Shared perceptual processes in phoneme and word perception: Evidence from aphasia

• ¹ Rice University, Psychology, United States

Introduction Dissociations between preserved word recognition with impaired phoneme perception have long been noted (e.g., Blumstein, Cooper, Zurif & Caramazza, 1977; Miceli, Gainotti, Caltagirone, & Masullo, 1980). This dissociation is surprising given the assumption that word perception depends on phoneme perception. Consequently, some researchers have claimed that different perceptual processes are involved in phoneme and word perception (Blumstein et al., 1977) and that there are two routes for speech perception – one sublexical and one lexical (Hickok & Poeppel, 2001). An examination of prior findings indicates, however, that perceptual discriminability of targets and distractors has often not been closely matched across the phoneme and word perception tasks (e.g., Blumstein et al., 1977). The present study revisited the issue of separate processes for phoneme and word perception using tasks with closely matched perceptual demands. Method Eight individuals with aphasia were tested on auditorily presented consonant discrimination, word discrimination, and lexical decision tasks using natural speech tokens. In the word discrimination task, subjects made same/different judgments to single syllable words that differed by one distinctive feature in either the initial or final phoneme (e.g., pat-bat, voicing), with one item spoken by a male and one by a female. To create the consonant discrimination stimuli, the initial or final phoneme of the word was deleted (e.g., /pæ/-/bæ/). The lexical decision task presented single spoken stimuli, using the words from the discrimination task and non-words created by changing a single distinctive feature of the word stimuli (e.g., bat-bap, place). To compare to prior studies, scores are also reported on auditory word recognition from the Western Aphasia Battery (WAB; Kertesz, 1982), which does not include phonologically close distractors, and on a picture-word matching task (PWM) with phonologically related distractors differing from targets by more than one distinctive feature. Results and Conclusions Replicating previous studies, performance on the two word recognition tasks without closely matched distractors (WAB and PWM) was at ceiling for some subjects with impairments on consonant discrimination (see Figures 1a/1b). However, as shown in Figures 1c/1d, for word processing tasks matched in phonological discriminability to the consonant discrimination task, scores on consonant discrimination and word processing were highly correlated, and no individual demonstrated substantially better performance on word than phoneme perception. One patient demonstrated worse performance on lexical decision (d’ = .21) than phoneme perception (d’ = 1.72), which can be attributed to impaired lexical or semantic processing. These data argue against the hypothesis that phoneme and word perception rely on different perceptual processes/routes for processing, and instead indicate that word perception depends on perception of sublexical units.