Abstract not found

We describe an account of lexically guided tuning of speech perception based on interactive processing and Hebbian learning. Interactive feedback provides lexical information to prelexical levels, and Hebbian learning uses that information to retune the mapping from auditory input to prelexical representations of speech. Simulations of an extension of the TRACE model of speech perception are presented that demonstrate the efficacy of this mechanism. Further simulations show that acoustic similarity can account for the patterns of speaker generalization. This account addresses the role of lexical information in guiding both perception and learning with a single set of principles of information propagation.

Symbols enable people to organize and communicate about the world. However, the ways in which symbolic knowledge is learned and then represented in the mind are poorly understood. We present a formal analysis of symbolic learning—in particular, word learning—in terms of prediction and cue competition, and we consider two possible ways in which symbols might be learned: by learning to predict a label from the features of objects and events in the world, and by learning to predict features from a label. This analysis predicts significant differences in symbolic learning depending on the sequencing of objects and labels. We report a computational simulation and two human experiments that confirm these differences, revealing the existence of Feature-Label-Ordering effects in learning. Discrimination learning is facilitated when objects predict labels, but not when labels predict objects. Our results and analysis suggest that the semantic categories people use to understand and communicate about the world can only be learned if labels are predicted from objects. We discuss the implications of this for our understanding of the nature of language and symbolic thought, and in particular, for theories of reference.

In the last ten years or so, exemplar theory has enjoyed much growth in the field of phonetics. More recently, attempts have been made to apply exemplar theory to syntactic phenomena. Thus far, the issue of unifying phonetic and syntactic exemplar-theoretic models has not been addressed. This paper presents a single overarching exemplar-based model of constituent interactions across both linguistic domains which represents a significant first step towards a unified account of exemplar theory. Our simulations for one phonetic and two syntactic phenomena provide insights into how a unified account can be achieved. In addition, the phenomena we investigate shed light on the role of prototypes in exemplar theory and on whether exemplar clouds are defined by a fixed radius or by a fixed number of nearest neighbors.

A number of recent theories of semantic representation propose two-way interaction of semantic and perceptual information. These theories are supported by a growing body of experiments that show widespread interactivity of semantic and perceptual levels of processing. In the current experiment, participants classified ambiguous colors into one of two color categories. The ambiguous colors were presented either as a color patch (no semantic context), as an icon representing an object that is strongly associated with one of the color options, or as a word referring to such an object. Although the iconic and lexical contexts were incidental and irrelevant to the color categorization task, participants ’ responses were consistently biased toward the context color. These results extend previous findings by showing that lexical contexts, as well as iconic contexts influence color categorization.

Abstract not found

Two experiments investigated the mechanism by which listeners adjust their interpretation of accented speech that is similar to a regional dialect of American English. Only a subset of the vowels of English (the front vowels) were shifted during adaptation, which consisted of listening to a 20-min segment of the “Wizard of Oz. ” Compared to a baseline (unadapted) condition, listeners showed significant adaptation to the accented speech, as indexed by increased word judgments on a lexical decision task. Adaptation also generalized to test words that had not been presented in the accented passage but that contained the shifted vowels. A control experiment showed that the adaptation effect was specific to the direction of the shift in the vowel space and not to a general relaxation of the criterion for what constitutes a good exemplar of the accented vowel category. Taken together, these results provide evidence for a context-specific vowel adaptation mechanism that enables a listener to adjust to the dialect of a particular talker.

Available online at www.sciencedirect.com

Word recognition is a balancing act: listeners must be sensitive to phonetic detail to avoid confusing similar words, yet, at the same time, be flexible enough to adapt to phonetically variable pronunciations, such as those produced by speakers of different dialects or by non-native speakers. Recent work has demonstrated that young toddlers are sensitive to phonetic detail during word recognition; pronunciations that deviate from the typical phonological form lead to a disruption of processing. However, it is not known whether young word learners show the flexibility that is characteristic of adult word recognition. The present study explores whether toddlers can adapt to artificial accents in which there is a vowel category shift with respect to the native language. Eighteen–twenty-month-olds heard mispronunciations of familiar words (e.g. vowels were shifted from [a] to [æ]: ‘dog ’ pronounced as ‘dag’). In test, toddlers were tolerant of mispronunciations if they had recently been exposed to the same vowel shift, but not if they had been exposed to standard pronunciations or other vowel shifts. The effects extended beyond particular items heard in exposure to words sharing the same vowels. These results indicate that, like adults, toddlers show flexibility in their interpretation of phonological detail. Moreover, they suggest that effects of top-down knowledge on the reinterpretation of phonological detail generalize across the phono-lexical system.

Purpose: Deficits in phonological working memory and deficits in phonological processing have both been considered potential explanatory factors in specific language impairment (SLI). Manipulations of the lexicality and phonotactic frequency of nonwords enable contrasting predictions to be derived from these hypotheses. Method: Eighteen typically developing (TD) children and 18 children with SLI completed an assessment battery that included tests of language ability, nonverbal intelligence, and two nonword repetition tests that varied in lexicality and frequency. Results: Repetition accuracy showed that children with SLI were unimpaired for short and simple high-lexicality nonwords, whereas clear impairments were shown for all low-lexicality nonwords. For low-lexicality nonwords, greater repetition accuracy was seen for nonwords constructed from high over low-frequency phoneme sequences. Children with SLI made the same proportion of errors that substituted a nonsense syllable for a lexical item as TD children, and this was stable across nonword length. Conclusions: The data show support for a phonological processing deficit in children with SLI, where long-term lexical and sublexical phonological knowledge mediate the interpretation of nonwords. However, the data also suggest that while phonological processing may provide a key explanation of SLI, a full account is likely to be multifaceted.