Towards a phonetic feature-based account of dyslexic speech perception deficits.

摘要

The most prominent and enduring theories of reading disability have typically attributed the speech perception deficits of dyslexics to poor phonological processing. Other theories, however, have focused on poor sensory acuity as the proximal cause of these deficits. A series of experiments was, thus, conducted to test the veracity of these claims. Experiment 1 demonstrated that both dyslexic and auditory neuropathy (AN) subjects are poor at processing a broad range of consonant-vowel (CV) syllables in speech-shaped noise. Both disorders have previously been associated with poor rapid auditory processing. Unlike AN subjects, for whom centralized speech processing mechanisms are purportedly intact, the dyslexics were ineffective in utilizing visual articulatory cues to circumvent difficulties in processing noise-masked speech. Experiment 2 extended this audio-visual twist to the standard categorical perception experimental paradigm to see if the dyslexics' categorization of synthetic VCV stimuli could be biased toward either endpoint of a place or manner of articulation contrast by corresponding visual cues. As expected, the dyslexics exhibited distorted categorical functions for an acoustic /ada/ vs. /aga/ place contrast in noise; again their categorizations were less influenced by the introduction of corresponding visual cues. However, the dyslexics showed better categorization of an acoustic /aba/-/awa/ manner contrast, and greater influences of the corresponding visual cues. Feature-based accounts of poor phonological processing in line with Elbro (1996) may best account for our observation that, aurally or visually, place of articulation contrasts are more problematic than manner contrasts for dyslexics. A final set of experiments investigated this further, examining dyslexics' processing of critical vocalic cues to word-final consonant place, manner, and voicing relative to age-matched and reading level controls. The dyslexics were sharper in their categorization of word-final consonants that, again, differed in manner rather than place. Although dyslexics were just as categorical as controls on continuous changes in vocalic duration---a strong cue to final consonant voicing---they were less influenced by manipulations of F1 vocalic offset cues than controls. Interestingly, these cues were less well-specified in dyslexic vocalizations. Implications for potential links between speech perception, speech production, and phonetic feature acquisition are discussed.