Related Experiment Videos
Competition and segmentation in spoken-word recognition
D Norris1, J M McQueen, A Cutler
1Medical Research Council (MRC), Applied Psychology Unit, Cambridge, United Kingdom.
Journal of Experimental Psychology. Learning, Memory, and Cognition
|September 1, 1995
Summary
Listeners easily identify words in speech due to a model combining lexical competition and prosodic structure sensitivity. This spoken-word recognition ability balances word candidate competition with metrical segmentation for accurate word identification.
Area of Science:
- Psycholinguistics
- Computational Linguistics
- Auditory Perception
Background:
- Listeners identify words in continuous speech despite limited boundary cues.
- Spoken-word recognition models are crucial for understanding auditory processing.
Purpose of the Study:
- To investigate how listeners identify words in continuous speech.
- To evaluate a model integrating lexical competition and prosodic structure.
Main Methods:
- A word-spotting experiment was conducted.
- Simulations using a computational model (Shortlist with Metrical Segmentation Strategy) were performed.
- A lexicon of over 25,000 words was utilized.
Main Results:
- Prosodic effects were most evident with high lexical competition.
- Lexical competition modulates the impact of prosodic structure on word recognition.
- The model accurately simulated experimental findings.
Conclusions:
- Spoken-word recognition relies on both lexical competition and prosodic structure.
- Integrating the Metrical Segmentation Strategy into the Shortlist model provides a robust account of word recognition.
- Auditory processing effectively combines bottom-up (lexical) and top-down (prosodic) information.