Statistical and connectionist models of speech perception and word recognition

M. Gareth Gaskell

Published in print August 2007 | ISBN: 9780198568971
Published online September 2012 | | DOI:
 Statistical and connectionist models of speech perception and word recognition

More Like This

Show all results sharing these subjects:

  • Psychology
  • Cognitive Psychology
  • Cognitive Neuroscience


Show Summary Details


This article reviews the current impact of connectionism in the area of speech perception and spoken word recognition. A major advance that connectionism provided was to highlight the value and power of statistical models of language processing. Therefore, some types of statistical model—particularly those stressing statistical learning—are reviewed alongside connectionist theories such as interactive activation and competition models, error-driven learning networks, and adaptive resonance theory. The article examines how connectionist models represent speech pre-lexically, and how such prelexical representations might develop and adapt to fit the requirements of the perceptual system. It also looks at the process of word segmentation, again addressing both acquisition issues and the degree to which connectionist models can explain performance in the adult system. Finally, the article considers the process of word recognition, as modeled in terms of lexical competition. Key issues here include whether distributed models can cope with the specific properties that are imposed by the speech medium, such as the drawn-out nature of the input and the consequent requirement to entertain multiple hypothesis (parallel activation) during recognition.

Keywords: statistical models; connectionist models; speech perception; word recognition; word segmentation; connectionism; learning networks; adaptive resonance theory; prelexical representations; lexical competition

Article.  10903 words. 

Subjects: Psychology ; Cognitive Psychology ; Cognitive Neuroscience

Full text: subscription required

How to subscribe Recommend to my Librarian

Buy this work at Oxford University Press »

Users without a subscription are not able to see the full content. Please, subscribe or login to access all content.