1-The cohort model predicts that the initial part of a word will be
more important for lexical access than its end,
2-Mispronunciations at the beginnings of words are detected more accurately
than are mispronunciations at the ends of words
3-The phoneme restoration effect is also more robust
when the missing phoneme is in the middle or at the end of a word
rather than at the beginning