This project is supported by the Social Sciences and Humanities Research Council (SSHRC).
Most terminological resources (specialized dictionaries and term banks) do not devote enough attention to the linguistic functioning of terms. This observation is valid for monolingual resources as well as multilingual resources. This project aims to provide insights for the development of multilingual terminological resources that take into consideration the linguistic properties of terms, and more specifically their syntactico-semantic properties. We focus on the interlinguistic equivalence relationships between predicative terms (verbs, adjectives and predicative nouns) that are not adequately described in terminology. Since terminology is usually interested in terms that denote entities (computer, data, ozone), it often ignores predicative terms (warming, configure, compatible) the characteristic of which is to require arguments (warming of X, X configures Y, X is compatible with Y). This research is based on Frame Semantics, FS (Fillmore 1977, 1982; Fillmore and Atkins 1992) and its implementation in FrameNet (Atkins et al. 2003; Fillmore 2003a; Ruppenhofer et al. 2006). The “FrameNet” methodology provides for a previous annotation of sentences that aims to give a full description of the syntactic and semantics properties of lexical units. Our project also has applied objectives. We will apply and adapt a methodology previously devised for French to other languages, i.e. English and Spanish. In addition, we will extend the computational methods developed for French with satisfactory results to these languages. This extension will inevitably require some substantial adjustments. We plan to use methods that are language-independent (and this becomes possible with machine learning techniques.