Impaired Resolution of Ambiguous Homographs in High-Functioning Individuals with Autism: An ERP Study

Background:  Autism spectrum disorders (ASDs) are partly characterized by deficits in communication and pragmatic language, particularly in determining the meaning of ambiguous words in context. For example, when reading sentences aloud, individuals with ASD are more likely than individuals with typical development (TD) to mispronounce ambiguous homographs like “lead” and “tear”. Traditionally this difficulty has been interpreted within the “weak-central coherence” theory of autism, which suggests that individuals with ASD have impaired contextual integration. However, difficulty with correctly reading homographs could also stem from executive control deficits, such that individuals with ASD have difficulty inhibiting the context-inappropriate meaning of the homograph.

Objectives:  We used the N400 event-related potential (ERP) to tease apart executive control and context sensitivity processes in individuals with ASD. Using sentence stimuli designed to distinguish between context sensitivity and executive control, we compared N400 responses to lexical ambiguity and semantic incongruity in ASD and TD groups. If impairments in ambiguity resolution in the ASD group result from difficulty attending to contextual clues, then these impairments should correlate with a reduced N400 to incongruous words. Conversely, a dissociation between these two mechanisms would suggest that individuals with ASD have a specific impairment in disambiguating meaning.

Methods:  Twenty adults with ASD and 20 matched TD adults (ages 18-68) read 120 sentences for comprehension, responding to relatedness probes after each sentence. Materials consisted of sets of four sentences, adapted from Sitnikova et al. (2002). All sentences had a first clause that biased either the dominant or subordinate meaning of a homograph. A second clause contained a target word associated with the homograph’s dominant meaning. Finally, a control word at the end of the sentence was either congruent or incongruent with the preceding sentence context. Sentences were presented in rapid serial visual presentation (RSVP) format during concurrent EEG recording.

Results:  Analyses were performed at nine regions across the scalp (representing frontal, central, and parietal sites and left, midline, and right lateralities). Repeated-measure ANOVAs were run with factors of congruity (congruent/incongruent), site (frontal/central/parietal), laterality (left/midline/right), and group (ASD/TD). At the control word, both groups showed a significant N400 effect from 300-500 ms over central and parietal scalp. The magnitude and topography of the N400 effects were comparable between groups. At the target word, the TD group showed a small N400 effect from 350-450 ms over the right hemisphere. In contrast, the ASD group did not show a significant N400 effect in response to the target word.

Conclusions:  Both groups showed an N400 effect in response to control words, indicating that the ASD group effectively attended to sentence context. In contrast, the ASD group showed no N400 effect to the target words, indicating inappropriate ambiguity resolution. The dissociation between these two effects suggests that inattention to context is an unlikely explanation for the disambiguation deficits in the ASD group. Instead, atypical ambiguity resolution may result from inappropriate executive control mechanisms, such that individuals with ASD are unable to suppress the context-inappropriate meaning of the homograph.