Conversation electrified: ERP correlates of speech act recognition in underspecified utterances

Area of Science:

• Psycholinguistics
• Cognitive Neuroscience
• Computational Linguistics

Background:

• Understanding speech acts (verbal actions) is vital for conversation, but utterances are often ambiguous.
• Listeners use conversational context to interpret underspecified speech acts.
• Early recognition of speech acts facilitates efficient conversational turn-taking.

Purpose of the Study:

• Investigate the timing of auditory speech act recognition in underspecified utterances.
• Examine how sequential conversational context influences speech act processing.
• Test the hypothesis of early speech act recognition for conversational efficiency.

Main Methods:

• Recorded event-related potentials (ERPs) while participants listened to dialogues.
• Participants performed an action categorization task on target utterances.
• Target utterances were identical but varied in speech act and contextual fit.

Main Results:

• Early frontal positivity (200 ms) indicated speech act recognition during partial utterance processing.
• Highly constraining contexts showed no late ERP effect, suggesting pre-final word recognition.
• Less constraining contexts elicited late posterior negativity, indicating additional processing for complex actions.

Conclusions:

• Speech act recognition begins very early in auditory processing.
• Listeners predict speech acts before utterance completion in constrained contexts.
• Conversational sentence comprehension relies on early verbal action recognition.