Idiom Variation: Experimental Data and a Blueprint of a Computational Model

Area of Science:

• Psycholinguistics
• Computational Linguistics
• Cognitive Science

Background:

• Corpus surveys reveal that idioms, or multi-word expressions, are not always realized in a fixed form.
• This variation in idiom realization challenges existing processing theories that assume fixed multi-word units.

Purpose of the Study:

• To investigate the varying acceptability of different idiom variant forms.
• To propose a computational model that explains how listeners comprehend these variants.

Main Methods:

• A rating experiment was conducted to assess the acceptability of idiom variants.
• A computational model simulating naive discriminative learning was developed to model comprehension.

Main Results:

• The study found that alternative realizations of idioms exhibit varying degrees of acceptability.
• The computational model successfully explains the differential acceptability of various idiom variant types.

Conclusions:

• Idiom variation is a key phenomenon that requires flexible processing models.
• The proposed computational model offers a viable explanation for idiom comprehension, challenging fixed-unit theories.