Les dictionnaires dans les systèmes de traduction automatique.

Machine translation (MT) is a relatively new field combining linguistics and computer science. From a linguistic perspective, one of the most important components of MT systems is undoubtedly dictionaries. Until now, few studies have been carried out on the role played by dictionaries in such systems. The objective of this thesis is therefore to advance research in machine lexicography, or more precisely, in bilingual machine lexicography. This thesis examines MT dictionaries in a certain number of systems in order to gain a better understanding of the role played by MT bilingual dictionaries in relation to other parts of the system. This is important since, unlike human translators who can choose to consult bilingual dictionaries whenever they wish, according to their own particular needs, MT systems dictate when consultation takes place and the type of information sought. Furthermore, the type of MT system (direct, transfer or interlingual) has a direct impact on the content and use of bilingual dictionaries. Bilingual dictionaries in direct systems, which do not contain unilingual dictionaries, usually contain more information than those in transfer or interlingual systems. However, the linguistic analysis of bilingual dictionaries in direct systems is less sophisticated than that of unilingual dictionaries in transfer or interlingual systems. This thesis also compares traditional bilingual dictionaries with MT bilingual dictionaries in order to determine to what extent it is possible to adapt a traditional bilingual dictionary entry for use in MT systems