Corpus exploitation for terminological purposes: a proposed term extraction process for bilingual specialised dictionary elaboration

The method and outcomes described in this research are part of a wider project aimed at creating a specialised, bilingual dictionary on industrial ceramics. One of the key stages in the elaboration of such a work is the term extraction process, since it determines which terminological units are part of the domain and should thus be included as dictionary entries. This paper deals with a proposal that is aimed at accurately detecting the units of specialised knowledge that constitute the terminology of a domain from a raw, bilingual corpus that was compiled ad hoc. This proposal is presented in the form of a series of stages in which prospective terms are progressively retrieved, identified and analysed with the concordance software program WordSmith Tools (WST) 5.0. In the method proposed here, the WST application WordList first generates monolexical frequency lists which are compared with the lists generated by the tool KeyWords, providing saliency data. Afterwards, Mutual Information lists provided once again by Wordlist constitute the first approach to the combinatorial aspect of terms. Subsequently, the WST application Concord, with its different options, provides further evidence on the way prospective terms collocate and combine as well as on their contextual nature, thus completing the term extraction process. This methodology, always combined with the terminographer’s “manual” work, observation and intuition, has proved to be effective for the dictionary under development