Combining Different Features of Idiomaticity for the Automatic Classification of Noun+Verb Expressions in Basque

摘要

We present an experimental study of how different features help measuring the idiomaticity of noun+verb (NV) expressions in Basque. After testing several techniques for quantifying the four basic properties of multiword expressions or MWEs (institutionalization, semantic non-compositionality, morphosyntac-tic fixedness and lexical fixedness), we test different combinations of them for classification into idioms and collocations, using Machine Learning (ML) and feature selection. The results show the major role of distributional similarity, which measures composi-tionality, in the extraction and classification of MWEs, especially, as expected, in the case of idioms. Even though cooccurrence and some aspects of morphosyntactic flexibility contribute to this task in a more limited measure, ML experiments make benefit of these sources of knowledge, allowing to improve the results obtained using exclusively distributional similarity features.