Gary S. Dell

The retrieval of phonological forms in production: tests of predictions from a connectionist model

This article reviews a model of retrieval processes in language production that accounts for phonological speech-error data (Dell, 1986, Psychological Review, 93, 283–321). A distinction is drawn between empirical phenomena that are built into the model and phenomena that can be considered predictions from it. Data from experiments creating phonological speech errors are presented as tests of those predictions.

Stages of lexical access in language production

We describe two primary stages in the top-down process of lexical access in production, a stage of lemma access in which words are retrieved as syntactic-semantic entities, and a stage of phonological access in which the forms of the words are fleshed out. We suggest a reconciliation of modular and interactive accounts of these stages whereby modularity is traceable to the action of discrete linguistic rule systems, but interaction arises in the lexical network on which these rules operate. We also discuss the time-course of lexical access in multi-word utterances. We report some initial production priming explorations that support the hypothesis that lemmas are buffered in longer utterances before they are phonologically specified. Because such techniques provide a relatively direct way of assessing activation at the primary stages of lexical access they are an important new resource for the study of language production.

Stages in sentence production: An analysis of speech error data

The hypothesis that sentence production is organized into independent positional and functional stages is tested using speech error data. Contrary to predictions from the hypothesis it was found that sound misordering errors tend to create words and that word errors, such as substitutions and misorderings, tend to involve similar sounding words. In addition, it was found that incorrectly substituted words often show both a semantic and phonological relationship to the intended words. A proposal regarding the stages of production is developed that accounts for the results. It is assumed that information can leak between stages by way of the mental lexicon and cause the decision making at a given stage to be affected in a probabilistic manner by information from other stages.

Mediated and convergent lexical priming in language production: a comment on Levelt et al. (1991).

Levelt et al. (1991) argued that modular semantic and phonological stage theories of lexical access in language production are to be preferred over interactive spreading-activation theories (e.g., Dell, 1986). As evidence, they show no mediated semantic-phonological priming during picture naming: Retrieval of sheep primes goat, but the activation of goat is not transmitted to its phonological relative, goal. This research reconciles this result with spreading-activation theories and shows how the absence of mediated priming coexists with the convergent priming necessary to account for mixed semantic-phonological speech errors. The analysis leads to the proposal that the language-production system may best be characterized as globally modular but locally interactive.

Effect of Ambiguity and Lexical Availability on Syntactic and Lexical Production

Speakers only sometimes include the that in sentence complement structures like The coach knew (that) you missed practice. Six experiments tested the predictions concerning optional word mention of two general approaches to language production. One approach claims that language production processes choose syntactic structures that ease the task of creating sentences, so that words are spoken opportunistically, as they are selected for production. The second approach claims that a syntactic structure is chosen that is easiest to comprehend, so that optional words like that are used to avoid temporarily ambiguous, difficult-to-comprehend sentences. In all experiments, speakers did not consistently include optional words to circumvent a temporary ambiguity, but they did omit optional words (the complementizer that) when subsequent material was either repeated (within a sentence) or prompted with a recall cue. The results suggest that speakers choose syntactic structures to permit early mention of available material and not to circumvent disruptive temporary ambiguities.

Anterior temporal involvement in semantic word retrieval: voxel-based lesion-symptom mapping evidence from aphasia

Analysis of error types provides useful information about the stages and processes involved in normal and aphasic word production. In picture naming, semantic errors (horse for goat) generally result from something having gone awry in lexical access such that the right concept was mapped to the wrong word. This study used the new lesion analysis technique known as voxel-based lesion-symptom mapping to investigate the locus of lesions that give rise to semantic naming errors. Semantic errors were obtained from 64 individuals with post-stroke aphasia, who also underwent high-resolution structural brain scans. Whole brain voxel-based lesion-symptom mapping was carried out to determine where lesion status predicted semantic error rate. The strongest associations were found in the left anterior to mid middle temporal gyrus. This area also showed strong and significant effects in further analyses that statistically controlled for deficits in pre-lexical, conceptualization processes that might have contributed to semantic error production. This study is the first to demonstrate a specific and necessary role for the left anterior temporal lobe in mapping concepts to words in production. We hypothesize that this role consists in the conveyance of fine-grained semantic distinctions to the lexical system. Our results line up with evidence from semantic dementia, the convergence zone framework and meta-analyses of neuroimaging studies on word production. At the same time, they cast doubt on the classical linkage of semantic error production to lesions in and around Wernickes area.

Structural Priming as Implicit Learning: A Comparison of Models of Sentence Production

Structural priming reflects a tendency to generalize recently spoken or heard syntactic structures to different utterances. We propose that it is a form of implicit learning. To explore this hypothesis, we developed and tested a connectionist model of language production that incorporated mechanisms previously used to simulate implicit learning. In the model, the mechanism that learned to produce structured sequences of phrases from messages also exhibited structural priming. The ability of the model to account for structural priming depended on representational assumptions about the nature of messages and the relationship between comprehension and production. Modeling experiments showed that comprehension-based representations were important for the models generalizations in production and that nonatomic message representations allowed a better fit to existing data on structural priming than traditional thematic-role representations.

Origins of Paraphasias in Deep Dysphasia: Testing the Consequences of a Decay Impairment to an Interactive Spreading Activation Model of Lexical Retrieval

This study investigates an account of atypical error patterns within the framework of an interactive spreading activation model. Martin and Saffran (1992) described a patient, NC, whose error pattern was unusual for the occurrence of higher rates of form-related than meaning-related word substitutions in naming and the production of semantic errors in repetition. They proposed that NCs error pattern could be accounted for by a pathologically rapid decay of primed nodes in the semantic-lexical-phonological network that shifts the probabilities of error outcome in lexical retrieval. In the present study, Martin and Saffrans account was tested and supported in a series of simulations that reproduce essential features of NCs lexical error pattern in naming and repetition. Also described here are the results of a longitudinal study of NCs naming and repetition, which revealed a shift in relative lexical error rates toward a qualitatively normal pattern. This change in error pattern was simulated by assuming that recovery reflects resolution of the rapid decay rate toward normal levels. The patient data and computational studies are discussed in terms of their significance for the understanding of aphasic impairments and their implications for models of lexical retrieval.

The activation of antecedent information during the processing of anaphoric reference in reading

The time course of the retrieval of antecedent information during the processing of anaphoric reference was examined in four experiments. In each experiment subjects read paragraphs that appeared one word at a time on a CRT screen. At unexpected times, they were given a single-word recognition test. Response times and error rates for these tests indicated that both the referent (e.g., car) of a superordinate anaphor (e.g., the vehicle) and concepts in the same proposition as the referent become activated as early as 250 milliseconds after the anaphor is read. The referent remains activated as the sentence is read, but the activation of other concepts dies away. The results are interpreted as support for the proposal that antecedent information is initially retrieved in the form of propositions, but only certain concepts from those propositions, those that are important for establishing text structure, remain activated.

The dark side of incremental learning: A model of cumulative semantic interference during lexical access in speech production

Naming a picture of a dog primes the subsequent naming of a picture of a dog (repetition priming) and interferes with the subsequent naming of a picture of a cat (semantic interference). Behavioral studies suggest that these effects derive from persistent changes in the way that words are activated and selected for production, and some have claimed that the findings are only understandable by positing a competitive mechanism for lexical selection. We present a simple model of lexical retrieval in speech production that applies error-driven learning to its lexical activation network. This model naturally produces repetition priming and semantic interference effects. It predicts the major findings from several published experiments, demonstrating that these effects may arise from incremental learning. Furthermore, analysis of the model suggests that competition during lexical selection is not necessary for semantic interference if the learning process is itself competitive.