Willem J. M. Levelt

A theory of lexical access in speech production

Preparing words in speech production is normally a fast and accurate process. We generate them two or three per second in fluent conversation; and overtly naming a clear picture of an object can easily be initiated within 600 msec after picture onset. The underlying process, however, is exceedingly complex. The theory reviewed in this target article analyzes this process as staged and feedforward. After a first stage of conceptual preparation, word generation proceeds through lexical selection, morphological and phonological encoding, phonetic encoding, and articulation itself. In addition, the speaker exerts some degree of output control, by monitoring of self-produced internal and overt speech. The core of the theory, ranging from lexical selection to the initiation of phonetic encoding, is captured in a computational model, called weaver ++. Both the theory and the computational model have been developed in interaction with reaction time experiments, particularly in picture naming or related word production paradigms, with the aim of accounting for the real-time processing in normal word production. A comprehensive review of theory, model, and experiments is presented. The model can handle some of the main observations in the domain of speech errors (the major empirical domain for most other theories of lexical access), and the theory opens new ways of approaching the cerebral organization of speech production by way of high-temporal-resolution imaging.

The spatial and temporal signatures of word production components

This paper presents the results of a comprehensive meta-analysis of the relevant imaging literature on word production (82 experiments). In addition to the spatial overlap of activated regions, we also analyzed the available data on the time course of activations. The analysis specified regions and time windows of activation for the core processes of word production: lexical selection, phonological code retrieval, syllabification, and phonetic/articulatory preparation. A comparison of the word production results with studies on auditory word/non-word perception and reading showed that the time course of activations in word production is, on the whole, compatible with the temporal constraints that perception processes impose on the production processes they affect in picture/word interference paradigms.

Monitoring and self-repair in speech.

Abstract Making a self-repair in speech typically proceeds in three phases. The first phase involves the monitoring of ones own speech and the interruption of the flow of speech when trouble is detected. From an analysis of 959 spontaneous self-repairs it appears that interrupting follows detection promptly, with the exception that correct words tend to be completed. Another finding is that detection of trouble improves towards the end of constituents. The second phase is characterized by hesitation, pausing, but especially the use of so-called editing terms. Which editing term is used depends on the nature of the speech trouble in a rather regular fashion: Speech errors induce other editing terms than words that are merely inappropriate, and trouble which is detected quickly by the speaker is preferably signalled by the use of ‘uh’. The third phase consists of making the repair proper. The linguistic well-formedness of a repair is not dependent on the speakers respecting the integrity of constituents, but on the structural relation between original utterance and repair. A bi-conditional well-formedness rule links this relation to a corresponding relation between the conjuncts of a coordination. It is suggested that a similar relation holds also between question and answer. In all three cases the speaker respects certain structural commitments derived from an original utterance. It was finally shown that the editing term plus the first word of the repair proper almost always contain sufficient information for the listener to decide how the repair should be related to the original utterance. Speakers almost never produce misleading information in this respect. It is argued that speakers have little or no access to their speech production process; self-monitoring is probably based on parsing ones own inner or overt speech.

Word frequency effects in speech production: Retrieval of syntactic information and of phonological form

In 7 experiments the authors investigated the locus of word frequency effects in speech production. Experiment 1 demonstrated a frequency effect in picture naming that was robust over repetitions. Experiments 2, 3, and 7 excluded contributions from object identification and initiation of articulation. Experiments 4 and 5 investigated whether the effect arises in accessing the syntactic word (lemma) by using a grammatical gender decision task. Although a frequency effect was found, it dissipated under repeated access to a words gender. Experiment 6 tested whether the robust frequency effect arises in accessing the phonological form (lexeme) by having Ss translate words that produced homophones. Low-frequent homophones behaved like high-frequent controls, inheriting the accessing speed of their high-frequent homophone twins. Because homophones share the lexeme, not the lemma, this suggests a lexeme-level origin of the robust effect.

Tonal consonance and critical bandwidth

Firstly, theories are reviewed on the explanation of tonal consonance as the singular nature of tone intervals with frequency ratios corresponding with small integer numbers. An evaluation of these explanations in the light of some experimental studies supports the hypothesis, as promoted by von Helmholtz, that the difference between consonant and dissonant intervals is related to beats of adjacent partials. This relation was studied more fully by experiments in which subjects had to judge simple‐tone intervals as a function of test frequency and interval width. The results may be considered as a modification of von Helmholtzs conception and indicate that, as a function of frequency, the transition range between consonant and dissonant intervals is related to critical bandwidth. Simple‐tone intervals are evaluated as consonant for frequency differences exceeding this bandwidth. whereas the most dissonant intervals correspond with frequency differences of about a quarter of this bandwidth. On the base of ...

The Time Course of Lexical Access in Speech Production: A Study of Picture Naming

Nijmegen University Nijmegen, The Netherlands Lexical access in object naming involves the activation of a set oflexical candidates, the selection of the appropriate (or target) item, and the phonological encoding of that item. Two views of lexical access in naming are compared. From one view, the 2-stage theory, phonological activation follows selection of the target item and is restricted to that item. From the other view, which is most explicit in activation-spreading theories, all activated lexical candidates are phonologically activated to some extent. A series of experiments is reported in which subjects performed acoustic lexical decision during object naming at different stimulus-onset asynchronies. The experiments show semantic activation of lexical candidates and phonological activation of the target item, but no phonological activation of other semantically activated items. This supports the 2-stage view. More- over, a mathematical model embodying the 2-stage view is fully compatible with the lexical deci- sion data obtained at different stimulus-onset asynchronies. One of a speakers core skills is to lexicalize the concepts intended for expression. Lexicalization proceeds at a rate of two to three words per second in normal spontaneous speech, but doubling this rate is possible and not exceptional. The skill of lexicalizing a content word involves two components. The first one is to select the appropriate lexical item from among some tens of thousands of alternatives in the mental lexicon. The second one is to phonologically encode the selected item, that is, to retrieve its sound form, to create a phonological represen- tation for the item in its context, and to prepare its articulatory program. An extensive review of the literature on lexicalization can be found in Levelt (1989). This article addresses only one aspect of lexicalization, namely its time course. In particular, we examine whether the selection of an item and its phonologi- cal encoding can be considered to occur in two successive, non- overlapping stages. We acknowledge the invaluable contributions of John Nagengast and Johan Weustink, who programmed the computer-based experi- ments; ofGer Desserjer and Hans Fransen, who ran the experiments and assisted in data analysis; and of lnge Tarim, who provided graphi- cal assistance. We also acknowledge Gary Dells and Picnic Zwitser- loods detailed comments on an earlier version of this article, as well as the thorough comments of an anonymous reviewer. Herbert Schriefers is now at Freie Universit~it Berlin, Berlin, Federal Republic of Germany, and Thomas Pechmann is now at Universit~it des Saarlandes, Saarbriicken, Federal Republic of Germany. Correspondence concerning this article should be addressed to Wil- lem J. M. Levelt, Max-Planck-lnstitut for Psycholinguistik, Wundtlaan 1, NL-6525 XD Nijmegen, The Netherlands. 122 This is by no means a novel concept. One should rather say that it is the received view in the psycholinguistic literature (see especially Butterworth, 1980, 1989; Fromkin, 1971; Garrett, 1975, 1976, 1980; Kempen, 1977, 1978; Kempen & Huijbers, 1983; Levelt, 1983, 1989; Levelt & Maassen, 1981; Morton, 1969; Schriefers, Meyer, & Levelt, 1990). The first stage, lexical selection, makes available a semantically specified lexical item with its syntactic constraints. Kempen (1977, 1978) called this a lemma. Lemmas figure in grammatical encoding, specifically in the creation of syntactic frames. During the second stage, phonological encoding, phonological information is retrieved for each lemma. These phonological codes are used to create the articulatory plan for the utterance as a whole. Both Garrett (1976) and Kempen (1978), following Fry (1969), have stressed that the grammatical encoding and phonological encoding of an utterance normally run in parallel. Grammatical encoding, of which lexical selection is a proper part, is just slightly ahead of phonological encoding. The phonological encoding of a given item overlaps in time with the selection of a subsequent item. Only at the level of individual lexical items can one speak of successive stages. An items semantic-syntactic makeup is accessed and used before its phonological makeup becomes available. Garrett (1975, 1976) argued for this separation of stages on the basis of speech error data. He distinguished between two classes of errors, word exchanges and sound exchanges, and could show that these classes differ in distributional properties. Word exchanges occur between phrases and involve words of the same syntactic category (as in this spring has a seat in it). Sound exchanges typically involve different category words in the same phrase (as in heft lemisphere). Word exchanges are

Models of word production

Research on spoken word production has been approached from two angles. In one research tradition, the analysis of spontaneous or induced speech errors led to models that can account for speech error distributions. In another tradition, the measurement of picture naming latencies led to chronometric models accounting for distributions of reaction times in word production. Both kinds of models are, however, dealing with the same underlying processes: (1) the speakers selection of a word that is semantically and syntactically appropriate; (2) the retrieval of the words phonological properties; (3) the rapid syllabification of the word in context; and (4) the preparation of the corresponding articulatory gestures. Models of both traditions explain these processes in terms of activation spreading through a localist, symbolic network. By and large, they share the main levels of representation: conceptual/semantic, syntactic, phonological and phonetic. They differ in various details, such as the amount of cascading and feedback in the network. These research traditions have begun to merge in recent years, leading to highly constructive experimentation. Currently, they are like two similar knives honing each other. A single pair of scissors is in the making.

Spoken word production: A theory of lexical access

A core operation in speech production is the preparation of words from a semantic base. The theory of lexical access reviewed in this article covers a sequence of processing stages beginning with the speakers focusing on a target concept and ending with the initiation of articulation. The initial stages of preparation are concerned with lexical selection, which is zooming in on the appropriate lexical item in the mental lexicon. The following stages concern form encoding, i.e., retrieving a words morphemic phonological codes, syllabifying the word, and accessing the corresponding articulatory gestures. The theory is based on chronometric measurements of spoken word production, obtained, for instance, in picture-naming tasks. The theory is largely computationally implemented. It provides a handle on the analysis of multiword utterance production as well as a guide to the analysis and design of neuroimaging studies of spoken utterance production.

An MEG Study of Picture Naming

The purpose of this study was to relate a psycholinguistic processing model of picture naming to the dynamics of cortical activation during picture naming. The activation was recorded from eight Dutch subjects with a whole-head neuromagnetometer. The processing model, based on extensive naming latency studies, is a stage model. In preparing a pictures name, the speaker performs a chain of specific operations. They are, in this order, computing the visual percept, activating an appropriate lexical concept, selecting the target word from the mental lexicon, phonological encoding, phonetic encoding, and initiation of articulation. The time windows for each of these operations are reasonably well known and could be related to the peak activity of dipole sources in the individual magnetic response patterns. The analyses showed a clear progression over these time windows from early occipital activation, via parietal and temporal to frontal activation. The major specific findings were that (1) a region in the left posterior temporal lobe, agreeing with the location of Wernickes area, showed prominent activation starting about 200 msec after picture onset and peaking at about 350 msec, (i.e., within the stage of phonological encoding), and (2) a consistent activation was found in the right parietal cortex, peaking at about 230 msec after picture onset, thus preceding and partly overlapping with the left temporal response. An interpretation in terms of the management of visual attention is proposed.

Effects of semantic context in the naming of pictures and words

Two experiments investigated whether lexical retrieval for speaking can be characterized as a competitive process by assessing the effects of semantic context on picture and word naming in German. In Experiment 1 we demonstrated that pictures are named slower in the context of same-category items than in the context of items from various semantic categories, replicating findings by Kroll and Stewart (Journal of Memory and Language, 33 (1994) 149). In Experiment 2 we used words instead of pictures. Participants either named the words in the context of same- or different-category items, or produced the words together with their corresponding determiner. While in the former condition words were named faster in the context of same-category items than of different-category items, the opposite pattern was obtained for the latter condition. These findings confirm the claim that the interfering effect of semantic context reflects competition in the retrieval of lexical entries in speaking.