideration the localization aspect. There are civilisation and cultural aspects that might need the translator’s attention. Therefore, without at least the whole paragraph, it is very hard, if not impossible to grasp the global meaning of the text, its formal or informal character and stylistic aspects of the language.

Nevertheless, one major advantage that Translation Memory Systems present today that might balance the inconvenient aspect of sentence-based segmentation is the possibility of TMs to be enhanced by syntactical and morphological parsers that would generate sentence fragments as basic units of translation. What a parser does is to analyze a text by looking at its syntax and grammar and divide it following a logical order, which would generate a higher quality of segmentation.

The reality is that, in the future, Translation Memories will have to adapt to each translator’s individual style and manner of segmentation. Young translators may learn to work in the TM’s way from the very beginning, but older professional translators will have a hard time adapting to this new technology if it is not improved to suit their needs.

Fulford/ Granell-Zafra

The two researchers were interested in a professional translator’s capacity to understand and adapt to information and communication technologies (which include Translation Memory Systems). The study they conducted in 2003 was the type of a survey among UK freelance translators. The sample group numbered 591 freelance translators from the United Kingdom.

Their findings were rather surprising: “TM technology and other CAT tools are less widely used than we might expect: just under half of the respondents … are not really familiar with CAT tools at all, and only 28% of respondents state that they actually use TMs and other CAT tools.”

What this meant was that Translation Memory terminology management tools were not so popular among independent professionals. It might be understandable, considering the amount of work they are hired to do is far lower than the activity of translators working in a corporation.

Dillon/Fraser

Dillon/Fraser (2006) provide a snapshot of UK-based professional translators’ personal views on TM technology. Reporting on one of very few studies that focus on the translators’ perspective, the paper sets up three hypotheses:

(1) Novice translators have more positive perceptions of TM technology than more experienced translators.

(2) Translators with TM experience have more positive perceptions of the technology than translators without such experience.

(3) Translators’ perceived IT proficiency is not the main influence on their perception.

While Dragsted’s professional subjects complain that the sentence-by-sentence segmentation enforced by Trados “has a constraining effect on their cognitive behavior and mental representation of the text and thus changes the translation task”, her student subjects are rather unaware of this constraint of TM technology.

Data are derived from an online questionnaire survey that was carried out in August 2004. The authors received 59 usable responses for measuring personal perceptions, Dillon/Fraser asked translators to respond to 24 statements that expressed various attitudes towards TM technology – such as, “I would lose out on work if I did not have TM software” or “The disadvantages of TM far outweigh the advantages”. According to their findings, the first two hypotheses may be true. Thus, newly qualified translators and translators with TM experience seem to be more positive towards TM technology than others.

The third hypothesis appears to be falsified, as translators with strong IT skills also appear to be more likely to have positive perceptions of TM technology. These findings lead the authors to suggest that a lack of understanding and knowledge of TM technology and its possibilities – rather than the nature and applications of TM technology itself – may be an important reason why some translators reject it altogether.

Lagoudaki

Lagoudaki (2006) reports on a survey of the adoption of TM technology and, like Dillon/Fraser (2006), she studies users’ attitudes towards TM technology.

By means of an online questionnaire, Lagoudaki obtained responses from 699 translation professionals (translators, terminologists, project managers, reviewers, subtitlers, etc) from 54 countries. Unlike in Fulford/Granell-Zafra’s (2005) survey, the adoption of TM technology appears to be considerable: the percentage of respondents using a TM system is 82.5%. In line with Fulford/ Granell-Zafra’s (2005) findings, Lagoudaki asserts that those who specialise in technical texts are more likely to use TM tools, followed by those who specialise in financial and marketing content.

Those who report a legal specialisation are also likely to use TM tools, but less so than the above-mentioned groups, which concurs with Christensen’s (2003) findings. In line with Christensen’s expectations, Lagoudaki finds a relationship between high levels of textual repetition and high adoption of TM technology. Unlike Dillon/Fraser (2006), Lagoudaki does not find any striking difference in TM adoption between different age groups and between those with and without work experience. However, like in Dillon/Fraser’s study, Lagoudaki finds that high IT proficiency is linked with high adoption of TM technology. When asked why they use a TM, most respondents answer that it saves time (86%), that it ensures consistency in terminology (83%), and that it improves quality (70%).

Other benefits are cost savings (34%) and the efficient exchange of resources such as glossaries and TM databases (31%). A rather surprising result is that, though they own a TM tool, some respondents (16%) have not been able to learn how to use it yet.

Mandreoli et al.

Mandreoli et al. (2006) evaluate the design of their own TM, namely the Example-based Translation Assistant (EXTRA). In comparison with most TM systems relying on artificial intelligence, the search engine of the EXTRA system is founded on advanced information retrieval techniques executing two processes: first, the document to be translated undergoes a syntactic analysis, and then it is compared with the TM data using so-called edit distance. This procedure is applied in order to ensure a good trade-off between the effectiveness of the results and the efficiency of the processes involved (2006: 169). The aim of the paper is to present, analyse and test the effectiveness and efficiency of the system . The investigation is based on theory and research on Example Based Machine Translation (EBMT). Data derive from statistical simulation experiments

which document that EXTRA is able to support effectively and efficiently the translation of texts in western languages (2006: 194). Based on a test run of the EXTRA system, the authors conclude that their “results show the goodness of EXTRA in retrieving useful suggestions and of the two processes constituting it, suggesting the suitability of their stand-alone employment also in contexts that are not strictly related to translation”.

We may not fully appreciate the technical intricacies of their investigation, but the authors are certainly right in asserting that CAT in general and TM in particular may currently “represent one of the most promising translation paradigms” .

4.8. O’Brien (2006) and (2008)

Investigating translators’ cognitive load in connection with various TM match types, O’Brien’s is

one of very few TM studies of translators’ interaction with TM tools. In her 2006 study, O’Brien

sets out to investigate whether eye-tracking in general is a useful research methodology for investigating

translators’ interaction with TM, and, in order to investigate this generic question,

whether eye-tracking makes it possible to identify differences in cognitive effort with different

TM match types. The study is an experimental pilot study involving four professional translators,

who translated a text using a TM (SDL Trados Translator’s Workbench) and then commented on

their translation process in retrospective verbalisations. The study investigates the cognitive effort

required from translators for different TM match types by analysing quantitative data from

eye-tracking (provided by Tobii 1750) supplemented with qualitative screen-capture data (provided

by Camtasia) and with subjects’ retrospective verbalisations about what they were doing

during the translation task (2006: 189). The cognitive effort is measured using processing speed

(words per second) and pupil dilation. The analysis is performed as a comparison of processing

speed for each match type and the percentage change in pupil dilation for each match type, supplemented

by the retrospective verbalisations. The fi ndings suggest a strong correlation between

percentage change in pupil dilation and processing speed for different match types. The study also

shows that exact matches exert the least cognitive load on translators, while no matches exert the

greatest load. Furthermore, though the relationship is not a linear one, it is demonstrated that the

cognitive load increases as fuzzy-match values decrease (2006: 199ff). As a general conclusion,

O’Brien suggests that her method of eye-tracking in combination with retrospective protocols is

well suited for translation process research (2006: 200).

Inspired by her 2006 fi ndings, O’Brien (2008) investigates in more detail why the relationship

between fuzzy-match values and cognitive effort is not as linear as one might expect. She carried

out an experimental study in which eight students translated a technical text using a TM (SDL

Trados Translator’s Workbench). The subjects’ eye movements and pupil dilations were recorded

using an eye tracker (Tobii 1750). Because only fi ve subjects’ eye movements could be accurately

tracked, the study was limited to these fi ve. Once the subjects had fi nished the translation

task, they were presented with a paper-based survey that included the same source segments and

fuzzy matches that they had just seen on the screen. Subjects were not shown the fuzzy-match

values that were assigned by the TM system, nor were any differences between the new source

text and the source text in the TM highlighted. Subjects were asked to rate their perceived editing

effort for each match using a fi ve-point scale. Based on processing speed alone, the study demonstrates

that decreasing fuzzy matches mean increasing effort, whereas the picture that emerges

if pupil dilation is used as a measure of cognitive load is less clear. When seen as a median measurement

across all subjects, dilations increase as match values decrease until the 60-69% match

class is reached. Below this match class, decreased pupil dilation is noted. According to O’Brien,

this might be due to the fact that subjects reached a baseline cognitive effort when they reached

the 60-69% match class.

4.9. Colominas (2008)

Like Dragsted (2004, 2006), Colominas (2008) assumes that sentence-based segmentation is unnatural

and should at least be supplemented with sub-sentential segmentation, as the repetition

of a whole sentence is generally rare. She therefore sets out to investigate the usefulness of subsentential

segmentation. With a view to evaluating both recall (target-text proposal) and precision

(usability), two experiments are carried out based on different TMs: a multilingual corpus

10

extracted from the proceedings of the European Parliament (the Europarl corpus) and an English-

Spanish corpus built up from United Nations documents (2008: 346). Segmentation below the

sentence level, especially noun-phrase (NP) segmentation, is employed and analysed. The experiments

are reported as showing that “sub-sentential segmentation clearly shows a signifi cant better

recall”, though NP segmentation “should be much more refi ned in order to be really useful”

(2008: 352). These fi ndings, which should be considered in future program improvements, are

logical and probable, but the experiments are discussed in fairly technical terms and concepts are

not suffi ciently defi ned and operationalised.

Conclusion

In conclusion, the spread and need of CAT tools in contemporary society is a reality which cannot be ignored. In older times, the traditional translator was satisfied with what he earned and was not bothered by the time consuming and exhausting method of work. But today, as we have observed through the analysis written in this paper, things are slowly but certainly changing.

Through this paper, I have tried to give the reader an insight into the world of translation, in general, but mainly into the future of translation. The reality is that the new technology is still developing, being updated and improved by the second, but its popularity will certainly reach high rates in the next years.

In order to prepare the younger generation for a technological enhanced society, I strongly believe that research should be conducted and encouraged in all parts of the world . An emphasis should be placed on motivating universities to insert CAT training in their curriculum with the purpose of preparing talented young people for the growing expectations of employers on the labour market or in order to be as successful as possible as freelancers.

In this paper, I have attempted to discover what the field of translation knows about the nature, applications and influences of TM technology, including translators’ interaction with it. Based on the literature on TM technology in general, we have sketched the history of the conception of TM tools as an aid for professional translators and we have provided an overview of some basic definitions and applications.

While TM systems may differ as far as applications, internal processes and interactive modes are concerned, all TM technology shares the basic function of deploying existing translation resources in a new translation project, and, though the translation process is controlled by a human translator, the assistance of the TM tool is always regarded as an integral part of the translation process.

Hoping to establish what has been documented so far by means of empirical TM studies, we analysed a collection of nine empirical TM studies. As current TM systems only became commercially available in the 1990s, we decided to concentrate on studies that were published in 2000 or later. Searching the TM literature at large, we came up with a modest list of nine empirical studies (some of them reported in two publications). This list may not contain all available empirical studies of TM, but we hope that it may be seen as representative of empirical TM research so far.

6.Bibliography

Works Cited:

Baron, Dumitra. "Les NTIC Dans Le Processus De La Traduction Spécialisée – Les Mémoires De Traduction SDL Trados." (2012): 1-3. 24 Mar. 2012. Web. 5 May 2016.

Peraldi, Sandrine. "Traduction Assistée Par Ordinateur : Entre Théorie Et Pratique." Les Cahiers Du GEPE No.2. N.p., 2010. Web. 2 June 2016.

Selected Bibliography:

Baron, Dumitra. "Les NTIC Dans Le Processus De La Traduction Spécialisée – Les Mémoires De Traduction SDL Trados." (2012): 1-3. 24 Mar. 2012. Web. 5 May 2016.

Peraldi, Sandrine. "Traduction Assistée Par Ordinateur : Entre Théorie Et Pratique." Les Cahiers Du GEPE No.2. N.p., 2010. Web. 2 June 2016.

7. Appendices

Figure 1.The Percentage of CAT users

Figure 2. The product used as a CAT tool

Figure 3. The number of text translated with a CAT tool

.

Figure 4. The productivity of CAT tool users

Figure 5. Monthly profits of CAT users