Susan Brady

Speech Perception Deficits in Poor Readers A Reply to Denenberg's Critique

We reply to Denenbergs (1999) recent critique of our work (Mody, Studdert-Kennedy, & Brady, 1997). Denenberg mounted two main lines of criticism, one concerning characteristics of the population sampled for the experimental group, and the other a statistical critique, concerning (a) violation of parametric assumptions for use of the F distribution and (b) our supposed acceptance of the null hypothesis of no differences between experimental and control groups. We show that the first criticism stemmed from a misunderstanding of the experimental hypothesis and that the second can be answered by both parametric and nonparametric comparisons across conditions within the experimental group, without reference to the control group. Thus, our original conclusion stands: The difficulty with rapid /ba/-/da/ discrimination that some children with reading impairment may experience does not stem from difficulty in discriminating the rapid spectral transitions at stop-vowel syllable onsets.

Vowel recognition: Inferences from studies of forward and backward masking

The recognition of brief vowels was studied in forward and backward masking tasks. In a series of experiments in which both target and mask parameters were systematically varied, two populations of subjects were identified. The majority (Nonmaskers) evidenced little masking at any interstimulus interval, while relatively fewer subjects (Maskers) evidenced backward masking (but not forward masking) over a 100-200 ms interval. Increasing target set size systematically increased masking for the Maskers but not for the Nonmaskers. Adding white noise to the targets increased the extent of masking for both groups. These results suggest (1) that masking does not impose a substantive constraint on vowel perception in running speech, and (2) that multiple strategies may exist for vowel recognition.

Two processes in vowel recognition: inferences from studies of backward masking

The recognition of brief vowels was studied in forward‐ and backward‐masking paradigms. In a series of experiments in which both target and mask parameters were systematically varied, two populations of subjects were consistently identified. The majority (nonmaskers) evidenced no masking at any interstimulus interval (ISI) while relatively fewer subjects (maskers) evidenced backward masking (but not forward masking) out to 100–200 msec ISI. Increasing target set size systematically increased the extent of masking for the maskers. Adding −15‐dB white noise relative to the vowel amplitude increased the extent of masking for both groups. These results suggest two parallel recognition routines for vowels, at least one of which may involve a serial search process.

A Response to Denenberg

et al., 2000) to his critique (Denenberg, 1999), that alternative nonparametric tests on the experimental group alone without reference to the control group yield experimental conclusions identical to those of the parametric tests he has rejected. Let us briefly review the facts. In Experiments la and 1b of Mody et al. (1997), we found that poor readers who had difficulty in both temporal order judgement (TOJ) and discrimination of two perceptually similar stopvowel syllables, /ba/-/da/, differing only in place of articulation and presented at very short interstimulus intervals (ISI), had no difficulty on either task for perceptually distinct stopvowel and fricative-vowel syllable pairs that differed in place, manner, and voicing, /ba-sa/ or / da-sa / . In our

Speech perception by less‐skilled readers

In the field of reading disability, the major finding of the last 25 years is that poor readers have difficulty attaining explicit awareness of the phonemic components of language. This insensitivity to the phonemic structure of words makes it correspondingly hard to master an alphabetic writing system representing those phonemes. An important question still under investigation is whether difficulty achieving phonemic awareness is entirely a metacognitive problem or whether it might stem from more basic deficits in perception. Using several paradigms (e.g., categorical perception tasks, speech perception in noise, word repetition), the research on this issue has generally supported two conclusions. First, less‐skilled readers apply similar strategies but do so less efficiently and/or less accurately than better readers. Second, these difficulties are restricted to the linguistic domain: less‐skilled readers perform as well as better readers on nonspeech control tasks. Current work is exploring the associa...