BILINGUAL REFERENCE CORPORA - A SYSTEM FOR PARALLEL TEXT RETRIEVAL

A system for the automatic creation and retrieval of parallel concordances from a Bilingual Reference Corpus is described. At present, the system runs on a test set of ltalian/English 'bilingual' texts. A description is given of the different stages of the procedure which aligns and links equivalent texts in the two languages and of the way in which the query system uses these links to construct parallel contexts. The procedure uses a number of the components of the PiSystem, an integrated set of tools for text processing and analysis, including morphological analyzers and generators, and monolingual and bilingual lexical database systems. The user can use the system to query either of the two sets of texts (Italian and English) and, for any form or cooccurrences of forms found in the set of texts for one language, can retrieve parallel contrastive contexts from the other. The system should be of particular interest to bilingual lexicographers, translators and linguists, and should also find applications in the office automation area and in the language industry in general.