Bilingual reference corpora for translators and translation studies

A recent trend in corpus linguistics is the construction of bilingual (or multilingual) reference corpora. So far most studies on bilingual text archives have been on parallel corpora, i. e.sets of translationally equivalent texts. Such corpora provide valuable indepth evidence of how texts written in one language can be rendered in another according to contextual factors, such as style, register, domain, etc., and are thus important sources of data when studying the translation processo However, a criticism made of the results of analyses based on these corpora is that a translation is alwaysinfluenced by the source text and, therefore, the target language text(s) can never be considered a true representation of that language in usage. It has been suggested that a more reliable source for certain studies, e.g. how a particular concept is expressed independently in different languages, is a comparable corpus. Comparable archives are sets of texts from pairs or multiples of languages that concem a given domain and can be contrasted because of their common features. However, the development of procedures to query data from this kind of corpus is far more problematic than when interrogating parallel archives as it is difficult to identify automatically equivalent expressions in the different languages. We have studied a solution to this problem. We will present procedures that have been developed at the Istituto di Linguistica Computazionale, Pisa, to construct and query both parallel and comparable bilingual text corpora and to extract significant data for translation purposes and contrastivetextual studies, giving examples of the different types of results that can be obtained. The procedures form part of an integrated mono- and bilingual text and lexical database system, implemented in a prototypal version at the Institute.