Poor sensitivity to sound statistics impairs the acquisition of speech categories in dyslexia

Research output: Contribution to journalReview articlepeer-review

24 Scopus citations

Abstract

Accumulative studies suggest that people with dyslexia have impaired categorical perception of speech. However, the origin of this deficit remains unknown. We now propose that impaired sensitivity to distributional statistics in dyslexia underlies this deficit. This proposal is based on the following rationale. The literature on acquiring speech categories in the general population attributes a major role to learning distributional statistics. Sensitivity to the statistics of stimuli used in perceptual discrimination tasks is impaired in dyslexia. Specifically, dyslexics’ implicit memory of previous stimuli decays faster than that of good readers and their stimulus-specific adaptation processes are shorter. Their faster decay may limit the temporal window over which they implicitly calculate stimulus distributions, and yield slightly impoverished speech categories. Since impaired sensitivity to stimulus statistics and impaired categorisation in dyslexia are domain general, we further propose that impaired categorical speech perception is one manifestation of a broader deficit in distributional learning.

Original languageEnglish
Pages (from-to)321-332
Number of pages12
JournalLanguage, Cognition and Neuroscience
Volume33
Issue number3
DOIs
StatePublished - 16 Mar 2018

Bibliographical note

Publisher Copyright:
© 2017 Informa UK Limited, trading as Taylor & Francis Group.

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

  1. SDG 4 - Quality Education
    SDG 4 Quality Education

Keywords

  • adaptation processes
  • anchoring hypothesis
  • Bayesian inference
  • Reading
  • skill acquisition
  • speech perception

Fingerprint

Dive into the research topics of 'Poor sensitivity to sound statistics impairs the acquisition of speech categories in dyslexia'. Together they form a unique fingerprint.

Cite this