Search

Purpose: The cross-modal picture-word interference task is used to examine contextual effects on spoken-word production. Previous work has documented lexical-phonological interference in children with specific language impairment (SLI) when a related distractor (e.g., bell) occurs prior to a picture to be named (e.g., a bed). In the current study, the authors examined whether interference also arises with nonwords as distractors., Method: In Study 1, children with SLI (N = 20; ages 7;1 [years;months] to 11;0) and age-matched controls named pictures accompanied by (a) phonologically related nonwords, (b) unrelated nonwords, or (c) the word go (baseline). Stimulus asynchrony (SA) varied across blocks with distractors occurring prior to (-300 ms, -100 ms) or after (+100 ms, +300 ms) the pictures. In Study 2, a cross-sectional sample of children (N = 48, 5;3 to 10;9) and adults (N = 16) performed the same task., Results: Child and adult control participants showed phonological priming (not interference) at early and late SAs, whereas children with SLI showed priming only at late SAs. Effect sizes correlated with language skills (Clinical Evaluation of Language Fundamentals-Fourth Edition scores; Semel, Wiig, & Secord, 2003). In the cross-sectional sample, anticipatory priming at SA -300 varied with age, with larger effects in older children., Conclusions: Children with SLI utilize phonological information when it is available just in time for word production but fail to anticipate upcoming stimuli. Poor anticipatory processing may adversely affect language fluency in children with SLI.

In this paper we present a self-organizing neural network model of early lexical development called DevLex. The network consists of two self-organizing maps (a growing semantic map and a growing phonological map) that are connected via associative links trained by Hebbian learning. The model captures a number of important phenomena that occur in early lexical acquisition by children, as it allows for the representation of a dynamically changing linguistic environment in language learning. In our simulations, DevLex develops topographically organized representations for linguistic categories over time, models lexical confusion as a function of word density and semantic similarity, and shows age-of-acquisition effects in the course of learning a growing lexicon. These results match up with patterns from empirical research on lexical development, and have significant implications for models of language acquisition based on self-organizing neural networks.

Eleven children with early focal lesions were compared with 70 age-matched controls to assess their performance in repeating non-words, in learning new words, and in immediate serial recall, a triad of abilities that are believed to share a dependence on serial ordering mechanisms (e.g.,; ). Results for the experimental group were also compared with other assessments previously reported for the same children by. The children with brain injury showed substantial impairment relative to controls in the experimental tasks, in contrast with relatively unimpaired performance on measures of vocabulary and non-verbal intelligence. The relationships between word learning, non-word repetition, and immediate serial recall were similar to those observed in several other populations. These results support previous reports that there are persistent processing impairments following early brain injury, despite developmental plasticity. They also suggest that word learning, non-word repetition, and immediate serial recall may be relatively demanding tasks, and that their relationship is a fundamental aspect of the cognitive system.

Much of the power of neural network modeling for language use and acquisition derives from a reliance on statistical regularities implicit in the phonological properties of words. Researchers have devised several methods for representing the phonology of words, but these methods are often either unable to represent realistically sized lexicons or inadequate in the ways they represent individual words. In this paper, we present a new phonological pattern generator (PatPho) that allows connectionist modelers to derive accurate phonological representations of the English lexicon. PatPho not only generates phonological patterns that can scale up to realistically sized lexicons, but also accurately and parsimoniously captures the similarity structures of the phonology of monosyllabic and multisyllabic words.

The authors investigated the on-line relationship between overt articulation and the central processes of speech production. In 2 experiments manipulating the timing of Stroop interference in color naming, the authors found that naming behavior can shift between exhibiting a staged or cascaded mode of processing, depending on task demands: An effect of Stroop interference on naming durations arose only when there was increased pressure for speeded responding. In a simple connectionist model of information processing applied to color naming, the authors accounted for the current results by manipulating a single parameter, termed "gain," modulating the rate of information accrual within the network. Results are discussed in relation to mechanisms of strategic control and the link between cognition and action.

Second through 6th graders were presented with nonword primes (orthographic, pseudohomophone, and control) and target words displayed for durations (30 and 60 ms) that were brief enough to prevent complete processing. Word reading skills were assessed by 3 word and nonword naming tasks. Good readers exhibited more orthographic priming than poor readers at both durations and more pseudohomophone priming at the short duration only. This suggests that good readers activate letter and phonemic information more efficiently than poor readers. Good readers also exhibited an equal amount of priming at both durations, whereas poor readers showed greater priming at the longer duration. This suggests that activation was not under strategic control. Finally, priming was reliable for both high- and low-frequency targets. This suggests that readers activate consistent information regardless of target word characteristics. Thus, quick, automatic, and general activation of orthographic and phonological information in skilled readers results from the precision and redundancy of their lexical representations.

In a recent issue of this journal, Pinker and Prince (1988) and Lachter and Bever (1988) presented detailed critiques of Rumelhart and McClelland's (1986) connectionist model of the child's learning of the phonological form of the English past tense. In order to address these criticisms, a new connectionist model was constructed using the back-propagation algorithm, a larger input corpus, a fuller paradigm, and a new phonological representation. This new implementation successfully addressed the criticisms of the phonological representation used by Rumelhart and McClelland. It did a much better job of learning the past tense using a fuller input set with realistic frequencies of occurrence. Ancillary simulations using the same network were able to deal with the homonymy problem and the generation of forms like "ated" from "ate". The one feature not provided by the new model was a way of modeling early correct production of irregular forms. The success of the new model can be used to help clarify the extent to which the published critiques apply to a particular connectionist implementation as opposed to fundamental principles underlying the broader connectionist conceptualization.