An Organisational Checklist
In 1996, the translation services of six Institutions and bodies of the European Union1 published a joint call for tenders covering the acquisition of a software package providing Integrated Translation Support Tools2. According to the specifications, this package had to include as its main component a piece of Translation Memory Software (TMS)
integrated with one or more word processing programs, and -as optional but highly valued components- a text alignment utility and a terminology management program.
The award procedure was completed during 1996, and a contract was signed with the successful tenderer in 1997. The specifications of this call for tenders included a long
questionnaire, the aim of which was to assess the fulfilment of the technical and functional requirements which the European Institutions had established. Given the state of the art back in February 1996 and, above all, the fact that the participating Institutions were operating on different technological platforms and within different organisational
frameworks, these requirements were in fact quite open, and the questionnaire was directed more at gathering comparable data about the existing products than at excluding them outright on the basis of any matters of principle.
For this reason, the questionnaire annexed to the specifications3 can in a sense be regarded as a technical and functional checklist which any Translation Organisation (TO) could use as a guide when considering the use of TMS. The functional part, in particular, remains almost entirely valid today; as far as the technical part is concerned,
some adaptations would of course be necessary (operating systems, word processor versions, required hardware, etc). The purpose of this article is not to update that information, but rather to provide a different kind of checklist, namely an organisational
one, which any TO considering the large-scale introduction of TMS might find worth browsing. Even if it is largely based on the experience of the Translation Service of the European Commission, it is intended to be as independent as possible of any specific assumptions, technological or otherwise. On the other hand, although a number of contributions in this volume tackle the integration of TMS with other linguistic tools from both the technical and the functional point of view, the organisational implications of such integration are left aside in this article. However, it is clear that many of the items mentioned in this checklist would require considerable modifications if, on top of the sentence-level resources, the TO in question had eg significant terminological resources or a machine translation system available.
1 Prior considerations
1 Why does my TO really want to use TMS?
1.1 Repetitiveness. Are the texts sufficiently repetitive? Is this repetition purely document-internal or across families of documents? How many such families or documents can be established? Is my TO sure that repetition does not take place mainly below sentence level (eg phrases), which would reduce the interest of a TMS-based
1.2 Consistency. Is consistency a real requirement for our customers beyond the terminological level? If not, is it a requirement for the TO itself? To what extent can consistency constrain creativity in the texts which my TO translates?
1.3 Cost-effectiveness. Has my TO evaluated all of the costs of the implementation of a TMS system, including training, support, etc? On the other hand, has it also evaluated its indirect and/or long-term benefits in terms of flexibility, productivity, house style, resource sharing, etc?
2 Has my TO considered any viable alternatives to TMS for all or a part of the documents identified as suitable? (Some examples: machine translation -raw or edited-, version-tracking software, definition of models in close contact with the customer, use of a restricted syntax by the customer combined with some kind of automatic processing, etc)
2 Building translation memories from existing data
1 Does my TO want to begin using TMS from scratch or can it take advantage of existing data?
2 If suitable data exist, is my TO sure that they are usable with a reasonable degree of effort? In particular:
2.1 Are both the original and translations available in electronic form?
2.1.1 If not, is scanning of at least some key documents a reasonable alternative?
2.1.2 If yes, does my TO have a convenient archiving system allowing for a quick and efficient retrieval of all the necessary documents? Are both originals and translations available, or can both be made available, in the same format? Even if they are or can be made available in the same format, is formatting relatively parallel across language versions?
2.2 If any further conversions are necessary so that the aligner can handle the documents, has my TO evaluated to what extent this process will deteriorate the results? Is this acceptable?
2.3 If the amount of data to be aligned is significant, can my TO accept the results of a raw or lightly edited batch alignment (possibly indicating the automatic origin in each record of the resulting translation memory), or is perfect quality of the alignment an
2.4 Can my TO really afford to take care of this time-consuming task, or would it be better to subcontract it?
3 If suitable data do not exist or are not worth converting, is my TO sure that suitable translation memories will result from interactive work exclusively, within a reasonable time framework? Will translators accept working with TMS without any significant benefits at the beginning?
3 Working with TMS
1 Once a new document has arrived for translation, who will determine, and according to what criteria, whether it is suitable for translation with TMS?
2 f the document is suitable for translation with TMS, how will my TO organise the preparation of actual interactive work? This task might include:
2.1 deciding which one is the most suitable translation memory if there are several of them, and/or preparing a tailored one, either from the various existing translation memories or from a central source;
2.2 checking whether this translation memory already contains all the necessary material (older versions, related dossiers, etc);
2.3 scanning the document for references pointing to extra documents worth aligning;
2.4 working on the formatting of the original in order to maximise gains arising from the use of TMS.
3 Will my TO want to consider batch processing of the original against a translation memory as an alternative and/or a complement to subsequent interactive work? If yes, will this be left as a choice for the individual translator or organised in some way?
3.2 Interactive work
1 Will all translators use TMS? Will this be left as a choice for the individual translator or imposed in some way?
2 Will the translators in my TO regard interaction with TMS, which usually involves translating sentence by sentence, as an acceptable requirement in order to ensure consistency and to avoid reinventing the wheel, or rather as an impediment distracting them from their intellectual activity?
3 Has my TO evaluated the ergonomic consequences of a working environment which includes TMS as an essential component? In other words, is the TO confident that its translators can cope with a significantly greater amount of on-screen information flow?
4 If provisions for preparation of dossiers and/or post-processing are adopted so that these tasks can be carried out by staff other than translators, how should the existing workflow be adapted to ensure that interactive work continues to receive enough time and resources?
5 Where length and/or urgency require that a document be shared among several translators, how will my TO organise data-sharing during interactive work?
3.3 Integrating revision into the TMS workflow
1 Will revisers be working with TMS themselves, with the word processor or by other means? Will this be left as a choice for the individual reviser or imposed in some way?
2 If the revisers work with TMS themselves, will they be updating the same translation memory used by the translator, or will they be using a copy thereof?
3 If the revisers do not work with TMS at all, or not with the same translation memory used by the translator, will the TO want to organise an update of the translation memory after revision of the initial translation anyway? How?
4 If the revisers work with TMS themselves or if the TO makes provision for a subsequent update of the initial translation memory after revision:
4.1 Who will take responsibility for the final version of the translation memory, as opposed to the document?
4.2 Will the TO want a single solution for each original segment, or will it want to keep track of any proposed alternatives, either by the reviser or by the translator?
5 What particular measures should be taken for revision of a document which has been translated by several translators?
1 Several situations might require post-processing of dossiers, especially when it is known that similar documents will re-occur in the future. How does my TO intend to cope with them? Some examples, apart from the ones already mentioned (batch processing against a translation memory, revision without TMS, etc):
1.1 A document has not been translated with TMS because it was not found to be sufficiently repetitive. However, it now appears that it was actually the first instance of a recurring class of documents.
1.2 A document has been translated with TMS into some languages but not into others.
1.3 Part, but not all, of a document has been translated with TMS into a given language.
2 Some of these post-processing jobs could be carried out off-line, ie after the translation has been delivered, but in that case they could easily be put off. Does the workflow at my TO require adaptations so that these tasks get an adequate priority?
3.5 Data sharing and validation
1 Has my TO assessed what kind of balance should be defined between these two desirable objectives for a TMS system? A maximum of data sharing would involve fewer management tasks but could jeopardise confidence in the data; a maximum of 7 validations, on the other hand, would ensure a high degree of data quality but could require heavy management.
2 Once this balance has been defined, has my TO estimated the resources necessary for the new management requirements? How will the workflow at the TO need to be adapted to these new tasks?
3 A usual way to enable data management and, at the same time, to increase user confidence in the data is to assign labels to each piece of information. In the case of TMS, such labels could include the translator’s name, the date, the document name or reference number, a validation code, a domain indicator, etc. Does my TO already have a classification system adaptable to this purpose? Who will create or adapt it? Is the TO confident that such classification is stable enough, but also sufficiently flexible?
4 Distribution of tasks
1 TMS allows saving time during translation, eg by recycling existing translations. However, it also creates new tasks, especially before and after the translation proper (see Preparation and Postprocessing above). Who will take care of these new tasks at my TO? Could it consider assigning some or all of these tasks to nontranslating staff? Would the creation of this “translation assistant” profile fit into the new workflow? Can existing resources be allocated to this profile?
2 If my TO works with several target languages and documents are usually translated to more than one of them, additional gains can arise from centralising some of the tasks described above. What adaptations does my TO require in order to take as much
advantage as possible of this potential?
3 If my TO subcontracts part of its workload to freelance translators, will it use TMS only in-house or will it consider giving out linguistic resources to its freelance translators as well? In the latter case:
3.1 According to what criteria will TMS-suitable texts be assigned to freelance translators or kept in-house?
3.2 How will the distribution of in-house linguistic resources to freelance translators be organised?
3.3 Should any special legal measures be considered (eg concerning the protection of the copyright of linguistic resources)? Also, in requiring the use of a particular TMS system, should my TO participate in the expenses incurred by its freelance translators?
3.4 What adaptations should my TO introduce into its pricing scheme when providing in-house linguistic resources which facilitate a translation job subcontracted to a freelance translator?
4 What kind of impact will the use of TMS have on training and support staff at my TO?
4.1 Will a generic initial training be enough? Or does the complexity of my TO justify a significant investment in in-house training? In that case, should the training be carried out by translators or by computer experts? And how will the training of the trainers be
4.2 Is my TO ready for taking on first-level user support concerning TMS? Should this task be carried out by translators or by computer experts? Should it include extra assistance for the constitution of linguistic data?
SANTIAGO DEL PINO