The use of the computer in translating natural languages ranges from that of a translator's aid for word processing and dictionary look-up to that of a full-fledged translator on its own. However the obstacles to translating by means of the computer are primarily linguistic. To overcome them it is necessary to resolve the ambiguities that pervade a natural language when words and sentences are viewed in isolation. The problem then is to formalize, in the computer, these aspects of natural language understanding. In this work the authors show how, from a linguistic point of view, one may form some idea of what goes on inside a system's black box, given only the input (original text) and the raw output (translated text before post-editing). Many examples of English/French translation are used to illustrate the principles involved.

Скачать книгу бесплатно (djvu, 2.13 Mb)

---

Читать «Machine Translations. Linguistic characteristics of MT systems»