Theory 03/Hearing & Auditory Science /Speech perception/3 min
Speech Is Tracked by Rhythm
We treat speech as a code of frequencies. It may be a code of beats.
Speech perception depends on cortical entrainment to syllable rhythm more than on spectral decoding, so comprehension should be predicted by neural phase locking to the syllable rate rather than by spectral resolution, and rhythm training should improve comprehension more than spectral training.
01The evidence
- Cortical entrainment to continuous speech has been characterised as a functional mechanism rather than an epiphenomenon, with neural phase locking to syllabic and prosodic rates proposed to segment the stream.Cortical entrainment review
- Speech perception is incremental and online, and listeners track several levels of linguistic organisation simultaneously, which requires a timing mechanism rather than a purely spectral one.Rhythmic coordination in speech
02The tension
Speech is modelled as a sequence of acoustic features decoded by spectral analysis, and hearing loss is characterised by spectral resolution. That model predicts that comprehension should track spectral acuity, which it does only weakly, especially in noise where rhythmic cues remain intact.
03The theory
Speech arrives as a modulated stream with a strong rhythm at the syllable rate. The auditory system phase-locks to that rhythm, and the phase relationship tells the listener where boundaries are, which is the segmentation problem that spectral models solve only implicitly.
If rhythm is the carrier, then comprehension in noise should depend on the integrity of the phase lock rather than on the signal-to-noise ratio in each frequency band. That predicts that listeners with poor spectral resolution but good entrainment can understand degraded speech, which is observed.
The prediction is that individual differences in comprehension should correlate with entrainment strength better than with audiometric thresholds, and that rhythmic training should transfer to comprehension while spectral training does not.
The clinical consequence is that auditory rehabilitation should include rhythmic training, and that hearing-aid fitting should preserve temporal envelope cues even at the cost of spectral fidelity.
04What this is not
Those models decode a sequence of spectral features and treat timing as a byproduct. This makes phase locking the mechanism and predicts that rhythm, not spectrum, predicts comprehension.
05Predictions
- P1Comprehension should correlate with syllable-rate entrainment better than with spectral acuity.
- P2Rhythm training should transfer to comprehension; spectral training should not.
- P3Preserving temporal envelope at the cost of spectrum should preserve intelligibility.
06What would refute it
Comprehension tracking spectral resolution at fixed entrainment would restore the spectral account.
07What it would change
It would change hearing-aid design and rehabilitation toward timing.
08Evidence base
- Cortical entrainment to continuous speech: functional roles and interpretationshttps://scholar.google.ro/citations?view_op=view_citation&hl=ru&user=Q_mMDVMAAAAJ&citation_for_view=Q_mMDVMAAAAJ:roLk4NBRz8UC
- Rhythmic coordination as a scaffold for communicative alignmenthttps://www.cambridge.org/core/services/aop-cambridge-core/content/view/6BE55F0B8BDA10A4C5FA91CB0EDB38D8/9781009295840c29_533-550.pdf