Mark S. Seidenberg

A Distributed, Developmental Model of Word Recognition and Naming

A parallel distributed processing model of visual word recognition and pronunciation is described. The model consists of sets of orthographic and phonological units and an interlevel of hidden units. Weights on connections between units were modified during a training phase using the back-propagation learning algorithm. The model simulates many aspects of human performance, including (a) differences between words in terms of processing difficulty, (b) pronunciation of novel items, (c) differences between readers in terms of word recognition skill, (d) transitions from beginning to skilled reading, and (e) differences in performance on lexical decision and naming tasks. The models behavior early in the learning phase corresponds to that of children acquiring word recognition skills. Training with a smaller number of hidden units produces output characteristic of many dyslexic readers. Naming is simulated without pronunciation rules, and lexical decisions are simulated without accessing word-level representations. The performance of the model is largely determined by three factors: the nature of the input, a significant fragment of written English; the learning rule, which encodes the implicit structure of the orthography in the weights on connections; and the architecture of the system, which influences the scope of what can be learned.

Understanding normal and impaired word reading: Computational principles in quasi-regular domains.

A connectionist approach to processing in quasi-regular domains, as exemplified by English word reading, is developed. Networks using appropriately structured orthographic and phonological representations were trained to read both regular and exception words, and yet were also able to read pronounceable nonwords as well as skilled readers. A mathematical analysis of a simplified system clarifies the close relationship of word frequency and spelling-sound consistency in influencing naming latencies. These insights were verified in subsequent simulations, including an attractor network that accounted for latency data directly in its time to settle on a response. Further analyses of the ability of networks to reproduce data on acquired surface dyslexia support a view of the reading system that incorporates a graded division of labor between semantic and phonological processes, and contrasts in important ways with the standard dual-route account.

The lexical nature of syntactic ambiguity resolution.

: Ambiguity resolution is a central problem in language comprehension. Lexical and syntactic ambiguities are standardly assumed to involve different types of knowledge representations and be resolved by different mechanisms. An alternative account is provided in which both types of ambiguity derive from aspects of lexical representation and are resolved by the same processing mechanisms. Reinterpreting syntactic ambiguity resolution as a form of lexical ambiguity resolution obviates the need for special parsing principles to account for syntactic interpretation preferences, reconciles a number of apparently conflicting results concerning the roles of lexical and contextual information in sentence processing, explains differences among ambiguities in terms of ease of resolution, and provides a more unified account of language comprehension than was previously available.

Computing the meanings of words in reading: cooperative division of labor between visual and phonological processes.

Are words read visually (by means of a direct mapping from orthography to semantics) or phonologically (by mapping from orthography to phonology to semantics)? The authors addressed this long-standing debate by examining how a large-scale computational model based on connectionist principles would solve the problem and comparing the models performance to peoples. In contrast to previous models, the present model uses an architecture in which meanings are jointly determined by the 2 components, with the division of labor between them affected by the nature of the mappings between codes. The model is consistent with a variety of behavioral phenomena, including the results of studies of homophones and pseudohomophones thought to support other theories, and illustrates how efficient processing can be achieved using multiple simultaneous constraints.

How Psychological Science Informs the Teaching of Reading

This monograph discusses research, theory, and practice relevant to how children learn to read English. After an initial overview of writing systems, the discussion summarizes research from developmental psychology on childrens language competency when they enter school and on the nature of early reading development. Subsequent sections review theories of learning to read, the characteristics of children who do not learn to read (i.e., who have developmental dyslexia), research from cognitive psychology and cognitive neuroscience on skilled reading, and connectionist models of learning to read. The implications of the research findings for learning to read and teaching reading are discussed. Next, the primary methods used to teach reading (phonics and whole language) are summarized. The final section reviews laboratory and classroom studies on teaching reading. From these different sources of evidence, two inescapable conclusions emerge: (a) Mastering the alphabetic principle (that written symbols are associated with phonemes) is essential to becoming proficient in the skill of reading, and (b) methods that teach this principle directly are more effective than those that do not (especially for children who are at risk in some way for having difficulty learning to read). Using whole-language activities to supplement phonics instruction does help make reading fun and meaningful for children, but ultimately, phonics instruction is critically important because it helps beginning readers understand the alphabetic principle and learn new words. Thus, elementary-school teachers who make the alphabetic principle explicit are most effective in helping their students become skilled, independent readers.

Semantic feature production norms for a large set of living and nonliving things

Semantic features have provided insight into numerous behavioral phenomena concerning concepts, categorization, and semantic memory in adults, children, and neuropsychological populations. Numerous theories and models in these areas are based on representations and computations involving semantic features. Consequently, empirically derived semantic feature production norms have played, and continue to play, a highly useful role in these domains. This article describes a set of feature norms collected from approximately 725 participants for 541 living (dog) and nonliving (chair) basic-level concepts, the largest such set of norms developed to date. This article describes the norms and numerous statistics associated with them. Our aim is to make these norms available to facilitate other research, while obviating the need to repeat the labor-intensive methods involved in collecting and analyzing such norms. The full set of norms may be downloaded from www.psychonomic.org/archive.

Pre- and postlexical loci of contextual effects on word recognition

The context in which a word occurs could influence either the actual decoding of the word or a postrecognition judgment of the relatedness of word and context. In this research, we investigated the loci of contextual effects that occur in lexical priming, when prime and target words are related along different dimensions. Both lexical decision and naming tasks were used because previous research had suggested that they are differentially sensitive to postlexical processing. Semantic and associative priming occurred with both tasks. Other facilitative contextual effects, due to syntactic relations between words, backward associations, or changes in the proportion of related items, occurred only with the lexical decision task. The results indicate that only associative and semantic priming facilitate the decoding of a target; the other effects are postlexical. The results are related to the different demands of the naming and lexical decision tasks, and to current models of word recognition.

On the nature and scope of featural representations of word meaning.

Behavioral experiments and a connectionist model were used to explore the use of featural representations in the computation of word meaning. The research focused on the role of correlations among features, and differences between speeded and untimed tasks with respect to the use of featural information. The results indicate that featural representations are used in the initial computation of word meaning (as in an attractor network), patterns of feature correlations differ between artifacts and living things, and the degree to which features are intercorrelated plays an important role in the organization of semantic memory. The studies also suggest that it may be possible to predict semantic priming effects from independently motivated featural theories of semantic relatedness. Implications for related behavioral phenomena such as the semantic impairments associated with Alzheimers disease (AD) are discussed.

Automatic access of the meanings of ambiguous words in context: Some limitations of knowledge-based processing ☆ ☆☆ ★ ★★

Abstract Five experiments are described on the processing of ambiguous words in sentences. Two classes of ambiguous words (noun-noun and noun-verb) and two types of context (priming and nonpriming) were investigated using a variable stimulus onset asynchrony (SOA) priming paradigm. Noun-noun ambiguities have two semantically unrelated readings that are nouns (e.g., PEN, ORGAN); noun-verb ambiguities have both noun and verb readings that are unrelated (e.g., TIRE, WATCH). Priming contexts contain a word highly semantically or associatively related to one meaning of the ambiguous word; nonpriming contexts favor one meaning of the word through other types of information (e.g., syntactic or pragmatic). In nonpriming contexts, subjects consistently access multiple meanings of words and select one reading within 200 msec. Lexical priming differentially affects the processing of subsequent noun-noun and noun-verb ambiguities, yielding selective access of meaning only in the former case. The results suggest that meaning access is an automatic process which is unaffected by knowledge-based (“top-down”) processing. Whether selective or multiple access of meaning is observed largely depends on the structure of the ambiguous word, not the nature of the context.

The time course of phonological code activation in two writing systems.

Abstract Models of visual word recognition differ in assumptions about the extent to which phonological information is used, and the processes by which it becomes available. These issues were examined in two studies of word recognition in two writing systems, English and Chinese, which are structured along different principles (alphabetic and logographic, respectively). The results indicate that in each writing system, a large pool of higher frequency words is recognized on a visual basis, without phonological mediation. Phonology only enters into the processing of lower frequency words. Thus, although there may be other differences among writing systems which influence processing, differences in the manner in which they represent phonology are not relevant to the recognition of common words. The results are consistent with a parallel interactive model of word recognition in which orthographic and phonological information are activated at different latencies.