Ch 3: Language Processing

Segmenting the Speech Stream
1. Results due to exposure to speech.
  1. 8-Month-old infants can pick out words within 2 mins of hearing artificial language.
2. Pre-Lexical
  1. Uses speech stream characteristics.
    1. Coarticulation makes it harder to detect.
    2. Some word boundaries harder to detect.
  2. English: Metrical foot.
    1. Stressed syllable, followed by one or more unstressed.
    2. Cutler & Norris: Wordspotting
    3. Rhythmic units different in other languages.
3. Lexical
  1. Phonological representation.
    1. Knowledge of what words sound like.
    2. Parallel Activation: Works like predictive text!
  2. Models include: TRACE, Cohort model.
Cohort Model
1. Parallel Activation
  1. Word-initial cohorts are activated
    1. Uniqueness Point is reached

Resource summary

	Created by Kev Campbell over 12 years ago