Finally, we should bear in mind that computer-assisted translation is a complex process consisting of a several stages. Managing large translation projects involves a number of phases and tasks which can be broken down into translation and non-translation tasks or pre-translation, translation and post-translation tasks. In such classifications, only the actual building of the TL equivalent of a SL text is regarded as a translation task, while all the remaining tasks, i.e. terminology management, desktop publishing (format conversions), text extraction3, proofreading and customer's review, are considered to be non-translation tasks.
3. Historical Background
The first efforts in terminology began in 1960's, probably as a result of the publication of the (in)famous ALPAC report which advocated developing software tools to aid translators instead of carrying out machine translation research (Feder 2001:15, Hutchins 1996, Palacz 2003:8). Terminology, however, had not been perceived as a discipline distinct from lexicology and other linguistic disciplines until the publication of the Einfuhrung in die allgemeine Terminologielehre und terminologische Lexikographie4 by Eugen Wuster in 1979 (Wuster 1979) (cf. Campenhoudt 2001).
The first terminology projects were only available for large organizations because the terminology management software required mainframe computers (Rico Perez 2001). This situation led to the development of large-scale termbanks, e.g. Termium, Eurodicautom, Banque de terminologie du Quebec (now Le Grande dictionnaire terminologique). The termbanks developed at this stage are still in use nowadays, although the systems underwent general overhauls, e.g. Eurodicautom now runs on entirely new platforms (Oracle and Fulcrum)5.
The 1980's saw the first electronic dictionaries and terminology management software developed for personal computers and available for individual translators, following the development of translation memory software. However, these tools had many limitations. First of all, they were not networkable i.e. it was impossible to share terminology collections over local area networks. The first generation of terminology management tools often offered only unidirectional searching e.g. EN-GR but not GR-EN. There were also restrictions on the number and type of data fields, as well as of storage capacity.
The new generation of terminology management tools followed the publication of the concepts of a three-level integrated translator's workstation (Melby 1992) (cf. Feder 2001).
The first release of MultiTerm for Windows and DOS Translator's Workbench package was in 1992. Another CAT tool which is now one of the market leaders - Deja` Vu, was first released in 1993. These tools offered more possibilities than the earlier generation and gained significance especially due to the fact that the developers e.g. Trados, recognized the opportunities lying in the growing popularity of local area networks (Brace&Joscelyne 1994).
Since that time, many new tools and new versions of the first CAT tools have been released, catching up with the developments in language engineering and computer technology. Currently, there are two main tendencies in CAT development. On the one hand, software developers tend to isolate the functionalities which used to be part of terminology modules into separate tools, e.g. term extraction module was part of MultiTerm 5.0 but is no longer in a package with Multiterm iX. On the other hand, there is the tendency for integration of the typical MAHT tools, including terminology management tools, with machine translation systems and localization tools (Melby&Wright 1999), resulting in the so-called hybrid systems (Feder 2001:32). The application of hybrid systems and highly integrated translation environment is usually most advanced in large institutions, e.g. European Commission (Blatt 1998, Hutchins 1989).
3If a project manager does not receive a source file in a text format, e.g. the source file is in a protected PDF format, the text that is to be translated will often be extracted from the file and delivered to the translator in a text format, in order to enable the translator to use CAT software. If the source text is delivered in hard copy, text extraction will refer to using optical character recognition (OCR) in order to receive the source text in the electronic form.
4General Theory of Terminology and Terminological Lexicography - An Introduction