Français Deutsch


Computer translation :
the status today

Contact us
Our services

Translation services


Quality assurance




Terminology services

Advice to clients

Quotation request

Consulting Services


Best Translation Tools

About languages and translation

Glossary-Keywords within translation

Computer translation

Translation free of charge

L'Europe c'est Babel


Swiss terms and acronyms




Déjà Vu Training




Work with us


Friendly links
Friendly links


It is important, first of all, to make a distinction between machine translation (MT) and computer-assisted translation (CAT).

Often a source of confusion for the general public, where the term "computer translation" is used to describe both types of technology, the two are in fact based on very different approaches. The respective results they are intended to achieve are also extremely different, since they use two quite separate, specific contexts.

Machine translation (MT)

On a schematic level, this method of translating involves the calculation speed of a computer in order to analyse the structure of each term or phrase within the text to be translated (source text). It then breaks this structure down into elements that can be easily translated, and recomposes a term of the same structure in the target language. In doing so, the method calls upon the use of highly voluminous, multi-lingual dictionaries plus sections of text that have already been translated.
Great hopes were placed in MT during the eighties, and huge investments were made in research into the subject. However, this was then to a large extent neglected in favour of CAT systems that were technically more realistic. Thanks to the development of the Internet over recent years, there is now a move towards both sites, and automatic translation systems, that make it possible to "get an idea" of the content of pages written in another language.
Apart from this function of providing a rough idea of what a given text contains and its utility when deciding whether not this would be worth translating, MT is only efficient where applied to texts with an appropriate degree of standardisation and coherency. In short, a text that can be translated by a computer must be written in a way that the computer can understand: there must be no ambiguity, and it must contain only terms contained in the computer's dictionary and which always have the same meaning.
This type of controlled language - which imposes major constraints on writers - has few areas of use beyond that of particular types of technical documentation that are sufficiently voluminous to justify the investment.
The best-known, and perhaps the most efficient of all MT systems is used in Canada for translating weather forecasts from English into French and vice-versa. These are created from a highly limited, self-contained unit of standard phrases.
Otherwise, MT is mainly used for a pre-translation phase that must be followed by intensive revision process in order to make the target text suitable for publication.
The extremely high costs of setting up and managing parameters for MT software and its dictionaries are a major obstacle when automatic translation.

Computer assisted translation (CAT)

Computer assisted translation is made up of a series of tools aimed at providing assistance for the translator with regards to both coherence (consistency) of his work and speed. The largest of these tools manage the specific terminology within the area of work as well as the translation memories.

Terminology management, above all, has the computer scan each word of the source text in order to locate them in the specialised dictionaries and, whenever possible, offers an equivalent to the translator, automatically and in the target language. The efficiency of this function is therefore basically determined by the quality and the volume of the specialised dictionary. The constitution of translation memories requires the creation of equivalency tables between the source text and the target text. In order to do so, the software breaks the text to be translated into segments. At the point where the translator validates the corresponding target text, the software memorises the source segment and the target segment as being linguistic equivalents. If the source segment then appears in the text again (repetitions may be frequent in technical texts), the software automatically proposes the memorised translation. When updating the source version of a text that has already been translated, the software automatically takes the parts already translated and alerts the translator in the case of any new or modified elements. The most sophisticated types of CAT software recognise segments that are approximately identical, and alert the translator by marking the elements that differ from the memorised segment.

Beyond managing terminology and translation memories, CAT tools offer various methods of managing translation projects and statistical analysis functions in order to determine - for example - the repetition rate within a text or carry out searches for terms in their various contexts, etc.

When is CAT a viable solution?

If the utility of CAT is quite evident in principle, we should not lose sight of the fact that its efficiency varies according to the type of text to be translated. Its implementation may demand major investments, and its profitability is far from immediate since the dictionaries and memories only increase in scope progressively. On the other hand, the use of translation memories does offer a solution to problems of deadlines when the text to be translated has undergone last-minute modifications; usually, the client waits until the text has been completed before sending it to be translated. And if the publication dates have already been fixed, and the last proofreader is running late, the translation deadline becomes so short that the work must be shared among several translators - meaning there will not only be a risk of inconsistency but that quality may be lost in other areas. In such cases, CAT makes it possible to translate a first draft and, when the final version of the source text is available, to bring up all the non-modified parts automatically before manually handling the modifications shown by the CAT software. Deadlines and quality thus cease to constitute a contradiction in terms. In order for CAT to be useful, the source text must possess certain qualities:

  • Terminological consistency: the same term is always used in the same sense, and the same object or action is always described using the same term.
  • Phraseological consistency: the same idea and the same action are always described in an absolutely identical manner. This is the prerequisite for increasing the number of repetitions and for the translation memories to become efficient.
  • Short, simple phrases: these also increase the probability of repetition and reduce cases of ambiguity.

The use of CAT is more justifiable for texts with a long life span and that may undergo several updates.


Whatever your project, contact the professional in the area of translation without delay, who will steer you towards the best approach for your particular needs.
© by FXM office[at]