Related Experiment Videos
Illusory continuity of interrupted speech: speech rate determines durational limits
J A Bashford1, M D Meyers, B S Brubaker
1Department of Psychology, University of Wisconsin, Milwaukee 53201.
The Journal of the Acoustical Society of America
|November 1, 1988
Summary
The phonemic restoration effect allows listeners to perceptually fill in missing speech sounds replaced by noise. Study shows continuity limits adapt to speech rate, suggesting a fixed information, not time, basis for this auditory illusion.
Area of Science:
- Auditory perception
- Psychoacoustics
- Speech processing
Background:
- The phonemic restoration effect demonstrates perceptual completion of speech segments masked by noise.
- Previous research linked continuity limits to average word duration in interrupted discourse.
Purpose of the Study:
- To investigate how varying speech playback rates affect the durational limits of phonemic restoration.
- To explore the relationship between temporal processing, auditory induction, and speech perception.
Main Methods:
- Recorded discourse was interrupted with noise at normal, faster (15% greater), and slower (15% less) playback speeds.
- Listeners' perception of illusory continuity was measured across different speech rates.
- Durational limits for apparent continuity were determined for each playback condition.
Main Results:
- At normal playback speed, continuity limits approximated average word duration, consistent with prior findings.
- Changes in playback rate resulted in proportional shifts in the measured continuity limits.
- This indicates that the temporal extent of illusory continuity is not fixed but scales with speech rate.
Conclusions:
- Phonemic restoration in speech is likely a specialized linguistic application of a general auditory induction mechanism.
- The limits of illusory continuity in discourse appear to be determined by a fixed amount of verbal information, not a fixed temporal duration.
- Auditory perception adapts to process speech information efficiently, even when segments are masked.