On predicting others’ words: Electrophysiological evidence of prediction in speech production

The present study investigated whether lexical processes that occur when we name objects can also be observed when an interaction partner is naming objects. We compared the behavioral and electrophysiological responses of participants performing a conditional go/no-go picture naming task in two different conditions: individually and jointly with a confederate participant. To obtain an index of lexical processing, we manipulated lexical frequency, so that half of the pictures had corresponding names of high-frequency and the remaining half had names of low-frequency. Color cues determined whether participants should respond, whether their task-partner should respond, or whether nobody should respond. Behavioral and ERP results showed that participants engaged in lexical processing when it was their turn to respond. Crucially, ERP results on no-go trials revealed that participants also engaged in lexical processing when it was their partner’s turn to act. In addition, ERP results showed increased response inhibition selectively when it was the partner’s turn to act. These findings provide evidence for the claim that listeners generate predictions about speakers’ utterances by relying on their own action production system.     

Dynamic action units slip in speech production errors

In the past, the nature of the compositional units proposed for spoken language has largely diverged from the types of control units pursued in the domains of other skilled motor tasks. A classic source of evidence as to the units structuring speech has been patterns observed in speech errors--"slips of the tongue". The present study reports, for the first time, on kinematic data from tongue and lip movements during speech errors elicited in the laboratory using a repetition task. Our data are consistent with the hypothesis that speech production results from the assembly of dynamically defined action units--gestures--in a linguistically structured environment. The experimental results support both the presence of gestural units and the dynamical properties of these units and their coordination. This study of speech articulation shows that it is possible to develop a principled account of spoken language within a more general theory of action.     

Constraints upon word substitution speech errors

We explore the features of a corpus of naturally occurring word substitution speech errors. Words are replaced by more imageable competitors in semantic substitution errors but not in phonological substitution errors. Frequency effects in these errors are complex and the details prove difficult for any model of speech production. We argue that word frequency mainly affects phonological errors. Both semantic and phonological substitutions are constrained by phonological and syntactic similarity between the target and intrusion. We distinguish between associative and shared-feature semantic substitutions. Associative errors originate from outside the lexicon, while shared-feature errors arise within the lexicon and occur when particular properties of the targets make them less accessible than the intrusion. Semantic errors arise early while accessing lemmas from a semantic-conceptual input, while phonological errors arise late when accessing phonological forms from lemmas. Semantic errors are primarily sensitive to the properties of the semantic field involved, whereas phonological errors are sensitive to phonological properties of the targets and intrusions.     

The production of determiners: evidence from French.

In numerous languages determiner forms depend not only on semantic information but also on several other kinds of information, such as the grammatical gender of the controlling noun or the phonological properties of the context. In the present research we contrasted two possible accounts of determiner retrieval: one in which every type of required information is bundled into a unitized representation for determiner retrieval and one in which each type of information can individually activate determiner forms. These alternative hypotheses were investigated in three experiments in which native speakers of French named pictures with simple [determiner + noun] or complex [determiner + adjective + noun] noun phrases. In the experiments, the properties of the contextual cues that drive the retrieval of the determiner were manipulated - for example, we manipulated the number of determiner forms that are compatible with a given grammatical gender and the number of grammatical genders that a given determiner form can be used with. Neither hypothesis can fully account for the results of the three experiments. However, a hybrid hypothesis that combines the principal features of the two hypotheses provides a good account of the data.     

The link between speech perception and production is phonological and abstract: evidence from the shadowing task

This study reports a shadowing experiment, in which one has to repeat a speech stimulus as fast as possible. We tested claims about a direct link between perception and production based on speech gestures, and obtained two types of counterevidence. First, shadowing is not slowed down by a gestural mismatch between stimulus and response. Second, phonetic detail is more likely to be imitated in a shadowing task if it is phonologically relevant. This is consistent with the idea that speech perception and speech production are only loosely coupled, on an abstract phonological level.     

Neighbourhood effects on error rates in speech production

Models of speech production differ on whether phonological neighbourhoods should affect processing, and on whether effects should be facilitatory or inhibitory. Inhibitory effects of large neighbourhoods have been argued to underlie apparent anti-frequency effects, whereby high-frequency default features are more prone to mispronunciation errors than low-frequency nondefault features. Data from the original SLIPs experiments that found apparent anti-frequency effects are analysed for neighbourhood effects. Effects are facilitatory: errors are significantly less likely for words with large numbers of neighbours that share the characteristic that is being primed for error ("friends"). Words in the neighbourhood that do not share the target characteristic ("enemies") have little effect on error rates. Neighbourhood effects do not underlie the apparent anti-frequency effects. Implications for models of speech production are discussed.     

Additivity of semantic and phonological effects: Evidence from speech production in Mandarin

A number of previous studies using picture–word interference (PWI) tasks conducted with speakers of Western languages have demonstrated non-additive effects of semantic and form overlap between pictures and words, which may indicate underlying non-discrete processing stages in lexical retrieval. The present study used Mandarin speakers and presented Chinese characters as distractors. In two experiments, we crossed semantic relatedness with “pure” phonological (i.e., orthographically unrelated) relatedness and found statistically additive effects. In a third experiment, semantic relatedness was crossed with orthographic overlap (phonological overlap was avoided), and once again we found an additive pattern. The results are discussed with regard to possible cross-linguistic differences between Western and non-Western languages in terms of phonological encoding, as well as concerning the locus of relatedness effects in PWI tasks.     

Phonotactic probability influences speech production

Speakers are faster and more accurate at processing certain sound sequences within their language. Does this reflect the fact that these sequences are frequent or that they are phonetically less complex (e.g., easier to articulate)? It has been difficult to contrast these two factors given their high correlation in natural languages. In this study, participants were exposed to novel phonotactic constraints de-correlating complexity and frequency by subjecting the same phonological structure to varying degrees of probabilistic constraint. Participants' behavior was sensitive to variations in frequency, demonstrating that phonotactic probability influences speech production independent of phonetic complexity.     

Effects of syllable structure in aphasic errors: implications for a new model of speech production

Current models of word production assume that words are stored as linear sequences of phonemes which are structured into syllables only at the moment of production. This is because syllable structure is always recoverable from the sequence of phonemes. In contrast, we present theoretical and empirical evidence that syllable structure is lexically represented. Storing syllable structure would have the advantage of making representations more stable and resistant to damage. On the other hand, re-syllabifications affect only a minimal part of phonological representations and occur only in some languages and depending on speech register. Evidence for these claims comes from analyses of aphasic errors which not only respect phonotactic constraints, but also avoid transformations which move the syllabic structure of the word further away from the original structure, even when equating for segmental complexity. This is true across tasks, types of errors, and, crucially, types of patients. The same syllabic effects are shown by apraxic patients and by phonological patients who have more central difficulties in retrieving phonological representations. If syllable structure was only computed after phoneme retrieval, it would have no way to influence the errors of phonological patients. Our results have implications for psycholinguistic and computational models of language as well as for clinical and educational practices.     

Perceptual Bias in Speech Error Data Collection: Insights from Spanish Speech Errors

This paper studies the reliability and validity of naturalistic speech errors as a tool for language production research. Possible biases when collecting naturalistic speech errors are identified and specific predictions derived. These patterns are then contrasted with published reports from Germanic languages (English, German and Dutch) and one Romance language (Spanish). Unlike findings in the Germanic languages, Spanish speech errors show many patterns which run contrary to those expected from bias: (1) more phonological errors occur between words than within word; (2) word-initial consonants are less likely to participate in errors than word-medial consonants, (3) errors are equally likely in stressed and in unstressed syllables, (4) perseverations are more frequent than anticipations, and (5) there is no trace of a lexical bias. We present a new corpus of Spanish speech errors collected by many theoretically naïve observers (whereas the only corpus available so far was collected by two highly trained theoretically informed observers), give a general overview of it, and use it to replicate previous reports. In spite of the different susceptibility of these methods to bias, results were remarkably similar in both corpora and again contrary to predictions from bias. As a result, collecting speech errors “in the wild” seems to be free of bias to a reasonable extent even when using a multiple-collector method. The observed contrasting patterns between Spanish and Germanic languages arise as true cross-linguistic differences.     

Phonological abstraction in processing lexical-tone variation: evidence from a learning paradigm

There is a growing consensus that the mental lexicon contains both abstract and word-specific acoustic information. To investigate their relative importance for word recognition, we tested to what extent perceptual learning is word specific or generalizable to other words. In an exposure phase, participants were divided into two groups; each group was semantically biased to interpret an ambiguous Mandarin tone contour as either tone1 or tone2. In a subsequent test phase, the perception of ambiguous contours was dependent on the exposure phase: Participants who heard ambiguous contours as tone1 during exposure were more likely to perceive ambiguous contours as tone1 than participants who heard ambiguous contours as tone2 during exposure. This learning effect was only slightly larger for previously encountered than for not previously encountered words. The results speak for an architecture with prelexical analysis of phonological categories to achieve both lexical access and episodic storage of exemplars.     

Cue integration with categories: Weighting acoustic cues in speech using unsupervised learning and distributional statistics

During speech perception, listeners make judgments about the phonological category of sounds by taking advantage of multiple acoustic cues for each phonological contrast. Perceptual experiments have shown that listeners weight these cues differently. How do listeners weight and combine acoustic cues to arrive at an overall estimate of the category for a speech sound? Here, we present several simulations using a mixture of Gaussians models that learn cue weights and combine cues on the basis of their distributional statistics. We show that a cue‐weighting metric in which cues receive weight as a function of their reliability at distinguishing phonological categories provides a good fit to the perceptual data obtained from human listeners, but only when these weights emerge through the dynamics of learning. These results suggest that cue weights can be readily extracted from the speech signal through unsupervised learning processes.     

Inner speech slips exhibit lexical bias, but not the phonemic similarity effect

Inner speech, that little voice that people often hear inside their heads while thinking, is a form of mental imagery. The properties of inner speech errors can be used to investigate the nature of inner speech, just as overt slips are informative about overt speech production. Overt slips tend to create words (lexical bias) and involve similar exchanging phonemes (phonemic similarity effect). We examined these effects in inner and overt speech via a tongue-twister recitation task. While lexical bias was present in both inner and overt speech errors, the phonemic similarity effect was evident only for overt errors, producing a significant overtness by similarity interaction. We propose that inner speech is impoverished at lower (featural) levels, but robust at higher (phonemic) levels.     

Perception of speech and nonspeech stimuli by children with and without reading disability and attention deficit hyperactivity disorder

The auditory temporal deficit hypothesis predicts that children with reading disability (RD) will exhibit deficits in the perception of speech and nonspeech acoustic stimuli in discrimination and temporal ordering tasks when the interstimulus interval (ISI) is short. Initial studies testing this hypothesis did not account for the potential presence of attention deficit hyperactivity disorder (ADHD). Temporal order judgment and discrimination tasks were administered to children with (1) RD/no-ADHD (n=38), (2) ADHD (n=29), (3) RD and ADHD (RD/ADHD; n=32), and (4) no impairment (NI; n=43). Contrary to predictions, children with RD showed no specific sensitivity to ISI and performed worse relative to children without RD on speech but not nonspeech tasks. Relationships between perceptual tasks and phonological processing measures were stronger and more consistent for speech than nonspeech stimuli. These results were independent of the presence of ADHD and suggest that children with RD have a deficit in phoneme perception that correlates with reading and phonological processing ability. (c) 2002 Elsevier Science (USA).     

How scenes containing visual errors affect speech fluency in young and older adults

We compared young and older adults’ speech during an error detection task, with some pictures containing visual errors and anomalies and other pictures error-free. We analyzed three disfluency types: mid-phrase speech fillers (e.g., It’s a little, um, girl), repetitions (e.g., He’s trying to catch the- the birds), and repairs (e.g., She- you can see her legs). Older adults produced more mid-phrase fillers than young adults only when describing pictures containing errors. These often reflect word retrieval problems and represent clear disruptions to fluency, so this interaction indicates that the need to form and maintain representations of novel information can specifically compromise older adults’ speech fluency. Overall, older adults produced more repetitions and repairs than young adults, regardless of picture type, indicating general age-related increases in these disfluencies. The obtained patterns are discussed in the context of the Transmission Deficit Hypothesis and other approaches to age-related changes in speech fluency.     

Lexical and post-lexical phonological representations in spoken production

Theories of spoken word production generally assume a distinction between at least two types of phonological processes and representations: lexical phonological processes that recover relatively arbitrary aspects of word forms from long-term memory and post-lexical phonological processes that specify the predictable aspects of phonological representations. In this work we examine the spoken production of two brain-damaged individuals. We use their differential patterns of accuracy across the tasks of spoken naming and repetition to establish that they suffer from distinct deficits originating fairly selectively within lexical or post-lexical processes. Independent and detailed analyses of their spoken productions reveal contrasting patterns that provide clear support for a distinction between two types of phonological representations: those that lack syllabic and featural information and are sensitive to lexical factors such as lexical frequency and neighborhood density, and those that include syllabic and featural information and are sensitive to detailed properties of phonological structure such as phoneme frequency and syllabic constituency.     

Picture–word interference reveals inhibitory effects of syllable frequency on lexical selection

While previous research has shown that high syllable frequency can facilitate speech production at the level of phonological/phonetic encoding, little is known about its influence on prephonological processes, specifically lexical selection. The current study used a picture–word interference (PWI) task to (a) shed light on the stages of lexical access where syllable frequency is relevant, and (b) inform as to whether lexical selection is accomplished via competition among activated word options. Participants named pictures whose names had high-frequency (HF) and low-frequency (LF) first syllables while ignoring phonologically related (same first syllable) or unrelated distractor words that were presented simultaneously. Word frequency was also manipulated, as half of the targets were HF words, and half were LF words. Results revealed inhibitory syllable frequency effects in all conditions, such that targets with HF first syllables were named more slowly than targets with LF first syllables. However, inhibitory syllable frequency effects were exacerbated in conditions thought to reflect heightened lexical competition, specifically in the presence of phonologically related distractors and for targets with low word frequency. These findings reveal novel evidence for first-syllable frequency effects on lexical selection and offer further support for models proposing delays at lexical selection due to activation of nontarget competitors.     

Set size and repetition in the picture--word interference paradigm: implications for models of naming

Caramazza and Costa (Cognition 75 (2000) B51) reported results which demonstrate that a semantically related word distractor interferes in picture naming even when it is not in the response set and there is no possibility for mediated interference. They interpreted the results to be problematic for the model of lexical access proposed by Levelt, Roelofs, and Meyer (Behavioral and Brain Sciences 22 (1999) 1). Roelofs (Cognition 80 (2001, this issue 283--90)) argues that those results are not inconsistent with Levelt et al.'s model when certain new assumptions about the mechanism of lexical selection are considered. Here we show that even with these assumptions the model still makes the wrong predictions. We report new results which demonstrate that the semantic interference and facilitation effects that are obtained respectively in the basic-level and category-level naming variants of the picture-word interference paradigm are not the result of response set size and response repetitions.     

Hemispheric asymmetries in semantic processing: evidence from false memories for ambiguous words

We used fMRI to investigate competition during language production in two word production tasks: object naming and color naming of achromatic line drawings. Generally, fMRI activation was higher for color naming. The line drawings were followed by a word (the distractor word) that referred to either the object, a related object, or an unrelated object. The effect of the distractor word on the BOLD response was qualitatively different for the two tasks. The activation pattern suggests two different kinds of competition during lexical retrieval: (1) Task-relevant responses (e.g., red in color naming) compete with task-irrelevant responses (i.e., the object’s name). This competition effect was dominant in prefrontal cortex. (2) Multiple task-relevant responses (i.e., target word and distractor word) compete for selection. This competition effect was dominant in ventral temporal cortex. This study provides further evidence for the distinct roles of frontal and temporal cortex in language production, while highlighting the effects of competition, albeit from different sources, in both regions.