Eye Gaze Provides a Window on Children's Understanding of Verbal Irony

Abstract

We investigated how children solve the interpretive problem of verbal irony. Children 5 to 8 years of age and a group of adults were presented with ironic and literal remarks in the context of short puppet shows. The speaker puppet's personality was manipulated as a cue to intent; that is, speakers were described as funny or serious. We measured all participants' interpretations of the remarks and also children's eye gaze and response latencies as they made their interpretations. As expected, children were less accurate than adults in their judgments of speaker intent. Although children took longer to judge speaker intent for ironic remarks than literal remarks, eye gaze data showed no evidence that children had a literal-first bias in their processing of ironic language. Instead, children's eye gaze behavior suggested that they considered an ironic interpretation even in the earliest moments of processing. We argue that these results are most consistent with a parallel constraint satisfaction framework for irony comprehension.

ACKNOWLEDGEMENTS

This research was supported by a Social Sciences and Humanities Research Council of Canada (SSHRC) research grant to P. M. P.

We thank Val San Juan for assistance with coding.

Notes

Note. Italics indicate conditions that were excluded from analyses because of low variance. Means in these conditions are presented here for purposes of visual comparisons.

Note. Mean proportions were calculated for trials on which speaker intent was correctly attributed. The “correct” inferences are that ironic remarks are intended to be funny whereas literal remarks are not. Italics indicate conditions that were excluded from analyses because of low variance or because there were too few correct speaker intent responses to warrant analyses. Means in these conditions are presented here for purposes of visual comparisons.

Note. Mean latencies were calculated for trials on which the speaker intent response was correct. Italics indicate conditions that were excluded from analyses because there were too few correct speaker intent responses to warrant analyses. Means in these conditions are presented here for purposes of visual comparisons.

Note. Means were calculated only for trials on which the speaker intent response was correct. Italics indicate conditions that were excluded from analyses because there were too few correct speaker intent responses to warrant analyses. Means in these conditions are presented here for purposes of visual comparisons.