How useful are CAT tools?

Abstract. It has been argued that the use of CAT software impacts translation work in ways that are at least three-fold: improved quantity, improved quality, and improved control over the whole translation process. Consequently, unrestricted adoption of CAT tools seemed to be the logical step to be taken by every professional in the field. However, such unanimous adoption of CAT tools has not yet become a reality. In spite of all its usefulness, criticism of CAT abounds. This paper aims at discussing the usefulness of CAT from a more balanced perspective. It presents and discusses the results of an informal investigation of the use translators are making of CAT utilities, based on the answers by members of an online group for professional translators to the question: How useful are CAT tools?

1. Introduction
In spite of the massive amount of time and money invested in the development of more efficient Machine Translation (MT) platforms in the last decades, it seems clear that a semantically consistent, discourse-aware, error-proof non-human translation system is still a goal to be pursued for many years ahead (Arnold et al, p. 01). However, research in this area has led to significant advancements in a correlated field, namely that of Computer-Assisted Translation (CAT)1, resulting in a large number of CAT software being released in the market. Such software, broadly grouped under the label of “CAT tools”, offer the translator an array of resources aimed at increasing translation quality and the translator’s productivity (Barrachina et al, p. 04; Hutchins, p. 11). The “CAT tool” label is in fact an umbrella term covering a very varied group of computer-based systems which seek to address translators’ most relevant necessities in their daily struggle to transfer concepts and ideas from a language to another. CAT tools, as their name suggest, are not intended at replacing the translator, taking over all the subtasks involved in the translation process (which is not yet possible, for reasons not to be discussed in this short article). Their purpose is more modest, and therefore more feasible: they focus on tackling some of these subtasks, thus reducing the cognitive load on the translator and boosting their translation output, as well as improving the quality of the final product (Craciunescu et al, p. 08; Lab, p.166). The number of tools in a CAT platform varies from package to package, but even the lowest priced systems usually include built-in spell checkers, grammar checkers, and facilitated search to electronic or online dictionaries, glossaries and, sometimes, linguistic corpora. More complex platforms – and that complexity usually results in much higher prices - offer a richer set of modules including terminology management tools, access to terminology databases, indexers (text search utilities), concordancers, text aligners, and perhaps the most valued of all CAT tools, translation
1 AKA Computer-Aided Translation, Machine-Aided Translation, Machine-Assisted Translation.

for instance. which the software also builds and manages automatically (Craciunescu et al. 09). Additionally. which certainly provides for faster completion of jobs and the possibility of new jobs being accepted by the translator. TM utilities suggest the right word to be used in every situation. which will certainly have an impact of the quality of the translated text. judging from the number of translation agencies working fully or mostly in CAT-based fashion. A final pro-CAT argument concerning quality touches a very sensitive area for many overworked professionals: a large part of the boring. provided that word has been encountered before and added to the translator’s terminological database. the introduction of task management modules into some of the most complex CAT platforms is said to result in better control over the whole translation process. p. and the market seems to have accepted them enthusiastically. allegedly allows the translation professionals to take in more tasks and have them accomplished faster. the case for such help as that provided by CAT tools is easily made. p. 04). aimed at helping translators deal with a several tasks at the same time or at assisting crew leaders at translation agencies/offices in the management of complex jobs involving various translation professionals. these quantitative benefits would not be so impressive if such increased turnout did not prove to be equally superior in terms of quality. However. thus contributing to an increase in the amount of time the translator spends effectively translating. from the point of view of the cognitive operations involved and the several specific abilities translators must develop to do their job properly (Craciunescu et al. Given the high level of complexity of the translation process. p. 08). . improved quality of the tasks completed. There are packages that go as far as offering users a number of project management tools. The use of CAT software. With time. mechanical typing is done by the software automatically. They are currently a well-established subject in many serious translator training programs all over the globe. consequently improving their productivity and their income. time-consuming. especially translation memories (TM). the translator can invest the time saved as a result of the adoption of CAT utilities to concentrate on further reviewing the final work. that is. checking work progress. It has been argued that the adoption of CAT software impacts translation work in ways that are at least three-fold: improved quantity of work done. The resulting boost in professional satisfaction is argued to have made a difference in the life of many translators. and improved control over the whole translation process. The use of CAT software is also presented as conducive to better final texts. Free-lance professionals as well as resident agency translators and their managers can benefit greatly from the automatization of such non-technical steps of the job such as assigning parts of a task to different translators. Finally. the non-stop growing of personal databases and the facilitated search to other databases online results in a substantial part of each task being pre-translated even before the translator starts working. given the higher terminological consistency resulting from the employment of TM tools.memories (Craciunescu et al. which starts with the reception on the originals and ends with the delivery of the translated text within the deadline agreed with the client (and the payment for the work done). performing at a more sophisticated level of text elaboration.

or have not considered it worth the effort to even start using such tools. The Professional Translators and Interpreters – ProZ.linkedin. and indicate that a more balanced evaluation of these computer-based utilities is in order. eventually. the arguments in favor of the use of CAT are undeniably strong enough to account for its widespread adoption by both freelancers and salaried translators worldwide.287 members. the adoption of CAT technologies by every translator and translation office has come to seem inevitable. Until the moment this article was finished.com/groups/Professional-Translators-Interpreters-ProZcom-138763 3 URL: http://www. The arguments presented by the respondents cast a lucid. The forum thread from which we collected the data presented here 2 URL: http://www. a professional online network.com)2 hosted by Linked In3. given its limited scope and the methods adopted for the collection of the data presented.com forum on Linked In. it was the logical next step to be taken by those wishing to survive in the market in the aftermath on the definitive CAT revolution. in Brazil and in other countries as well. were solely to give translation professionals and students some grounds on which to base their reflections on the alleged universal applicability and usefulness of the group of computer-based translation utilities referred to as CAT tools. As a consequence. It is therefore a very simple. far from bringing up an exhaustive discussion of the topic. which started on July 10. a well-known professional online network. therefore. used and given up MT and other related software either partially or completely. It was just a matter of giving the market the necessary time to adapt the new work settings brought up by this new set of digital translation tools. In spite of all its usefulness. criticism of CAT software abounds. years have passed and such unanimous adoption of CAT tools has not yet become a reality.linkedin. This paper presents the results of an informal investigation of the use translators are making of CAT utilities. However. and in recent years many clients have turned the file formats yielded by the most famous CAT tools obligatory for the delivery and acceptance of translated texts. hosted by Linked In. based on the answers 45 members of an online group for professional translators have given when asked the question that names this short article: How useful are CAT tools? The question was posted as a thread on an international forum for translators (Professional Translators and Interpreters – ProZ. critic light on CAT advantages and weak points. Some agencies have taken that use to the point of making it a requirement for anyone wishing to become part of its translation staff. Methods This study was based on the responses 45 (forty-five) members of an online group for professional translators provided to the question “How useful are CAT tools?”. unpretentious investigation. 2008. 2.com . Our aims.watching out for approaching deadlines and keeping records of payments received and due. In a quick analysis. had 30.com). posted on an international forum for translators (Professional Translators and Interpreters – ProZ. and translation professionals all over the world have tested.

Comments varied greatly in length. as follows: • • • • • Group 1 – 18 participants Group 2 – 2 participants Group 3 – 4 participants Group 4 – 9 participants Group 5 – 12 participants . having tried them and found them not worth the trouble. we divided the positions taken by the 45 participants of the forum thread considered for this study into the 5 groups above. since the thread is still on and continuously collecting more comments). Discussion As a prominent member of the forum thread (in term on number of comments posted 11 out of a total 149 responses) pointed out. according to the tasks at hand. Although the amount and nature of the data collected for this paper does not allow for proper statistical treatment. and those who have not directly contributed to the discussion. forty-five different members of the forum contributed to the thread. the participants’ positions concerning the usefulness of CAT tools seemed to fall neatly into three categories: (1) those who use CAT software for all or most translation jobs they receive and see more advantages than disadvantages to their use. namely (4) those who believe that CAT tools are useful in some situations but not so much in others. (2) those who will not use CAT in any situation. and (3) those who do use CAT when requested to do so (by agencies or clients). since their arguments are derived from other sources of information than their own experience. The content of the messages referred to different aspects of the topic in order. until the moment this paper was finished. but would rather not use them. including initial comments on the subject and responses to the comments by the other participants. or who are trying to use them as a way to adapt to the new job market and the new technologies. and the main points highlighted by the thread participants are discussed in the following section. according to each participant’s opinion about the usefulness of CAT tools. and (5) those who have never tried any CAT tools and therefore can not really contribute for the discussion. one single comment contained responses to several other comments. The comments were grouped according to the main aspect addressed. and therefore will alternate their use with more traditional translation methods. and has received 149 comments by members from all over the world (again. from one line to several paragraphs of text. Analyzing all the comments. 2013. 3. however. we consider that three other groups could be added to the that initial categorization. on March 21. but are not yet fully convinced CAT tools are really helpful. which greatly enriched the discussion. posting comments which did not address the question of the usefulness of CAT software. During that short period of time. Quite often.was started a month before this paper was written.

The use of translation memories (TM) means reduced need to search for terminological solutions on previous translations. at least for themselves. due to lack of relevant contribution to the discussion at hand. the second most important group in term of number of members is Group 4. in an increase in the number of jobs a translator can take. These participants of the forum thread report to have tested one or several CAT packages available on the market and detected considerable benefits in their use. however. based on the belief that the use of such tools reduces the difficulty and the time spent on each translation job (which is not always true). their reflections led them to evaluate the level of usefulness of computerbased translation tools in terms of the specific features of different translation. Some of the arguments pro-CAT mentioned in the introduction to this paper. In terms of quantity. Instead of taking grounds definitively for or totally against CAT use. First. As a consequence. therefore presenting in their comments some of the most balanced and relevant views of the topic. for several reasons. translators may find themselves reviewing and doublechecking the final work to get a grasp of the entirety of the written piece. The database itself is managed and updated automatically. display strongly favorable views of unlimited CAT tools use. if we disconsider the responses by the members of Group 5. satisfied users claim that the adoption of CAT has permitted them to do their work faster. which are the backbone of pro-CAT campaigns. On the other hand. since the software automatically consults the database and provides the appropriate term for each situation. so new terms entered by the translation in a new job are retrieved and stored inside the database for future use. “… there is no doubt that these tools can save a lot of time for the translator. which results in no real time saving in the end. By contrast. due to the amount of pre-translation performed by the software. some professionals are not sure if the time they are saving by using CAT software really results in income gains. provided that the database contains that information. some of which we have already presented in the Introduction to this paper. therefore. have been reinforced by the experiences reported by the forum thread participants. And time is money…”. however.The majority of the responses. Still on the “time saving” aspect of CAT adoption. As a participant stated. participants have made comments which are not directly related to the tools themselves but are worth discussing all the same. alongside with significant pitfalls. The least numerous groups are constituted by the reluctant users of CAT (Group 3) and by the detractors of such use (Group 2). for they are derived from problems found during real use of CAT tools in their daily translation practice. as a consequence. given the frequency such matters have been presented alongside the forum thread. the alignment of blocks of original text and translated texts presented by most CAT tools is felt to be confusing by some. that dealing with their multiple functionalities in fact slowed them down as compared to the “old. whose arguments are nevertheless worth considering. and some feel it prevents them from seeing the text as a whole. of course. which directly affects the professional’s income. Additionally. especially the quantity and quality gains these utilities can bring to the translator. That results in ever-growing translation speed and. traditional way”. some translators have reported finding out. ask for discounts from . after some time trying out different software. since clients/agencies.

On that matter. which we address in a later moment in this paper. since they may come to rely on the software original/target text matching to a point where they may overlook minor mistakes in the results presented by the tool.translators who work with the help of CAT. which are undisputedly less powerful than the most basic computer. though) and most participants reported on having difficulties to learn how to use the software. This is the strongest argument in favor of a more balanced use of CAT put forward by member of Group 4 – absolute consistency is not always desirable. and that has an impact on their financial gains. but until that day comes one solution is getting some training and then assess their usefulness based on that knowledge”. a participant has reported. translators are faced with the need to invest significant amounts of “unpaid time”. A participant reported having tried CAT tools and finding them 4 Although it refers more to the quality of the translated text. There [shouldn’t it be THEY?] are usually full of excruciatingly bad translations that I am ashamed to have associated with my name…”. 5 The second most important argument in favor of the adoption of computer-aided translation is the improvement in the quality of the texts translated with the help of such software packages. Another discussion is how easy or not they are to learn”. the gains in consistency are not universally relevant. .4 The most complex platforms are expensive for the average freelancer (although there are demonstration packs available. TMs make it possible for a translator to use always the same translated term for the same term in the original. however. or require the use of CAT so as to be able to reduce the amounts paid for the translations. The importance of this perception of the difficulties involved in learning a CAT tool and the amount of time that must be invested in training to use them can not be disconsidered. 5 He continues: “[…] Personally I still think that most CAT tools should be more user-friendly than they are. in some cases. On the other hand. since it may take long for the translation [shouldn’t it be “translators”?] to recover that investment. as a participant very appropriately pointed out. depending on the volume of work they usually get. the pro-CAT participants of the forum thread confirmed what the theory has affirmed. The initial investment in the purchase of a CAT package has also been mentioned as problematic. as participants in Group 2 and 4 have insisted to point out. given the different translation areas. lead translators to making mistakes. due to their user-unfriendly interfaces. sparing the professional the effort of searching dictionaries/glossaries several times or having to trust their own memories. as a participant put it. especially as far as the consistency element is concerned. Sometimes they work for more than an agency. “One discussion is how useful CAT tools are. Secondly. as well as having to use TMs supplied by agencies alongside with the originals. For some. and are requested to fully master the software adopted by each different employer. The very fact that the term translations offered by TMs result from a search on a database of previously translated jobs may keep the translator from realizing that the context of the new translation at hand asks for a different term. to “… hate having to use client TMs. trying out different CAT tools in order to find the one that best suits their needs and/or budgets and learning how to make effective use of the different modules offered by CAT packages. we are not develop this topic further since. with limited resources. on that matter. with results in the translator losing their independence towards their own work. CAT tools may.

This is true. First and more obviously. Although there is a clear movement towards inter-intelligibility.extremely useful for IT texts. there are document format considerations to be made. and since some freelance-affordable CAT tools operate over web servers. like quil [my guess is that it’s QUILL. However. another participant states his point of view as follows: “Translating works of fiction needs proper setting. that many times it is the agency. for example. and the consequent back and forth convertions [pretty sure the word is “conversions”] add an extra load of work on the translator. Some CAT platforms search external TMs and online TMs (sometimes sent by agencies alongside with originals or made available on the agency’s portal. but with real and serious impact on the spread on computer-based translation – apart the cost element. there is some grounds for the feeling. their use is not a choice for those translators operating on slow. or “anywhere where consistent terminology is used: legal. since TMs are updated according to the volume of translation work done. who really benefits the most from the adoption of CAT – the number of works managed by an agency surely justifies the investment. expressed by some participants. Finally. . Again. Originals are often sent in in formats that a certain tool does not deal with. Participants of the forum thread have reported not being able to benefit from CAT simply because they could not find a package that deals with the languages they work with. not the translator. and can lead professionals to giving up CAT whatsoever. as well as non-stop Internet connection availability. for instance. according to the comments gathered for this paper. medical. as mentioned before. no rarely the professional does not have the choice. or any other straightforward narrative or description”. unsteady connections. with double LL xD] and scent of paper. or could not afford the most complex platforms that do have resources on those languages. as a participant has put it. I then use also good old dictionary in print. Aware of 6 On that matter. 6 It is also important to keep in mind that. some tools still yield outputs in formats that can only be processed within that specific tool. CAT tools are commonly referred to as if they came prepared to tackle any given language pair or pairs – which is not true at all. expensive machines. which are important national and international languages. Since agencies often impose on the use of TMs available online on their websites. can be a hindrance to CAT use in poorer areas – the most powerful packages require equally powerful. All the other works are best done with trusty CAT”. as mentioned before). an individual translator has to take a large number of jobs or concentrate on only one subject field in order to be able to build a TM which is large enough to be as helpful in terms of consistency as the time/money invested on the CAT tool would suggest. The type of original can be an important element to influence a translator’s choice to use or not to use a CAT tool – provided that there is really a choice based only on textual features or personal preferences. That can be a serious problem for translators working for several different agencies. Some limitations to the use of CAT are rather prosaic. for German and Portuguese. but not so much for literature. due to some conditions the use of CAT require and that are not always within a translator’s reach. already discussed at length before. but the translator may not agree with the terminological choices resulting from search on outer sources. feeling that the resulting translated text looks like “a Frankenstein monster”. Hardware requirement.

F. CAT software developers have invested in equipping their packages with several conversion and text formatting tools. but I’m not sure. Volume 8. Again. (2004). Epistemologie. the more complete they will be. Lab. but the analysis of the results has indicated that the use of CAT is not yet consensus inside the area – and sometimes not even feasible for all members of the professional translation field. 2nd edition. “Machine Translation and Computer-Assited Translation: A New Way of Translating?” In: Translators and Computers. Machine Translation: An Introductory Guide.. S. et al. D. 03. Volume 22. “Statistical Approaches to ComputerAssisted Translation”. another word missing. 01. Quoting one of the participants of the present study. 4. 01. Gerdin-Salas. although it is not yet available for all CAT software in the market. L. Fréderique (1988).. et al (2008). Blackwell Publishers. Conclusion Given the modest. the more expensive. we have tried to build on the comments made by the participants of the forum thread we used as data [I think there’s something missing here. (2004). Craciunesco. however.this fact. In: Histoire. if they are [I have a feeling that there’s a word missing here. S.. In: Bulletin de l’EPI. it doesn’t say anything about it]. Stringer-O’Keeffe. Barrachina. “Human intelligence. References Arnold. Volume 35. No. it is important to keep in mind that their purpose is not to replace the translator. London. and this is making the translation quite hard – my guess is “willing”] to make their software more user-friendly and more widely applicable and usable. “La Traduction Automatique”. which is a much welcome improvement of their systems. “Machine Translation Over Fifty Years” (2001). and the next sentence is different. (2001). No. this time I think it is “do”] his job more efficiently. However powerful these tools may become in the future. O. there is still a long road ahead for CAT developers. pages 1-11. a whole idea. Language. pages 165170. professionalism and experience combined with technology certainly make for success”. . Our sample was limited and therefore statistically irrelevant. No. Although the majority of the opinions gathered point clearly in a proCAT direction. unpretentious scope of the present study. like. but to help him to [again. In: Computational Linguistics. Balkan. I couldn’t translate this sentence properly – I couldn’t answer this question that I’ve made: “what did we try to build on the comments?”. W. Hutchins. Casacuberta. pages 328. C. and use the information thereby provided as the background for a brief discussion of the usefulness of CAT tools as felt by their primary target users – professional translators. pages 7-31.