The role of idiom length and context in spoken idiom comprehension

Abstract

Two cross-modal lexical decision experiments investigated the role of the length of the idiom string (Experiment 1) and of prior sentential context (Experiment 2) in spoken idiom recognition. The idiomatic meaning was activated at the offset of long idioms but not of short idioms when the idiom was preceded by a neutral context. The idiomatic meaning of short idioms was instead activated at the string offset when the idiom was preceded by an idiomatic context. The results support the Configuration Hypothesis (Cacciari & Tabossi, 1988).

Keywords:

• Idiom
• Spoken language
• Comprehension
• Lexical processing
• Cross-modal lexical decision

Acknowledgements

We thank Vinanda Var for collecting and analysing the data of Experiment 1. The present research was supported by two grants PRIN 2005 to PT and to CC, respectively.

Notes

¹An example of the experimental sentences of Titone and Connine's study is as follows: The new assignment was a piece of cake compared to the last one. The idiom-related and idiom-unrelated targets were EASY and LIVE, respectively.

²To control for decomposability, we followed the same instructions and procedure of Gibbs et al. (1989): 15 participants were presented with a written list containing the 36 idioms and were asked to decide whether each idiom was decomposable or nondecomposable. Two examples were provided. Six short, seven long unpredictable, and six long predictable expressions were judged decomposable, and the remaining nondecomposable (average agreement 70.04%). To assess the role of decomposability, we performed a new set of ANOVAs adding decomposability (decomposable vs. nondecomposable idioms) to the factors of Experiment 1. Decomposability was within-subject in the by-participant analysis and within-subject in the by-item analysis. We replicated the previous results. The decomposability factor did not enter in any significant interaction and was reliable as a main factor only in by-participant analysis, F ₁(1, 49) = 7.93, MSE=5850, p=.007; F ₂<1, with the targets for decomposable idioms faster than for nondecomposable idioms.

³In Experiment 2 we had six decomposable and six nondecomposable idioms. To control for an effect of this factor, we performed a new set of ANOVAs adding decomposability (decomposable vs. nondecomposable idioms) to the idiom/target relationship factor. Decomposability was within-subject in the by-participant analysis and within-subject in the by-item analysis. We replicated the results relative to the idiom/target relationship factor and found a significant main effect of the decomposability factor (M=622 ms, SD=77 ms; M=678 ms, SD=76 ms, respectively), F ₁(1, 26) = 15.8, MSE=84802, p<.001; F ₂(1, 10) = 8.3, MSE=21416, p<.02; MinF'(1, 21) = 5.44, p=.03. However, the interaction was not significant, F ₁ and F ₂<1, suggesting comparable context effects for decomposable and nondecomposable idioms.