Phraseology and Cognitive Entrenchment : Corpus-based Evidence and Applications for Language Teaching and Translation

Cognitive entrenchment, originating from cognitive grammar, actually comes very close to other theoretical notions such as reproducibility, fixedness or idiomaticity. By the means of experiments carried out on huge linguistic corpora, computational phraseology makes it possible to find partial evidence for the theoretical notions, and to offer at the same time practical tools to language users in general. This paper provides evidence for the probabilistic nature of the network of constructions. Indeed, the same statistical score, the cpr-score, developed in the first place for the extraction of phraseology, turns out to yield significant results for other types of constructions: lexical ones (in the case of Chinese word segmentation), cultural references and named entities, and even more schematic or abstract patterns underlying syntactic constructions