The effect of shared structure and content on reading nonwords: evidence for a CV skeleton.

Four experiments examined the effect of shared skeletal structure versus content overlap on naming printed nonwords. Experiments 1-2 compared priming among nonwords sharing either skeletal structure and content (e.g., dus-DUS) or structure alone (e.g., pid-BAF) with controls that differed from the target in the number of skeleton slots (e.g., pid-BAF vs. plid-BAF). Conversely, in Experiments 3-4, same-versus different-structure primes contrasted only in the ordering of CV skeletal slots (e.g., fap-DUS vs. ift-DUS). Priming effects were modulated by shared content and skeletal similarity. The sensitivity of skeletal priming to the abstract arrangement of consonants and vowels suggests that skeletal representations assign distinct slots for consonants and vowels. Readers' sensitivity to skeletal structure in nonword identification indicates that assembled phonological representations are constrained by linguistic knowledge.

The temporal unfolding of local acoustic information and sentence context.

The purpose of this study was to investigate the temporal unfolding of local acoustic information and sentence context using both cross-modal interference (CMI) and word-monitoring tasks. The timing of sentence context effects have important theoretical implications for models of language processing (e.g., initial context independence vs. initial interaction). Yet, different tasks tend to yield different results. For both experiments, stimuli from an acoustically manipulated "goat-to-coat" continuum were embedded in sentences whose interpretation was biased toward either "goat" or "coat." In experiment 1 (CMI), the primary task was listening to sentences for comprehension; the interference task was a word/nonword decision to an unrelated visual probe that appeared at one of three positions within the sentence. Results showed immediate effects of the acoustic manipulation, but only delayed effects of sentence context. These results were interpreted to indicate that phonological processing is initially context-independent but is followed by rapid context integration. Experiment 2 used a word-monitoring task: Response times were significantly longer when sentence context was incongruent with the monitoring target, showing an immediate effect of context. The apparently contradictory results of the two experiments together support an account of language processing in which phoneme categorization is initially independent of sentence context unless an explicit judgment about the identity of the target is required.

Categorization of ambiguous sentences as a function of a changing prosodic parameter: a dynamical approach.

The present study investigates the dynamics of changes in interpretation of ambiguous sentences. The sentences used had alternative interpretations with different surface structures (bracketing). Continua were created by systematic manipulation of prosodic cues (relative foot duration), resulting in stimulus sentences that spanned the range between the two interpretations. Continua were presented to subjects who were requested to indicate as quickly as possible which meaning they perceived. The pattern of responses and response times revealed the presence of hysteresis. That is, when the values of prosodic parameters are congruent with both interpretations, the individual's recent history decides which meaning will be perceived. Thus, we can treat this categorization process as a transition from an initially stable meaning that loses stability with variations in prosody (our control parameter). The same underlying dynamics have been observed in studies of perception of syllables and aspects of visual perception. Apart from demonstrating the same characteristics of pattern formation at various levels of cognition, the study points to the usefulness of the dynamical approach in the investigation of language understanding.

"How to milk a coat:" the effects of semantic and acoustic information on phoneme categorization.

This study examined the effect of sentence context and local acoustic structure on phoneme categorization. Target stimuli from a 10-step GOAT-COAT continuum, differing only on a temporal cue for voice onset time (VOT), were embedded in carrier sentences that biased interpretation toward either "goat" or "coat." While subjects listened to the sentences they also responded as quickly as possible to a visual probe by indicating whether the probe matched the target stimulus they heard. Results showed that the interaction of VOT and sentence context significantly affected both identification and RT for stimuli near the perceptual boundary; the identification function showed a boundary shift in favor of the biased context and peak response times for each context reflected the shifted identification boundaries. In addition, response times were faster for identification of stimuli near the category boundary when responses were congruent, rather than incongruent with the sentence context. The response time differences for congruent versus incongruent responses in the boundary region are interpreted as depending on the results of initial phonological analysis; potentially ambiguous categorizations may be subject to additional evaluation in which a context-congruent response is both preferred and available earlier.

Temporal patterning in an auditory illusion: the verbal transformation effect.

When the same syllable is presented repeatedly to a human subject, it undergoes perceptual changes known as verbal transformations. However, the dynamics of such perceptual changes is poorly understood. In the present work we show that the main organization of the perceptual transitions is into pairs. This pairwise coupling is pronounced in the frequency of switching to pair members, but not in their dwell times (the time spent perceiving a given phonemic form before switching to another form). We show that the paircoupled transforms of perception have a faster and more stable dynamic than the nonpaircoupled transforms. We also demonstrate that the pairwise coupling is stronger than would be expected from random arrangement of small numbers of transforms. These characteristic patterns of verbal transforms have been predicted by a mathematical model, first proposed as a model of perceptual alternations of ambiguous visual figures.

A synergetic model for the verbal transformation effect.

We describe a nonlinear dynamical model based on synergetics for the auditory perceptual illusion known as the verbal transformation effect. The model is an extension of a synergetic model of perceptual oscillations of visual ambiguous figures. The main extension is connected to the number of reported alternative phonemic structures, which is typically much greater than the two or three alternatives usually reported in experiments with visual ambiguous figures. The properties of the model, which are derived using basic psychophysical principles, are presented and evaluated on the basis of the fit to earlier empirical work. It will be shown that there is very good agreement between the empirically observed properties of the verbal transformation effect and the properties detected by the model.

Fractal timing of verbal transforms.

The paper deals with the timing pattern of perceptual change elicited by multiple repetitions of a syllable (the verbal-transformation effect). We show that the distribution of the dwell time, the time spent perceiving a given phonemic form before switching to another form, obeys a power law with an exponent valued between 1 and 2. This result is robust, occurring for meaningless syllables and for English words of different initial phonemic salience. Experiment 2 demonstrates that subjects are aware of the temporal dynamics of perceptual change. On the basis of these results, it is argued that within this paradigm the notion of a mean dwell time is ill defined and there is apparently no characteristic time scale for perceptual change.

Ambient noise interferes with auscultatory blood pressure measurement during exercise.

This study was designed to investigate whether the acoustical characteristics of the Korotkoff sounds (K-sounds) were altered during exercise and/or masked by the ambient noise. After signing informed consent, 11 subjects (8 females, 3 males; 27 +/- 2 yr; 166.2 +/- 3.2 cm; 62 +/- 5 kg; means +/- SD) underwent a cycle ergometer exercise test that increased in workload by 30 W every 3 min until volitional fatigue. Heart rate, auscultatory systolic (SBP) and diastolic blood pressure (DBP), and oxygen consumption were monitored 1 and 2 min into each work stage. The auscultatory K-sounds were recorded with a microphone mounted in a stethoscope tube for later frequency (Hz) and sound pressure level (dB SPL) analysis. Frequency and SPL of ambient noise (99 +/- 13 Hz and 64 +/- 1 db at maximum, respectively) increased during the exercise test to magnitudes similar to the SBP and DBP K-sounds (166 Hz, 66 db; and 128 Hz, 69 db, respectively). Additionally, the ambient noise was responsible for a significant damping of the frequency and SPL of the measured blood pressure K-sounds and a rise in the measured frequency of the SBP K-sounds. Furthermore, we observed "inaudible" K-sounds at lower frequencies than adjoining audible K-sounds (100 Hz vs 126 Hz), supporting the known underestimation of SBP by auscultation. The increase in ambient noise during exercise testing dampens and may mask the auscultatory K-sounds, thus making detection of the proper K-sounds during exercise difficult at best. Furthermore, the presence of inaudible K-sounds may further explain the published discrepancies between auscultatory and intraarterial blood pressure measurements during exercise.

Prosodic influences on the resolution of temporary ambiguity during on-line sentence processing.

We present three experiments designed to investigate the role of prosody during sentence processing. The first investigated the question of whether an utterance's prosodic contour influences its comprehension on-line. We spliced the beginning and end portions of direct object and embedded clause sentences and observed the consequent effects on comprehension using a dual-task procedure to measure processing load. Our second experiment sought to determine whether the constituent structure of these sentences could be reliably predicted using prosodic information. We found that the duration and F0 contour associated with the main-clause verb and the following NP reliably distinguished between the direct object and embedded clause constructions. In the final experiment, we manipulated the duration of the main-clause verb and found that subjects used this information to guide their initial parse during on-line sentence comprehension. The need for a model of sentence processing that addresses the use of prosodic information is discussed.

Evaluation of a dynamical model of speech perception.

Previous work (Tuller, Case, Ding, & Kelso, 1994) has revealed signature properties of nonlinear dynamical systems in how people categorize speech sounds. The data were modeled by using a two-well potential function that deformed with stimulus properties and was sensitive to context. Here we evaluate one prediction of the model--namely, that the rate of change of the potential's slope should increase when the category is repeatedly perceived. Judged goodness of category membership was used as an index of the slope of the potential. Stimuli from a "say"-"stay" continuum were presented with gap duration changing sequentially throughout the range from 0 to 76 to 0 msec, or from 76 to 0 to 76 msec. Subjects identified each token as either "say" or "stay" and rated how good an exemplar it was of the identified category. As predicted, the same physical stimulus presented at the end of a sequence was judged a better exemplar of the category than was the identical stimulus presented at the beginning of the sequence. In contrast, stimuli presented twice near the middle of a sequence with few (or no) stimuli between them, as well as stimuli presented with an intervening random set, showed no such differences. These results confirm the hypothesis of a context-sensitive dynamical representation underlying speech.

The nonlinear dynamics of speech categorization.

Little is known about the processes underlying the nonlinear relationship between acoustics and speech perception. In Experiment 1, we explored the effects of systematic variation of a single acoustic parameter (silent gap duration between a natural utterance of s and a synthetic vowel ay) on judgements of speech category. The resulting shifts in category boundary between say and stay showed rich dynamics, including hysteresis, contrast, and critical boundary effects. We propose a dynamical model to account for the observed patterns. Experiment 2 evaluated one prediction of the model, that changing the relative stability of the two percepts allows categorical switching. In agreement with the model; an increase in the number of stimulus repetitions maximized the frequency of judgments of category change near the boundary. Thus, a dynamical approach affords the rudiments for a theory of the effects of temporal context on speech categorization.

The production and perception of syllable structure.

Much research on speech over the years has focused on uncovering examples of the nonlinear relationship between acoustics and perception (e.g., so-called "categorical perception") and between articulation and acoustics [as described by Stevens's (1972) quantal theory]. In the present experiment we demonstrate that naturally occurring linear changes in articulation may also be perceived discontinuously. Specifically, linear changes in relative phase of glottal and oral movements are perceived as categorical changes in the location of syllable juncture. Thus, phase transitions observed during speech demarcate a change in syllabic organization.

An evaluation of an alternating magnetic field device for monitoring tongue movements.

Alternating magnetic field devices hold much promise for tracking movements of multiple articulators, including the tongue, in the midsagittal plane. Here, the accuracy, repeatability, and linearity of one such device, the Articulograph AG-100 (Carstens Medizinelektronik GmbH, Göttingen, West Germany), are evaluated. The results indicate that with proper precautions in transducer calibration and placement, the Articulograph can be a useful tool in speech production research.

Environmentally-specified patterns of movement coordination in normal and split-brain subjects.

Rhythmic movement patterns between the hands in response to environmental signals are studied in normal (musicians and nonmusicians) and split-brain subjects. Only two phase-locked states - in-phase and anti-phase - are shown to be stable for all subjects. Split-brain subjects show an even greater attraction to these patterns, thus providing no support for the notion that reduced cortical interaction between the hemispheres allows for independent visuomotor control of the hands in such tasks. Moreover, differences in trajectories produced by normal individuals and those without an intact corpus callosum are remarkable. The resultant patterns of coordination afford 1) a generalization of previous results on intrisincally generated rhythmic behavior to environmentally-specified movement patterns; and 2) a discussion of neural mechanisms underlying stable phase relations.

Patterns of interarticulator phasing and their relation to linguistic structure.

Work by Tuller and Kelso [J. Acoust. Soc. Am. 76, 1030-1036 (1984)] and Kelso et al. [J. Phon. 14, 29-59 (1986)] has demonstrated stable relations between jaw and lip movements in (bV#CVb) utterances across rate and stress conditions. Specifically, the onset of lip movement toward the intervocalic consonant was found to be constant with respect to the vowel-to-vowel jaw cycle in both time and relative phasing. An attempt was made to replicate and extend this work by investigating interarticulator phase relations for utterances having a broader range of linguistic organization: In addition to rate and stress, syllable structure (open versus closed syllables) and identity of the intervocalic consonant (/p/ vs /m/) were manipulated. Results showed that the upper lip's lowering onset varied systematically with respect to the jaw vowel cycle as a function of both rate and stress. In addition, syllable structure and consonant identity influenced the relation of lip and jaw gestures. There was a general tendency for any condition that shortened the first vowel to produce earlier onsets of the upper lip relative to the jaw. However, the within-condition jaw cycle duration variability did not correlate with the within-condition variability in phase. Thus it seems that stable interarticulator phase relations maintain not only the integrity of phonological structure, as suggested by Kelso et al., but structural integrity at other levels of linguistic organization as well.

Functionally specific articulatory cooperation following jaw perturbations during speech: evidence for coordinative structures.

In three experiments we show that articulatory patterns in response to jaw perturbations are specific to the utterance produced. In Experiments 1 and 2, an unexpected constant force load (5.88 N) applied during upward jaw motion for final /b/ closure in the utterance /baeb/ revealed nearly immediate compensation in upper and lower lips, but not the tongue, on the first perturbation trial. The same perturbation applied during the utterance /baez/ evoked rapid and increased tongue-muscle activity for /z/ frication, but no active lip compensation. Although jaw perturbation represented a threat to both utterances, no perceptible distortion of speech occurred. In Experiment 3, the phase of the jaw perturbation was varied during the production of bilabial consonants. Remote reactions in the upper lip were observed only when the jaw was perturbed during the closing phase of motion. These findings provide evidence for flexibly assembled coordinative structures in speech production.

The timing of articulatory gestures: evidence for relational invariants.

In this article, we examine the effects of changing speaking rate and syllable stress on the space-time structure of articulatory gestures. Lip and jaw movements of four subjects were monitored during production of selected bisyllabic utterances in which stress and rate were orthogonally varied. Analysis of the relative timing of articulatory movements revealed that the time of onset of gestures specific to consonant articulation was tightly linked to the timing of gestures specific to the flanking vowels. The observed temporal stability was independent of large variations in displacement, duration, and velocity of individual gestures. The kinematic results are in close agreement with our previously reported EMG findings [B. Tuller et al., J. Exp. Psychol. 8, 460-472 (1982)] and together provide evidence for relational invariants in articulation.

Converging evidence in support of common dynamical principles for speech and movement coordination.

We suggest that a principled analysis of language and action should begin with an understanding of the rate-dependent, dynamical processes that underlie their implementation. Here we present a summary of our ongoing speech production research, which reveals some striking similarities with other work on limb movements. Four design themes emerge for articulatory systems: 1) they are functionally rather than anatomically specific in the way they work; 2) they exhibit equifinality and in doing so fall under the generic category of a dynamical system called point attractor; 3) across transformations they preserve a relationally invariant topology; and 4) this, combined with their stable cyclic nature, suggests that they can function as nonlinear, limit cycle oscillators (periodic attractors). This brief inventory of regularities, though not mean to be inclusive, hints strongly that speech and other movements share a common, dynamical mode of operation.

On categorizing aphasic speech errors.

Acoustic studies of voice-onset-time in aphasics' speech suggest that fluent aphasics' errors are misselected phonemic targets whereas nonfluent aphasics' errors are of articulatory origin. However, we must be cautious when extrapolating a theory from only one measure of articulation. In this experiment, I examined utterances produced by five fluent aphasics, five nonfluent aphasics and two controls. First, the voice-onset-time findings were replicated. Second, I examined the duration of vowels preceding word-final stop consonants as an index of the consonant's voicing category. The pattern of voice-onset-times produced did not predict the pattern of vowel durations. Thus, voice-onset-time cannot be used to characterize more generally the output of the speaker.

Converging evidence for the role of relative timing in speech.

In an earlier article (Tuller, Kelso, & Harris, 1982a) we suggested that the timing of consonant-related muscle activity was constrained relative to the period between onsets of muscle activity for successive vowels. Here, we reexamine those data based on reservations posed by Barry (1983). Next, we present a kinematic study of articulation that extends and strongly supports our original observations. Finally, we very briefly survey some converging lines of evidence for a functionally significant vowel-to-vowel period in speech and how this may relate to the role of temporal invariance in motor skills in general.