On the assessment of stability and patterning of speech movements.

Speech requires the control of complex movements of orofacial structures to produce dynamic variations in the vocal tract transfer function. The nature of the underlying motor control processes has traditionally been investigated by employing measures of articulatory movements, including movement amplitude, velocity, and duration, at selected points in time. An alternative approach, first used in the study of limb motion, is to examine the entire movement trajectory over time. A new approach to speech movement trajectory analysis was introduced in earlier work from this laboratory. In this method, trajectories from multiple movement sequences are time- and amplitude-normalized, and the STI (spatiotemporal index) is computed to capture the degree of convergence of a set of trajectories onto a single, underlying movement template. This research note describes the rationale for this analysis and provides a detailed description of the signal processing involved. Alternative interpolation procedures for time-normalization of kinematic data are also considered.

Prosodic influences on speech production in children with specific language impairment and speech deficits: kinematic, acoustic, and transcription evidence.

It is often hypothesized that young children's difficulties with producing weak-strong (iambic) prosodic forms arise from perceptual or linguistically based production factors. A third possible contributor to errors in the iambic form may be biological constraints, or biases, of the motor system. In the present study, 7 children with specific language impairment (SLI) and speech deficits were matched to same age peers. Multiple levels of analysis, including kinematic (modulation and stability of movement), acoustic, and transcription, were applied to children's productions of iambic (weak-strong) and trochaic (strong-weak) prosodic forms. Findings suggest that a motor bias toward producing unmodulated rhythmic articulatory movements, similar to that observed in canonical babbling, contribute to children's acquisition of metrical forms. Children with SLI and speech deficits show less mature segmental and speech motor systems, as well as decreased modulation of movement in later developing iambic forms. Further, components of prosodic and segmental acquisition develop independently and at different rates.

Metrical effects on speech movements in children and adults.

The present study investigates motor processes underlying the production of iambic and trochaic metrical forms for children and adults. Lower lip movement was recorded while 16 children between the ages of 3;10 and 4;9 (years; months) and 8 adults produced iambic (e.g., [see text]) and trochaic (e.g., [see text]) nonce words. For both children and adults, movement patterns for iambic and trochaic words are well differentiated, but in qualitatively different ways. Most notably, children do not produce amplitude modulated forms for trochees, perhaps reflecting a reliance on early developing rhythmic patterns such as those seen in canonical babbling. In contrast, movements corresponding to iambs are well modulated and particularly stable for both groups of speakers, suggesting that they require increased movement specificity. It appears that metrical forms are perceptually and linguistically established and that the child finds the means available within his or her existent motor repertoire to produce adequately differentiated movements corresponding with iambs and trochees.

Development and phonetic differentiation of speech movement patterns.

It is often hypothesized that speech production units are less distinctive in young children and that generalized movement primitives, or templates, serve as a base on which distinctive, mature templates are later elaborated. This hypothesis was examined by analyzing the shape and stability of single close-open speech movements of the lower lip recorded in 4-year-old, 7-year-old, and adult speakers during production of utterances that varied in only a single phoneme. To assess the presence of a generalized template, lower lip movement sequences were time and amplitude normalized, and a pattern recognition procedure was implemented. The findings indicate that speech movements of children already converged on phonetically distinctive patterns by 4 years of age. In contrast, an index of spatiotemporal stability demonstrated that the stability of underlying patterning of the movement sequence improves with maturation.

Stability and patterning of speech movement sequences in children and adults.

Children (aged 4 and 7 years) and young adults produced a six-syllable utterance 15 times. The displacement of the lower lip was recorded with an Optotrak system and analyzed in a number of ways. First, using a procedure recently developed in our laboratory, displacement records from the 15 repetitions were amplitude- and time-normalized, and the spatiotemporal index (the STI) was computed. The STI reflects the degree to which repeated performance of a task produces movement trajectories that converge on a single pattern. Children produced less stable movement trajectories, as reflected in higher values on the STI. In a second analysis, standard measurements of amplitude and peak velocity were made for two opening and two closing lip movements. These measures suggested that, relative to the size of their oral structures, children have large movement ranges in speech. Also, children tend to move with a lower peak velocity. This large-amplitude, low-velocity movement style may reflect different underlying control processes. Finally, another analysis focused on open-close movement sequences associated with two words of the utterance. A pattern-recognition algorithm applied to the normalized waveforms from the open-close sequences revealed that children and adults produced equally distinctive movement trajectories for the two syllables. Taken together, these preliminary results suggest that nonlinear and nonuniform changes occur in components of the speech motor system during development.

Information level and young children's phonological accuracy.

The influence of information level on the production accuracy of 20 children (22 to 28 months) was examined. The data were children's productions of nouns in sets of utterances referring to triplets of pictures representing noun-verb-noun utterances. In each triplet one noun remained the same, thus decreasing in information value from the first to the third picture (new, moderately old, and old information). Words representing new information were produced more accurately than words representing old information. The types of errors did not differ. Further evidence of this effect was provided by an examination of the duration of new versus old word productions by 12 of the children. Productions encoding new information were consistently longer on average than those encoding old information. The result provide experimental evidence of an effect observed in adults that indicates early sensitivity to information level. However, because of the children's young age, the effects are characterized as a speaker-internal process that only serendipitously corresponds to listener needs.

Young children's production of syllable stress: an acoustic analysis.

The acoustic characteristics of stress were examined in young children's productions of minimal pairs of novel words (e.g., sofi versus so'fi). Fourteen 2-year-olds participated as subjects. Their productions were analyzed in terms of vowel duration, syllable duration, peak amplitude, and peak fundamental frequency. The analyses revealed that children produced stressed and unstressed syllables distinctly along each of the dimensions examined. The absolute and relative (unstressed/stressed) values of the children's productions were compared to those of the single adult experimenter, who modeled the novel words, permitting a unique comparison of input to children's productions. One systematic difference was the relative values; the children's stressed and unstressed syllables were less distinct than the adults along each of the acoustic correlates. Furthermore, the acoustic features of both stressed and unstressed syllables appear to be subject to developmental change. The results are discussed in terms of their implications for young children's production capabilities and for the relationship between input and children's production characteristics.

Orofacial muscle activity of children who stutter: a preliminary study.

This study was a preliminary investigation of the relations between stuttering development and the maturation of speech motor processes. Electromyographic (EMG) activity was recorded from the orofacial muscles of children who stutter and their normally fluent peers during fluent and disfluent speech. Nine children who stutter (8 boys and 1 girl), ranging in age from 2:7 to 14:0, and 9 age- and sex-matched children who do not stutter were subjects. Pairs of surface EMG electrodes were placed on children's faces overlying the anterior belly of the digastric (ABD), levator labii superior (ULIP), and orbicularis oris inferior (LLIP) muscles. Twenty segments of stuttered (for the children who stutter) and perceptually fluent speech were extracted from children's conversational speech samples. Spectra of the amplitude envelopes of the EMG activity were computed. The 3 oldest children who stutter showed evidence of tremorlike oscillations of EMG activity in the 5 to 15 Hz range during stuttering in either ULIP, LLIP, or ABD muscles. The younger children who stutter and the children who do not stutter demonstrated primary spectral peaks in the 1 to 4 Hz range during stuttered and/or perceptually fluent speech. It is hypothesized that the emergence of tremorlike instabilities in the speech motor processes of children who stutter may coincide with aspects of their general neural maturation and with the development of stuttering.

Metrical patterns of words and production accuracy.

This investigation examined the influence of metrical patterns of words (syllable stress and serial position) on the production accuracy of 20 children (22 to 28 months). The data were productions of six pairs of individualized two-syllable experimental words that referred to unfamiliar objects. Members of each pair differed only in the placement of stress (e.g., ['soti] vs. [so'ti]). Unstressed syllables were much more likely to be omitted, particularly at the beginning of words. Very few stressed syllables and unstressed second position syllables were omitted. One fourth of the word initial unstressed syllables were omitted. Consonant omissions, though few in number, tended to occur in initial position. Assimilation errors were not influenced by stress or serial position. When segmental errors due to syllable omissions were excluded, other consonant errors were not affected by stress or serial position. These findings indicate that young children's productions of syllables are influenced by the metrical patterns of words. However, the trochaic pattern of English is a statistical tendency, not an absolute constraint on two-syllable words. Metrical pattern also does not affect the consonant accuracy in syllables produced.

Spatiotemporal stability and patterning of speech movement sequences.

In order to examine the stability and patterning of speech movement sequences, movements of the lip were recorded as subjects produced a phrase at normal, fast, and slow rates. Three methods of analysis were employed. First, a new index of spatiotemporal stability was derived by summing the standard deviations computed across amplitude- and time-normalized displacement records. This index indicated that normal and fast rates of speech production result in more stable movement execution compared to slow rates. In the second analysis, the relative time of occurrence of the peak velocity of the three middle opening movements of the utterance was measured. For each of the three peaks, the preservation of relative timing was assessed by applying Genter's (1987) slope test. The results clearly indicate that the relative timing of these events does not remain constant across changes in speech rate. The relative timing of the middle opening gestures shifted, becoming later as utterance duration increased. In a third analysis, pattern recognition techniques were applied to the normalized displacement waveforms. A classification algorithm was highly successful in sorting waveforms into normal, fast, and slow rate conditions. These findings were interpreted to suggest that, within a subject, three distinct patterns or movement templates exist, one for each rate of production. Speech rate appears to be a global parameter, one that affects the entire command sequence for the utterance.

Human perioral muscle activation patterns.

Task-dependent human motor organization in the perioral region was examined in eight normal adults who performed oral tasks including lip protrusion, chewing, and speech. Zero phase-lag correlations among EMG signals recorded from quadrants surrounding the lips were calculated in order to determine patterns of motor coupling. Results indicated that the perioral musculature is flexible in output organization. Activity in all quadrants was highly positively correlated during the protrusion task. During the chewing task, correlations were moderate, with a stronger pattern bilaterally across the upper and lower lips. The speech tasks showed lower levels of correlation among the quadrants, but again the pattern was more highly correlated bilaterally than ipsilaterally. Results are compared to studies of oral muscle innervation in humans and animals and also are related to hypotheses of cortical control patterns for oral movement.

Motor unit territories in the human perioral musculature.

The perioral region was divided into four quadrants, and electromyograms (EMGs) were recorded from each area. The coherence function (i.e., the squared cross correlation between two signals computed at each frequency in the spectrum) was used to determine aspects of the organization of motor unit territories and to examine potential higher level sources of input in speech and nonspeech tasks. Coherence functions were computed between pairs of EMGs and were examined for significant values in the range of 20-240 Hz. When two pairs of electrodes were intentionally placed to record the activity from a common subset of motor units in a single quadrant of the lower lip, all subjects exhibited significant broad-band coherence in every frequency in all experimental tasks. Thus, the presence of such a pattern of broad-band significant coherence for EMG pairs recorded from different quadrants would indicate that single motor unit territories extended across perioral quadrants. When separate EMG recordings were obtained from the four quadrants of the lips, coherence functions computed between EMG pairs were typically zero across the entire frequency range. These findings suggest that perioral motor unit territories are organized into nonoverlapping quadrants. Further, the present analyses suggest that, unlike bilateral pairs of jaw-closing muscles during chewing, these motor units are not driven by any correlated oscillatory activity in chewing or other oral motor tasks.