The role of working memory and idiom compositionality in idiom comprehension

Abstract

Figurative language use is not limited to poetry or literature but is a ubiquitous part of speech. Studies that looked at figurative language comprehension have shown that some cognitive mechanisms, such as working memory, may be involved in figurative language comprehension. For example, individuals with high working memory span tend to produce deeper metaphor interpretations. The current work was interested in how working memory is involved in a particular figure of speech comprehension: idioms. An idiom is a phrase whose meaning cannot be simply deduced from the literal meanings of the words that comprise that idiom. Idioms can vary according to their compositionality, which refers to the extent with which meanings of the idiom constituents provide cues for the idiom's idiomatic meaning. A number of researchers agreed upon certain idioms being decomposable and other idioms being fixed. The two different types were used in the Main Study. Models of idiom comprehension also vary from traditional "lexical look-up" models that consider idioms as multi-word lexical units stored as such in speakers' mental lexicons to "nonlexical" models, such as the Configuration Hypothesis, that states that an idiom as a whole does not have a separate lexical representation in the mental lexicon. Both models are considered in this work. Finally, understanding idiomatic expressions may require inhibiting irrelevant literal information. For example, literal meanings of the words dogs and cats in an idiom it is raining cats and dogs have to be inhibited in order to gather the figurative meaning of the expression. Thus, the main objective of the current work was to assess the role of working memory in idiom comprehension, as well as to explore whether idiom compositionality had an effect on how fast idioms were interpreted, while also considering implications for the two main models of idiom comprehension. A Preliminary Study narrowed down the list of idioms to the 26 that were used in the Main study, ensuring that both types of idioms did not differ in familiarity or length. The Main Study consisted of four tasks: working memory (Operation span task), inhibition (reading with distractions), idiom comprehension, and familiarity. Seventy-three general psychology students participated in the Main Study. The data were analyzed by several regression analyses and t-tests. The main finding was that there seems to be a difference in a way the two accepted types of idioms are interpreted: fixed idioms were interpreted faster than decomposable idioms. This is consistent with the lexical lookup hypothesis but only for fixed idioms and suggests that readers may not have to analyze the literal word meanings of fixed idioms when interpreting them, thus making their interpretation faster, since retrieving is faster than computing. Neither familiarity nor idiom length could account for this difference. On the other hand, neither operation span nor the number of critical errors committed by participants on the inhibition task predicted how long it took participants to interpret either type of idioms. Several possible explanations for such results are discussed, as well as the limitations and future directions.