Probability and surprisal in auditory comprehension of morphologically complex words

Area of Science:

• Psycholinguistics
• Cognitive Science
• Computational Linguistics

Background:

• Auditory word recognition involves identifying spoken words from continuous speech.
• Morphologically complex words present unique processing challenges due to their constituent morphemes.
• Existing models of auditory comprehension, like Shortlist B, offer frameworks for understanding word recognition.

Purpose of the Study:

• To investigate the processing of morphologically complex words in auditory lexical decision tasks.
• To identify critical points in word recognition where probabilistic information shifts significantly.
• To explore the role of surprisal and information theory in auditory word comprehension.

Main Methods:

• Two auditory lexical decision experiments were conducted.
• Participants responded to target words presented auditorily.
• Analysis focused on response latencies in relation to word properties like uniqueness points and surprisal.

Main Results:

• Two key points were identified: initial uniqueness point (UP1) and complex uniqueness point (CUP).
• Later UP1 and CUP were associated with longer response latencies.
• Cumulative surprisal throughout the word also influenced response latencies.

Conclusions:

• Uniqueness points and surprisal significantly impact auditory word recognition.
• The findings challenge the Shortlist B model.
• A Bayesian approach to auditory comprehension may need integration with information theory to account for cognitive costs.