Global Similarities and Multifaceted Differences in the Production of Partner-Specific Referential Pacts by Adults with Autism Spectrum Disorders

Abstract

Over repeated reference conversational partners tend to converge on preferred terms or referential pacts. Autism spectrum disorders (ASD) are characterized by pragmatic difficulties that are best captured by less structured tasks. To this end we tested adults with ASD who did not have language or intellectual impairments, and neurotypical comparison participants in a referential communication task. Participants were directors, describing unlexicalized, complex novel stimuli over repeated rounds of interaction. Group comparisons with respect to referential efficiency showed that directors with ASD demonstrated typical lexical entrainment: they became faster over repeated rounds and used shortened referential forms. ASD and neurotypical groups did not differ with respect to the number of descriptors they provided or the number of exchanges needed for matchers to identify figures. Despite these similarities the ASD group was slightly slower overall. We examined partner-specific effects by manipulating the common ground shared with the matcher. As expected, neurotypical directors maintained referential precedents when speaking to the same matcher but not with a new matcher. Directors with ASD were qualitatively similar but displayed a less pronounced distinction between matchers. However, significant differences and different patterns of reference emerged over time; neurotypical directors incorporated the new matcher's contributions into descriptions, whereas directors with ASD were less likely to do so.

Keywords: audience design; autism spectrum disorders; common ground; language production; lexical entrainment; partner-specificity; referential pact; referential precedent.

Figure 1

Tangram stimuli used in our task, Set A (top row) and Set B (bottom row). Each set was used for one condition (same vs. new matcher).

Figure 2

The top panel shows the same matcher condition, where dyads in both groups get much faster over 5 rounds of discussing the figures, reflecting lexical entrainment. Results differ however in the new matcher condition in the bottom panel, where dyads in both groups show a disruption of lexical entrainment when a new matcher is introduced on the fourth round. Gray dots indicate jittered data points, black dots indicate outliers.

Figure 3

Average number of exchanges required for the matcher to locate the figure, by condition and group.

Figure 4

Difference in duration of Round 1 vs. Round 4, where the new matcher switch occurred, by condition and group. Positive values indicate speeding up over four rounds of referring to the same figures, and a 0 duration difference indicates taking the same time on Round 4 as on the first presentation of the cards on Round 1.

Figure 5

Types of descriptions, with respect to the referential precedent of the prior round, given on critical Round 4 to the same matcher (left) vs. new matcher (right).

Figure 6

Maintenance of referential precedent from Round 3 on Round 4, by condition and group.

Figure 7

Relation between maintenance of referential precedent on Round 4 with new matcher and the duration difference between Round 1 and Round 4 (positive values indicate speeding up over repeated reference).

Figure 8

Maintenance of referential precedent from Round 4 on Round 5, by condition and group. Note: The NT group displayed little variability, with an interquartile range of 1 that was too small to appear in this boxplot.