Jennifer S. Carlson1 and Lisa R. LASALLE2
1Claire Santagati Vatz, M.A., C. C. C., INC., Private Practice, 250 Mt. Lebanon Blvd, Suite 41], Pittsburgh, PA 15234, USA
2University of Wisconsin-Eau Claire, Communication Disorders Department, 239 Water Street, Eau Claire, WI 54702, USA
The purpose of this study was to determine if clinicians’ slow speech rates facilitate fluency of preschoolers who stutter. Seven preschoolers who stutter and their clinicians served as participants. Clinicians’ speech rates were categorized into “slow” ( 3.0 syllables per second [SPS]) or “fast” ( 3.33 SPS), and the children’s adjacent utterances as “stuttered” or “normally disfluent/fluent.” As a group, observed probabilities did not differ from expected where the child either stuttered or spoke fluently, based on clinician rate of speech. One boy, however, stuttered significantly more when his clinician spoke quickly, while another boy stuttered significantly more when his clinician spoke slowly.
Brown (2002) found that stuttering children spoke more fluently when a clinician modeled a slow speech rate, but only in four of the six (67%) clinician-child dyads investigated, and Zebrowski et al. (1996) found this was true in only three of the five (60%) mother-child dyads investigated. To date, there has not been enough evidence to explain why approximately 30-40% of children who stutter do not increase their fluency in response to slow adult models. It is intriguing to note that the two children in Brown’s (2002) study who did not speak more fluently when their clinician slowed had a concomitant phonological disorder. There are problems in interpreting Brown’s (2002) finding, however, because in children’s overall fluency data, both linguistically constrained and spontaneous utterances were included. Spontaneous child utterances that immediately follow clinicians’ slow utterances are needed to learn about the individual contribution of a clinician’s slow rate model to a child’s fluency since linguistically constrained utterances (e.g., carrier phrases, listing) are often fluent due to linguistic simplicity, not due to slow rate model effects.
Researchers, as well as authors who reviewed stuttering parent-child interaction literature (Nippold & Rudzinski, 1995; Zebrowski, 1995) advocate caution in recommending that adults simply slow their rate as a way to get children to speak more fluently. Further support is needed in order to claim that adults’ slower speech facilitates greater fluency in children who stutter. Better rationale is needed for both recommending that parents of children who stutter slow their speech rate and clinicians also slow their speech rate during therapy. Because the "adult-slow-child-fluent” phenomenon appears not to occur in 30-40% of cases (Brown, 2002; Zebrowski et al., 1996), we need to understand what it is about these children (e.g., increased risk factors such as the presence of a concomitant phonological disorder) that might inhibit fluent responses.
The primary purpose of the present study was to assess children who stutter and the fluency of their spontaneous utterances that immediately follow clinicians’ utterances during treatment sessions. It was hypothesized that the children would be more likely to stutter following clinicians’ fast utterances and more likely to speak fluently following clinicians’ slow utterances, and that this likelihood would exceed expected levels through chi-square analyses. A secondary purpose was to explore the impact of risk factors (e.g., longer time since onset, awareness of stuttering, severity, genetic predisposition, and presence of disordered phonology) on the fluency responses of these children to their clinicians’ slow rate model.
The goal of speech therapy for the seven children during the time they were enrolled in the UWEC CCD was to facilitate normally fluent spontaneous speech. An “indirect” treatment approach was used (Zebrowski & Kelly, 2002), meaning that clinicians attempted to: (a) model a consistently slow speech rate (e.g., 160 - 200 SPM, or 2.66 - 3.33 SPS); (b) recast children’s stuttered utterances in a slow, fluent manner; and (c) keep the linguistic level simple (e.g., planning games that use listing and carrier phrases). Each of the clinician-child pairs was audio taped during twice-weekly therapy sessions held across three to twelve months between 1999-2001, using high quality audio recording equipment (Marantz & Corntek wireless microphone). For each child participant, two audio taped sessions were randomly selected so that they were distanced by at least a week.
The method of adjacent utterance pair analysis was borrowed from Yaruss and Conture (1995) who used it to compare articulatory speech rates of mothers to rates of their children who stutter. Adjacent utterance pairs (AUPs) were identified as a child’s utterance (e.g., a response) that immediately follows and is perceived to pertain to the clinician’s preceding utterance (e.g., Clinician says, “I like your shirt.” Child responds, “lt’s a blue one.”). Orthographic transcription and stopwatch timing (Guitar & Marchinkoski, 2001) of the clinicians’ utterances allowed for speech rate measures to be made in SPS. For inclusion as an AUP, a clinician’s utterance had to be intelligible, uninterrupted, 3+ words and fluent, and followed by a child utterance that met the same criteria, except that the child’s utterance had to be spontaneous and perceptibly fluent or disfluent. The children’s utterances were coded into four categories: perceptibly fluent utterances; utterances including one or more stutter; utterances including one or more between-word disfluency; utterances including at least one of each type of disfluency. If the child spoke in multiple utterances, only the first utterance was included in the AUP (Yaruss & Conture, 1995). The average length of clinicians’ utterances was 6.5 (Range = 3 - 28) syllables, the average length of the child’s utterances was 5.7 (Range = 3 - 14) fluent syllables, and the average AUP was 12.2 syllables.
For determining the intra- and inter-judge reliability of clinician speech rate measurement and fluency coding, six utterance pairs, representing 20% of the corpus, were randomly selected from each session for re-measurement. The first judge was the first author and the second judge was a speech-language pathologist who was trained in, the methods used by the first author. One hundred percent agreement was obtained within the first author’s judgments and between the first and second judges for coding the four utterance fluency categories. For syllable counts of child utterances, both intra- and inter-judge reliability for clinician utterances was 98.8%. For syllable counts of clinician utterances, intra-judge agreement was 100% and inter-judge was 99.7%. Taken with duration of fluent clinician utterance measures, this meant that intra-judge differences averaged +/- 0.08 SPS (Range = 0 - 0.42 SPS), and inter-judge differences averaged +/- 0.27 SPS (Range = 0 - 0.7 SPS). Both sets of differences are comparable to other stopwatch speech rate measures (Guitar & Marchinkoski, 2001). AUPS with clinician speech rate differences off by more than +/- 0.7 SPS between judges that could not be resolved to a smaller difference by consensus were excluded from data analysis (N = 7).
Clinicians’ speech rates averaged 4.23 SPS (Range = 1.2 - 13.7 SPS). Clinicians’ speech rates were significantly (Z = -5.641; p < 0.05) faster in the first therapy session (Mdn = 4.7 SPS; Range = 1.2 - 13.7 SPS) than in the second therapy session (Mdn = 3.6 SPS; Range = 1.4 - 9.0 SPS). Because chi-square analysis requires discrete categories of “clinician slow” vs. “clinician fast,” 25 of the 398 AUPS (6%) were excluded because clinicians spoke at a questionable rate between ideally slow (3.0 SPS) and too fast ( 3.33 SPS). Thus, 373 AUPS were submitted to Pearson chi-square, and Fisher’s exact chi-square test (two-tailed) was used for the individual dyad analyses because more than one-ï fth of the cell categories had expected frequency values of less than five in some of the participant dyads.
When individual chi-squares for each of the seven dyads were computed, two of the seven dyads showed statistically significant results. In one dyad, Clinician D and Child A, the percent of fast clinician and stuttered child utterances (23/55 = 42%) and the percent of slow clinician and fluent child utterances (18/55 = 32%) were significantly (p = 0.000) greater than would be expected by chance. Compared to the other six children, Child A showed the following clinical profile differences: (a) Relatively short time since onset of stuttering of 9 months; (b) Highest baseline stutter frequency, averaging 28 percent syllables stuttered (Range = 25 - 32); (c) Normal phonological development; and (d) Self-aware of stuttering. Child M was the only other child to also show normal phonology and self-awareness of stuttering, but he had a familial history of stuttering whereas Child A did not.
In contrast, another “dyad, M-Jl, showed significant results in the opposite direction that was hypothesized. That is, the percent of fast clinician utterances and stuttered child utterances (5/30 = 17%) and the percent of slow clinician and fluent child utterances (3/30 = 10%) were significantly (p = 0.023) less frequent than would be expected by chance. Thus, for J 1, 47% (14/30) of the AUPS were fast clinician and fluent child utterances and 27% (8/30) were slow clinician and stuttered child utterances. Clinical profile differences that child J 1 showed in comparison to other children were: (a) Relatively long time since onset of stuttering of 26 months; (b) Lowest baseline stutter frequency, averaging 7 percent syllables stuttered (Range = 5 - 10); (c) Presence of a concomitant phonological disorder; and (d) Unaware of his stuttering. Child B was the only other child to also show a phonological disorder and to be unaware of his stuttering, but Child B was the youngest participant at age 36 months, with the shortest time since onset of 3 months.
The other five of the seven dyads showed no significant differences between expected and observed probabilities in the 2x2 table of the two types of clinician speech rates and the two types of child fluency responses (Fisher’s p-levels ranged from 0.224 to 1.00).
The secondary purpose of the present study was to assess whether certain risk factors impact children’s fluency responses to their clinicians’ slow rate model. There were several clinical profile differences between Child A, who responded as hypothesized, and Child J 1, for whom about half of the AUPS produced with his clinician were “fast clinician - fluent child” in type. The first difference between these two children was in” time since onset of stuttering, where Child A’s was much shorter (9 months) than Child I 1 (26 months). Perhaps children who have been stuttering for a shorter period of time are more responsive to a slow rate model. However, this possibility is minimized due to the present finding that Child B, who had been stuttering for a shorter time (3 months) than Child A (9 months), showed no fluency differences in response to slow vs. fast clinician speech.
The second difference between Child A and J 1 was in the baseline frequency of stuttering or percent syllables stuttered (%SS). Present results could suggest that because Child A had a higher baseline of 28 %SS compared to Child J1 who showed only 7 %SS, Child A may simply have had more room for fluency improvement. Again, Child B was the closest comparison to J 1 other than in terms of younger age and shorter time since onset because he also showed a low stutter frequency baseline of 7 %SS, and yet there were no significant relationships between his fluency with his clinician’s speech rate categories. Thus, explaining Child A and Child J1’s differences only through suggesting that short time since onset and/or higher baseline %SS helps a child’s fluency responsiveness to slow rate model lacks support in the present findings. Also, Child A was estimated to be aware and Child 1 1 was unaware, but again, the likelihood of this being the most important clinical profile difference is minimized by the findings that two other children (Children E & M) were aware and did not respond as favorably as Child A. The two other children were unaware (Children J2 & S) and did not respond opposite as was hypothesized as did Child J 1. 208 Theory, research and therapy in fluency disorders
The most important difference appears to be the presence of a phonological disorder in Child J1, not present in Child A. As Child J1’s results suggest, presence of a phonological disorder may decrease the likelihood to stutter immediately following a fast clinician utterance. There were two other children (E & B) in the present study who presented with a concomitant phonological disorder and both showed no difference in their fluency when their clinicians spoke in fast VS. slow utterances. Child E was a girl who differed from the other two phonologically disordered stutterers in both gender and self-awareness of stuttering, differences which may help explain her fluency responsiveness to the slow rate model at least in the direction hypothesized, as found in Guitar et al.’s (1992) investigation of a girl who stuttered. The other child with a phonological disorder was Child B who was younger and had a shorter time since onset but was otherwise similar in profile to Child J1. It could be that this shorter time since onset in Child B counteracted the potential to find significant effects in the opposite direction hypothesized as with J1. Finding that a phonological disorder relates to a different overall fluency response corroborates previous research findings (Brown, 2002; LaSalle, 2003). Perhaps a phonological disorder is the primary communication disorder, thus stuttering onset occurs for some children in a secondary manner or as a side-effect. These children would then respond more fluently once intelligibility increases and slow rate would not immediately assist fluency. Alternatively, there may be no interaction between stuttering and phonology (Nippold, 2002). Perhaps children who both stutter and show a phonological disorder are a subgroup of children who primarily stutter, with progress that is complicated by issues of phonology and intelligibility, and this is why they do not appear to respond more fluently to a slow rate model. Further research is needed to provide support for these different perspectives.
Caveats in the present study suggest future directions. First, Child B received therapy for only nine sessions, which limited selection of sessions. Collection of adjacent utterance pairs should occur across a larger number of conversational samples, and these should be as standard as possible in terms of treatment duration across children who stutter. Second, Child E had two different clinicians due to university clinic restrictions. Future measures of clinicians’ speech rate should be standardized, even to the point of employing the same clinician for all the children who stutter recruited for a study such as this one. Greater restrictions on the person delivering the treatment approach would increase treatment fidelity. Also, to the degree that a clinician’s slow rate model could be fluency facilitating, as it was for Dyad D-A, future researchers should control potential generalization effects from linguistically constrained activities to spontaneous ones. Adjacent utterance pair analysis holds promise for understanding immediate fluency effects, but it also has limitations in terms of whether there are lasting effects of an adult model or linguistically constrained utterance effects across several child utterances or more, as is the case in prior research (Brown, 2002; Zebrowski et al., 1996).
In conclusion, a clinicians’ slow speech rate model is not pivotal for fluency facilitation and does not carry the same weight for increasing fluency in all children who stutter. It appears that a clinician slow rate model helps to facilitate some children’s fluency, especially those children who have a relatively short time since onset of stuttering, high stutter frequency, self-awareness of stuttering, and/or normal phonological devei opment. For children who are phonologically disordered in addition to stuttering, it appears from present findings that clinicians should not rely on their slow rate model to facilitate fluency. It is commonly accepted among stuttering experts that stuttering is widely variable. Therefore, it should also be accepted that the fluency-facilitating techniques used for treating preschoolers who stutter, such as a slow speech rate model, may vary as well.
Ambrose, N. G., and Yairi, E. (1994).. The development of awareness of stuttering in preschool children. Journal of Fluency Disorders, 19, 229-245.
Brown, R. (2002). Responses of preschoolers who starter to clinicians ‘slow speech rate. Unpublished master’s thesis, University of Wisconsin-Eau Claire, Eau Claire, WI. Section 4. Outcomes and Efiicacy of Interventions 209
Couture, E.G. (2001). Stuttering: Its nature, diagnosis, and treatment. Needham Heights, MA: Allyn & Bacon. Guitar, B. (1998). Stuttering: An integrated approach to its nature and treatment. (2”" ed.). Baltimore, MD: Williams & Wilkins.
Guitar, B., Kopff Schaefer, H., Donahue-Kilburg, G. & Bond, L. (1992). Parent verbal interactions and speech rate: A case study in stuttering. Journal of Speech and Hearing Research, 35, 742- 754.
Guitar, B. & Marchinkoski, L. (2001). Influence of mother’s slow speech rate on their children’s Speech rate. Journal of Speech, Language, and Hearing Research, 44, 853-861.
Helmer, L. (1995). Efiects of maternal speech rate and MLU reduction on speech fluency of children who stutter. Unpublished master’s thesis, University of Wisconsin-Eau Claire, Eau Claire, WI.
LaSalle, L. (2003, August). Clinicians slow speech rate model and children’s fluency. Paper presented at the meeting of the 4”â World Congress on Fluency Disorders, Montreal, Canada.
Nippold, M. (2002). Stuttering and phonology: Is there an interaction? American Journal of Speech- Language Pathology, 11, 99-110.
Nippold, M. & Rudzinski, M.A. (1995). Parents’ speech and children’s stuttering: A critique of the literature. Journal of Speech and Hearing Research, 38, 978-989.
Stephenson Opsal, D. & Bernstein Ratner, N. (1988). Maternal speech rate modification and childhood stuttering. Journal of Fluency Disorders, 13, 49-56.
Yaruss, J .S. & Conture, E.G. (1995). Mother and child speaking rates and utterance lengths in adjacent fluent utterances: Preliminary observations. Journal of Fluency Disorders, 20, 257- 278.
Yaruss, J.S., LaSa1le, L.R., and Conture, E.G. (1998). Evaluating stuttering in young children: Diagnostic data. American Journal of Speech Language Pathology, 7, 62-76.
Zebrowski, P. (1995). Temporal aspects of the conversations between children who stutter and their parents. Topics in Language Disorders, 15, 1-7.
Zebrowski, P.M. & Kelly, E.M. (2002). Manual of stuttering intervention. New York, NY: Singular - Thompson Learning.
Zebrowski, P.M., Weiss, A.L., Savelkoul, E.M. & Hammer, C.S. (1996). The effect of maternal rate reduction on the stuttering, speech rates and linguistic productions of children who stutter: Evidence from individual dyads. Clinical Linguistics and Phonetics, 10, 189-286.