Ludo Max1,3, Vincent L. Gracco2,3, and Anthony J. Caruso4
1University of Connecticut, Department of Communication Sciences, 850 Bolton Road Unit 1085, Storrs, CT 06269-1085, USA
2McGill University, School of Communication Sciences and Disorders, 1266 Pine Avenue West, Montreal, Quebec, H3G 1A8, Canada
3Haskins Laboratories, 2 70 Crown Street, New Haven, CT 0651] -6695, USA
4Kent State University, School of Speech Pathology and Audiology, Kent, OH44242-001, USA
Lip and jaw speech movements, lip and jaw nonspeech movements, and finger movements were analyzed to examine whether stuttering adults with no recent speech therapy differ from nonstuttering adults in the sequencing of peak velocity across effectors. Number of movements within a trial and location of the target movements within a trial were experimentally manipulated. Sequencing patterns for stuttering and nonstuttering individuals were similar for two movement types (closing/flexion and opening/extension) in all conditions of the three tasks. Specifically, the order of the most frequently used sequencing patterns was identical for the groups in each task. We conclude that intragestural motor timing as examined here is not impaired in adults who stutter.
De Nil and Abbs (1991) published data suggesting that such articulatory sequencing differences may be an artifact of speech rate differences: nonstuttering adults used the UL-LL-J peak velocity sequence less consistently than previously reported, and sequencing became more variable at slower rates. However, a later study failed to find an effect of speech rate on peak velocity sequencing (Ward, 1997).
McClean et al. (1990) studied nonstuttering adults, stuttering adults with no recent treatment, and stuttering adults with recent treatment. The UL-LL-J sequence -was the most typical pattern for nonstuttering individuals, but the same was true for stuttering individuals with no recent speech therapy. It should be noted that these measures were obtained by averaging the latencies from UL to LL and J peak velocity, respectively, across trials. Story and Alfonso (1989; cited in Alfonso, 1991) used similar measures averaged over multiple productions, but two of three stuttering subjects showed sequencing patterns that were different from those of two nonstuttering control subjects both pre- and post-treatment.
De Nil (1995) then investigated productions of sapapple in a carrier phrase as well as productions of emma papa and emma mafiia in a reading passage. Interestingly, findings showed a difference between stuttering and nonstuttering groups for sequencing of UL, LL, and J peak velocities in sapapple but not in emrna papa and emma maflia. Two additional studies with different target words also failed to find between-group differences in the sequencing of kinematic articulatory events (Jaincke et a1., 1995; Ward, 1997).
Several questions regarding intragestural articulatory timing in individuals who stutter remain unanswered. The purpose of this work was to examine (a) whether differences in kinematic event sequencing exist in the perceptually fluent speech of nonstuttering adults versus stuttering adults who had not participated in stuttering treatment for several years, (b) whether differences in kinematic event sequencing between these two groups of subjects exist for nonspeech motor tasks and systems, and (c) whether the presence or absence of possible differences in kinematic event sequencing between these two groups depends on specific task requirements, in particular trial length and location of the target movements within the trial.
Ten adults who stutter (27-45 years of age, M = 34.3, SD = 5.7) and ten adults who do not stutter (26-46 years of age, M = 33.9, SD = 5.9) participated. Each group consisted of 7 male and 3 female subjects, all native speakers of American English. Individuals who stutter reported that their stuttering started during childhood and that they had no present or previous neurological or communication disorders other than stuttering. Priorto data collection, they had not participated in stuttering treatment for at least eight years. The stuttering of five subjects was mild, that of four subjects was moderate, and that of one subject was severe (SSI-3; Riley, 1994). Nonstuttering subjects were individually matched (+/-3 years) with the stuttering subjects.
Participants completed three tasks in a single session: 21 Speech Task (ST), an Orofacial Nonspeech Task (ONT), and a Finger Movement Task (FMT). The FMT was always completed last whereas the order of the ST and ONT was randomized for each subject. Within each task, target movements were performed across four conditions that were selected to represent three levels of trial length and two levels of target position within the trial (here referred to as short, intermediate, long-initial, and long-final conditions). Each different utterance or movement trial within each task was elicited five times by custom-written computer software (with the order of the total set of trials within the task being completely randomized). If speech disfluencies or errors were noticed by either the experimenter or the subject during the experimental session, the trial was repeated until it was agreed that no disfluencies or errors were present. Only fluent and correct trials (as agreed upon by subject and experimenter during the session and a subsequent repeated listening/review by the experimenter) were included for analysis.
During the ST, subjects produced meaningful utterances in all conditions: in the short condition, the word my was followed by a target monosyllabic noun (e.g., My bob); in the intermediate condition, my was followed by a trisyllabic noun with the same initial syllable as the noun produced in the short condition (e.g., My bobby-pin); in the long-initial condition, my and the same trisyllabic noun occurred in the initial positions of a sentence (e.g., My bobby-pin was designed by an artist); and in the long-final condition, my and the same trisyllabic noun occurred in the final positions of a sentence (e.g., An artist designed my bobby-pin). Four different monosyllabic nouns were used, all with /p/ or /b/ as the initial consonant, /A/ or /Al/ as the vowel/diphthong, and /p/, /b/, or /m/ as the final consonant. Sequencing of UL, LL, and J peak velocities was determined for the closing and opening movements associated with the initial stop consonant in the target noun (i.e., the first consonant after my). Available for analysis were 789 closing and 789 opening movements from the nonstuttering group, and 773 closing and 773 opening movements from the stuttering group. During the ONT, subjects performed nonspeech orofacial movements: in the short condition, a “popping” sound was made by successively performing a bilabial closing and opening movement that resulted in a slight decrease of intra-oral air pressure; in the intermediate condition the same bilabial gesture was performed as the first of three successive nonspeech movements (e.g., bilabial popping, tongue protrusion, lip spreading); in the long-initial and long-final conditions, the same gestures were produced first or last, respectively, in a longer sequence of movements. The presented stimuli for this task consisted of descriptions of the required movements. For example, for the long-initial condition, the screen showed the following description: popping-tongue out corners back-tongue right-tongue 1eft-1ips round. For stuttering subjects S5 and S6, insufficient data were available to include these individuals (and matched nonstuttering subjects N5 and N6) in the analyses for opening and closing movements, respectively. A total of 168 closing and 167 opening movements from the nonstuttering group and 169 closing and 169 opening movements from the stuttering group were analyzed.
For the FMT, target movements consisted of thumb-index finger opposition movements (i.e., fine pinch) with the preferred hand. The short condition required the successive performance of a flexion (closing) and extension (opening) movement with the index finger and thumb; that is, from extension of both the index finger and thumb to flexion of both fingers (such that contact was made between the tips of these two fingers) and back to an extended position. In the intermediate condition, the index finger-thumb gesture from the short condition was the first in a series of three successive finger gestures. The second and third gestures consisted of the same fiexion and extension movements but this time performed with the middle finger and thumb and with the ring finger and thumb, respectively. In the long-initial and long-final conditions, the gestures from the short condition were performed first or last, respectively, in a longer sequence of six finger gestures.
Required responses for the FMT were presented on the computer monitor as a series listing those fingers that had to be used sequentially. For example, the following series was presented for the long-initial condition: index-middle-ring-little~-ring--middle. Each trial was started with the index finger and thumb in the fine pinch position. Technical difficulties made it necessary to exclude subject N10 (and matched subject S10). Analyses were based on 173 flexion and 173 extension movements from the nonstuttering group, and 171 flexion and 171 extension movements from the stuttering group.
Data acquisition and processing
Superior-inferior UL, LL, and J movements were transduced with a strain gauge system. Signals were amplified and low-pass filtered at 50 Hz (Model 205, Biocommunication Electronics), and 16-bit digitized at 3000 samples per second (MP100WSW, Biopac Systems). The J signal was online subtracted from the combined LL and J signal (transducer on the lower lip) to obtain a signal reflecting net LL movement. Signals were further digitally low-pass filtered at 20 Hz (FIR, 200 coefficients, Hamming window) and differentiated (three point central difference) and smoothed (10 ms window) to yield traces of movement velocity.
For the FMT, electrogoniometers were built using small, ultra-1ightweight potentiometers (Model 3352, Bourns) and nylon extensions. A two-joint model transduced angular displacement at the metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints of the index finger whereas a sing1e-joint model was used simultaneously for the interphalangeal (IP) joint of the thumb. The potentiometers were connected to Wheatstone bridges and the output signals were digitized, differentiated, and smoothed as described above.
Measurements and analyses Automatic software routines (Acknowledge, Biopac Systems) extracted the time of closing and opening peak velocity for UL, LL, and J in the ST and ONT, and fiexion and extension peak velocity for MCP, PIP, and IP in the FMT. For each movement, peak velocity sequencing was then determined by automatically ranking the respective peak velocity times for the three contributing effectors or joints.
Results for bilabial closing movements in the ST are shown in the top half of Figure 1 (due to space constraints, the long-final condition is not shown). Bars represent the percentage of trials produced with the different sequences listed on the category axis. For example, UL-LL-J indicates that peak velocity of UL was reached first, peak velocity of LL was reached second, and peak velocity of J was reached last (note that “/” indicates that the time of occurrence of two peak velocities could not be distinguished with the present temporal resolution).
Distribution of the trials across different sequencing patterns was highly similar for the stuttering and nonstuttering individuals in all conditions. Both groups preferred UL-lead sequences for the speech closing movements regardless of utterance length or target location within the utterance. The most frequent sequence for both groups in all conditions was UL-J-LL, and the proportion of such sequences was approximately the same across the groups. Between-group differences in use of this sequence were only 5.3%, 13.8%, 3.0%, and 3.8%, for the short, intermediate, long-initial, and long-finol conditions, respectively, and the differences were not consistent in direction. Combined, UL-J-LL and UL-LL-J sequences accounted for 73-85% (across conditions) of utterances produced by the nonstuttering group and 79-91% of utterances produced by the stuttering group.
The data from both groups were also similar with regard to the types of sequences that were commonly used as well as with regard to the rank order of those types based on frequency of occurrence. That is, both groups used almost exclusively UL-J-LL, UL-LL-J, or J-UL-LL sequences across all four conditions, with other sequences (such as both LL-lead sequences, J -LL-UL sequences, and sequences with simultaneously reached peak velocities) each occurring in less than 4% of the utterances from each group. With regard to the frequency of occurrence of the three most commonly used sequences, both groups showed the same rank order, and this rank order was the same in all four conditions. From the most frequently used to the least frequently used sequence, the order for both groups was (a) UL-J-LL, (b) UL-LL-J, and (c) J -UL-LL.
The bottom half of Figure 1 summarizes the results for opening movements in the ST. It can be seen that (a) sequencing patterns for the opening movements differed from those described above for the closing movements, and (b) distribution of the trials across the different sequencing patterns was again remarkably similar for the stuttering and nonstuttering individuals in all conditions. Whereas both groups most frequently used UL-lead sequences for speech closing movements, they most frequently used J -lead sequences for speech opening movements, and this finding was again independent of utterance length and target location. The most frequently used sequence in all four conditions was J -LL-UL, and the percentage of utterances with this sequence was almost identical for the two groups. Specifically, the between-group differences in use of the J-LL-UL sequence were only 7.3%, 0.9%, 0.3%, and 2.8%, for the short, intermediate, long-initial, and long-final conditions, respectively, and the direction of these differences was not consistent across the conditions. Combined, J -LL-UL and J -UL-LL sequences accounted for 65-72% (across conditions) of all utterances produced by the nonstuttering group and for 69-77% of all utterances produced by the stuttering group.
It can also be seen in Figure 1 that the stuttering and nonstuttering groups were similar with regard to the rank order of the sequencing patterns that were most commonly used for these bilabial opening movements during speech. Most typical for both groups across all four conditions were I-LL-UL, J-UL-LL, and LL-Jâ_ UL sequences, respectively. The reversal in direction of the difference between J-UL-LL and LL-J-UL within the nonstuttering group was related to a difference of only .5% or one single utterance.
Figure l. Stuttering and nonstuttering groups’ peak velocity sequencing for bilabial closing (top) and opening (bottom) movements in the short, intermediate, and long-initial conditions of the Speech Task.
Orofacial Nonspeech Task
The top panels in Figure 2 display the results for bilabial closing movements in the ONT. These nonspeech data are similar to the speech data in that, regardless of trial length and target location within the trial, closing movements were most frequently performed with UL-lead sequences, and this was the case for stuttering as well as nonstuttering individuals. However, the frequency difference between UL-lead sequences and tie second most frequently used pattern (J-lead) was smaller for the ONT than for the ST (especially for nonstuttering individuals in the long-initial condition). Another difference between the ONT and ST closing movements is that both groups used a larger number of different sequencing patterns during the ONT. Speech movements were almost always performed with one of the three most common sequencing patterns, and the frequency of all remaining patterns was very low (always less than 4% of all movements). The orofacial nonspeech movements, on the other hand, were performed with much more variability across individual trials. Although UL-lead sequences were used most frequently, several other sequences occurred with frequencies in the 2-30% range.
Overall, the distribution of trials across different patterns was again similar for both groups. Subjects most frequently used UL-lead sequences for the nonspeech closing movements in all conditions. Combined, UL-J-LL and UL-LL-J sequences accounted for 52-73% (across conditions) of all movements of the nonstuttering group and 50-63% of all movements of the stuttering group. In the intermediate and long-initial conditions, however, the groups differed in preference for the two possible UL-lead sequences: the UL-J-LL pattern was most common for nonstuttering subjects whereas the UL-LL-J pattern was most common for stuttering subjects.
Results for ONT opening movements are shown in the bottom half of Figure 2. Opening movements, too, showed both similarities and differences when comparing nonspeech versus speech movements and stuttering versus nonstuttering subjects. A similarity between the tasks is that, regardless of trial length and target location within the trial, both groups performed the opening most frequently with a J-lead sequence. One difference, however, is that both groups used the J- UL-LL sequence more consistently than J -LL-UL for nonspeech opening whereas the opposite was true for speech opening. A second difference is that, following J -lead sequences, UL-lead sequences were the second most frequently used pattern for nonspeech opening movements whereas LL-lead sequences were the second most frequently used pattern for speech opening movements.
Distribution of trials across the possible sequencing patterns was again similar for both groups. As already mentioned, stuttering as well as nonstuttering participants most frequently used J-lead sequences for these nonspeech opening movements in all conditions. Combined, J-UL-LL and J-LL-UL sequences accounted for 59-71% (across conditions) of all utterances produced by the nonstuttering group and for 45-59% of all utterances produced by the stuttering group. The discrepancy in these percentages for the stuttering versus nonstuttering group resulted from the fact that the stuttering individuals’ percentage of Head sequences in each condition was 10-20% lower than tha: of the nonstu :tering individuals. This, in turn, was due to an approximately equally frequent use of the J -UL-LL sequence but a less frequent use of the J-LL-UL sequence by the stuttering individuals. Instead, the stuttering individuals used more UL-lead sequences than the nonstuttering individuals.
Figure 2. Stuttering and nonstuttering groups’ peak velocity sequencing for bilabial closing (top) and opening (bottom) movements in the short, intermediate, and long-initial conditions of the Finger Movement Task
Results for MCP, PIP, and IP flexion movements in the FMT are shown graphically in the top half of Figure 3. The data from the two groups were similar to the extent that PIP-lead sequences were the most frequently used pattern in all four conditions for both the stuttering and the nonstuttering individuals. Combined, PIP-lead sequences accounted for 50-58% (across conditions) of all trials performed by the nonstuttering group and for 70-85% of all trials performed by the stuttering group.
Clearly, the stuttering group was more consistent than the nonstuttering group in the use of PIP-lead sequences. The difference between the two groups in use of the PIP-MCP-IP sequence was generally rather small, but the nonstuttering individuals used considerably fewer PIP-IP-MCP sequences than the stuttering individuals with differences of 26.2%, 20.5%, 23.6%, and 10.0% in the short, intermediate, long-initial, and long-final conditions, respectively. Instead, the nonstuttering individuals used more MCP-lead sequences (primarily MCP-PIP-IP) and more IP-lead sequences (primarily IP-PIP-MCP). Thus, trials performed by the nonstuttering group were distributed more equally over several sequence patterns whereas trials performed by the stuttering group contained a high proportion of PIP-lead sequences and a low proportion of all other patterns.
The bottom panels of Figure 3 show the results for finger extension movements. Neither group showed a preference for either MCP-lead, PIP-lead, or IP-lead sequences. Rather, all three types of sequencing patterns were commonly used across all conditions by both the stuttering and nonstuttering individuals. Additionally, the data did not provide any clear evidence for the use of different peak velocity sequencing patterns by stuttering versus nonstuttering individuals. For example, in the short condition, the frequency of MCP-lead, PIP-lead, and IP-lead sequences was 31.8%, 31.9%, and 36.4%, respectively, for the nonstuttering group and 35.5%, 37.8%, and 26.6%, respectively, for the stuttering group.
Figure 3. Stuttering and nonstuttering groups ‘ peak velocity sequencing for flexion (top) and extension (bottom) movements in the short, intermediate, and long-initial conditions of the Finger Movement Task.
This work was funded, in part, by NIH grant DC 03102.
Alfonso, P. J. (1991). Implications of the concepts underlying task-dynamic modeling on kinematic studies of stuttering. In H. F. M. Peters, W. Hulstijn, & C. W. Starkweather (Eds.), Speech motor control and stuttering (pp. 79-100). Amsterdam: Elsevier Science Publishers. 322 Theory, research and therapy in fluency disorders
Caruso, A. J ., Abbs, J . H., & Gracco, V. L. (1988). Kinematic analysis of multiple movement coordination during speech in stutterers. Brain, 1]], 439-456.
De Nil, L. F. (1995). The influence of phonetic context on temporal sequencing of upper lip, lower lip, and jaw peak velocity and movement onset during bilabial consonants in stuttering and nonstuttering adults. Journal of Fluency Disorders, 20, 127-144.
De Nil, L. F., & Abbs, J . H. (1991). Influence of speaking rate on the upper lip, lower lip, and jaw peak velocity sequencing during bilabial closing movements. Journal of the Acoustical Society of America, 89, 845-849.
Jancke, L., Kaiser, P., Bauer, A., Kalveram, T. (1995). Upper lip, lower lip, and jaw peak velocity sequence during bilabial closures: No differences between stutterers and nonstutterers. Journal of the Acoustical Society of America, 97, 3900-3903.
Max, L. (in press). Stuttering and internal models for sensorimotor control: A theoretical perspective to generate testable hypotheses. In B. Maassen et al. (Eds.), Speech motor control in normal and disordered speech. Oxford, UK: Oxford University Press.
Max, L., & Gracco, V. L. (2003). Coordination of oral and laryngeal movements in the perceptually fluent speech of adults who stutter. Paper submitted for publication.
Max, L., Gracco, V. L., Guenther, F. H., Ghosh, S., & Wallace, M. E. (2004). A sensorimotor model of stuttering: Insights from the neuroscience of motor control. In A. Packman et al. (Eds.), Proceedings of the 4”â World Congress on Fluency Disorders, Montreal, Canada.
Max, L., Guenther, F. H., Gracco, V. L., Ghosh, S. S., & Wallace, M. E. (in press). Unstable or insufficiently activated internal models and feedback-biased motor control as sources of dysfluency: A theoretical model of stuttering. Contemporary Issues in Communication Sciences and Disorders.
Max, L., & Yudman, E. M. (2003). Accuracy and variability of isochronous rhythmic timing across motor systems in stuttering and nonstuttering individuals. Journal of Speech, Language, and Hearing Research, 46, 146-163.
McClean, M. D., Kroll, R. M., & Loftus, N. S. (1990). Kinematic analysis of lip closure in stutterers’ fluent speech. Journal of Speech and Hearing Research, 33, 755-760.
Riley, G. (1994). Stuttering Severity Instrument for Children and Adults (3rd ed.). Austin: Pro-Ed; 1994.
Ward, D. (1997). Intrinsic and extrinsic timing in stutterers’ speech: Data and implications. Language and Speech, 40, 289-310.