By a synthesis of two complementary theories of language this DFG-funded project is pursuing an innovative approach to modelling lexical-semantic relations of collocations.

Academic training, academic career, academic research – these are three of the many typical word combinations with the head-word ‘academic’. New technical opportunities make it possible today to use large corpora to identify very large numbers of typical word combinations, in the form of word profiles. Out of this stock of word combinations lexicographers select the so-called collocations, in order to make clear the contextual meaning of a head-word, and group them in a dictionary entry on the relevant head-word.

The task of the MoKo project is to make it possible to show the collocations of a head-word in a way that, for example, goes beyond mere alphabetic listing. The goal is, rather, to group the collocations identified for a particular head-word in an intelligent way, according to particular aspects of meaning, using so-called lexical-semantic categories.

For the modelling of the internal semantic structure of the collocations, the project is using two complementary reference models: the Meaning-Text Theory of Igor Mel’čuk (lexical functions) and the Generative Lexicon Theory of James Pustejovsky (modification of qualia roles). It will be assessed empirically whether it is possible to achieve a way of modelling collocations that is appropriate for the comprehensive description of larger sections of the lexicon by using a synthesis of those theories. As well as the documentation of the research results, the project also hopes to provide a guide for the targeted recording and detailed description of collocations in dictionaries.