Marrying cognitive-linguistic theory and corpus-based methods: On the compositionality of English V NP-idioms

This paper presents a corpus-linguistic compositionality measure for V NP-constructions (make a point, take the plunge) that goes beyond previous approaches in trying to implement fundamental assumptions of cognitive-linguistic/constructionist approaches to language: Complex constructions are regarded as manifestations of several smaller constructions, all of which contribute to the phrasal meaning. Accordingly, the measure presented here quantifies the contributions made by both the verb and the noun to the V NP-construction. Constructions are assumed to be differently entrenched in the mental lexicon, depending – among other things – on their frequency. Accordingly, differences in entrenchment between component words of an idiomatic construction are expected to influence the relative weight of their contributions, which is accommodated by the measure proposed here. The measure accommodates a potential backward influence of the construction’s semantics on the (weightings in the often polysemous network of) the constituent word’s semantics by weighting the contribution of each component word not only in terms of how much of the construction’s meaning is accounted for, but also how much of itself each component word brings in. In take the plunge, e.g., plunge brings in nearly all of its semantics, whereas take only contributes a fraction of its meaning potential. The results obtained for more than 13,000 tokens of 39 V NP-constructions extracted from the British National Corpus tie in well with established findings from psycho-linguistic research. Ultimately, the model strengthens the case for usage-based approaches to grammar, demonstrating that a seemingly intuition-based, complex phenomenon can be modeled bottom-up using performance data.