ABSTRACT
Transmission of a natural-language text into an equivalent text of another natural language with the aid of a computer program. Such programs have (with varying specializations and success) lexical, grammatical, and, in part, encyclopedic knowledge bases. Machine-aided translation consists of three components: (a) analysis of the source language by means of parsing; (b) transfer: the transmission of information from the source language into the target language; (c) synthesis: the generation of the target language. The systems vary according to whether they translate directly from one language into another or whether the text in the source language must first be translated into a neutral interlingua and then into the target language, which makes particular sense if the source language is to be translated into several target languages. Linguistic problems associated with machine-aided translation arise principally from the different lexical structure of the given vocabularies (e.g. Ger. kennen, können, wissen for Eng. know), from various grades of grammatical differentiation (e.g. aspect differentiation in Slavic as compared to Germanic languages), from need for encyclopedic knowledge to disambiguate ambiguous forms, and from the necessity to rely on experience and standard assumptions when interpreting vagueness. Programs for machine-aided translation have been applied with fair success; in most cases, however, it is necessary to pre-and post-edit the texts.
