Dictionary Speech, Phraseological Mold and Text Corpus

Starting from the computerized dictionary Le Tresor de la langue francaise informatise as lexicographic corpus, we shall automatically retrieve transparent phraseologisms which have indicators of varying origins, and to extract phraseological molds from them. Then, the dictionary data will be used again to automatically generate phraseologism candidates according to semantic criteria. Finally, the generated phraseologisms will be projected into a large corpus for validation. The objective of our experience is to develop a method and a tool to, on the one hand, simulate the human process in linking dictionary data, and on the other hand, to model the different types of dictionary content in automatic recognition and generation of phraseologisms. This would highlight the role that phraseologisms play in linking semantically lexical units.