Abstract
Even though the outcome of the perception of phonological patterns is categorical, this process might still arise from continuous dynamics. Here, we propose a unified dynamical account of three types of behavior that are usually studied in isolation: short-term perceptual competition, long-term perceptual habituation, and even longer-term perceptual learning. We develop a model and test its predictions in two speech identification tasks on an acoustic continuum between the French words [sɛp] and [stɛp]. When presenting stimuli sequentially from one end of the continuum to the other, we found that the presentation order systematically changed the position of the perceptual switch from one word to the other. We also found that response times were slower and more variable around this perceptual switch, regardless of its position on the acoustic continuum. And, throughout the experiment, participants became more sensitive to small acoustic differences between stimuli. Our model can account for these results and for a surprising finding, namely that the initial presentation order affected responses even late in the experiment. Overall, our results point to the importance of the relation between fast processes responsible for competition, and slow processes responsible for habituation and learning in explaining how listeners can perceive speech categorically in a way that is both flexible and robust.
Original language | English |
---|---|
Journal | Laboratory Phonology |
Publication status | Published - 2013 |
Keywords
- speech perception
- speech
- perception
- learning
- Computational Modeling