توانایی تکرار غیرکلمه ای کودکانی که لکنت زبان دارند: یک مطالعه اکتشافی

Past research has suggested that children who stutter (CWS) may have less well-developed language skills than fluent children, and that such relative linguistic deficiencies may play a role in precipitating their disfluencies. However, data to support this position are primarily derived from results of standardized diagnostic inventories, which are originally designed to identify frank language impairment. Nonword repetition has emerged as a more sensitive measure of children's linguistic abilities. In this exploratory study, eight CWS (mean age 5:10, range 4:3–8:4) were compared to eight normally developing children (ND) (mean age 5:9, range 4:1–8:4) in their ability to repeat the nonwords of the Children's Test of Nonword Repetition. CWS performed more poorly than NS on measures of Number of Words Correct and Number of Phoneme Errors at all nonword lengths, although statistical differences were observed only for 3-syllable nonwords. When lexical stress of the nonwords was varied to a non-English stress pattern, all participants repeated the stimuli with less accuracy, and the CWS again exhibited more errors than NS. Fluency for the CWS group did not change systematically with increasing nonword length. These preliminary findings are interpreted in light of a number of extant theories of the underlying deficit in childhood stuttering. We conclude that children who stutter may have diminished ability to remember and/or reproduce novel phonological sequences, and that further investigation into this possibility may shed light on the emergence and characteristics of childhood stuttering.

Research correlating stuttering with linguistic demands falls broadly into two categories: one which contrasts the linguistic abilities of children who stutter (CWS) and those who do not, and one which correlates the existence of dysfluencies in individual children's utterances to specific linguistic demands. Both have produced findings suggestive of a link among linguistic capacities, demands and fluency of speech-language production. Results of standardized language tests have shown that CWS score lower than their nonstuttering peers (Byrd & Cooper, 1989; Murray & Reed, 1977; Ryan, 1992). Further, language skills appear to predict which children will recover spontaneously from stuttering and which children will stutter chronically; children with stronger language skills appear to have a higher likelihood of recovery (Yairi, Ambrose, Paden, & Throneburg, 1996). In addition, relative but subclinical depression in lexical and syntactic skills have been found in children (Bernstein Ratner, & Silverman, 2000; Wall, 1980) and adults who stutter (Homzie, Lindsay, Simpson, & Hasenstab, 1988; Prins, Main, & Wampler, 1997; Watson et al., 1994) when compared to fluent peer groups. However, not all studies that compared stuttering and fluent populations find differences in their language or phonological abilities (Nippold, 1990 and Nippold, 2002), and interpretation of differences has been controversial (Watkins & Johnson, 2004). Aside from the real question of whether such differences truly exist, one potential reason for this may be the typical methodology of studies that contrast groups of stuttering and nonstuttering children (Watkins & Johnson, 2004). For example, among other concerns, they tend to employ standardized language tests or spontaneous speech sampling. As has been noted (Bernstein Ratner, 1997), standardized language tests are primarily designed to identify frank language disability for diagnostic and therapeutic purposes, and so are unlikely to provide the more precise discrimination between groups that may be required when either subtle depression of skills or weakness in a very specific domain of language exists. Trends reported in studies that have not found significant differences between groups on a variety of measures suggest that this may be the case (Miles & Bernstein Ratner, 2001). Spontaneous language sampling, while a sensitive descriptive measure of children's expressive abilities, is not necessarily a valid way of distinguishing groups of children's language abilities: “… standardized tests are designed to tell whether a child is different from other children. Speech-sample analysis, on the other hand, is not constructed psychometrically for this purpose” (Paul, 2001, p. 319). Many of the measures that flow from language sampling (e.g., MLU, TTR), have been criticized as having large ranges of normal performance at an array of ages that impede their ability to discriminate between or among groups of children (Eisenberg, Fersko, & Lundgren, 2001; Klee, 1992 and Watkins and Kelly, 1995). For this reason, researchers and clinicians have continued to search for more sensitive measures that are capable of distinguishing between groups of children, particularly on the basis of language ability. A promising measure that has gained strength in the research literature over the past decade is nonword repetition. 1.1. Phonological coding in children and the nonword repetition task (NRT) A growing literature has examined the phonological aspects of language learning in children. Clearly, there is a cognitive/semantic component to learning new words or structures, but there also is evidence for a separate system known as the phonological loop. The phonological loop is part of working memory, a limited-capacity system that supports both the storage and processing of information; it is a temporary storage area for incoming verbal information (e.g., Baddeley, 1986; Gathercole, Hitch, Service, & Martin, 1997; Gathercole, Willis, Baddeley, & Emslie, 1994). The phonological loop is thought to contain two components: a phonological store and an articulatory rehearsal mechanism, and it provides temporary storage of speech code while the main working memory in the central executive controls “modality-free” reasoning or cognition (Baddeley, 1986). Children are presumed to use the phonological loop to store word forms in addition to using their current vocabulary knowledge when learning new words (Baddeley, Gathercole, & Papagno, 1998; Gathercole et al., 1997). Phonological short-term memory has been extensively studied in typically-developing, language-delayed and specifically language-impaired (SLI) children (e.g., Edwards & Lahey, 1998; Gathercole et al., 1994 and Sahlén et al., 1999). It is believed that repetition of novel stimuli (nonword repetition), which is thought to require the temporary storage of an unfamiliar phonological sequence, relies on the phonological loop, and that success on the task depends on the capacity of the short-term storage area of the phonological loop (Gathercole et al., 1994). Numerous studies have examined the performance of children with SLI on nonword repetition tasks. Relatively consistent findings have correlated nonword repetition task ability with measures of language ability. Nonword repetition ability is highly correlated with vocabulary and reading skills in both children with SLI (Gathercole et al., 1994) and in normally developing children (Gathercole, Service, Hitch, Adams, & Martin, 1999). It also is correlated with comprehension of grammar (Gathercole et al., 1994). An impressive finding by Dollaghan and Campbell (1998) was that a nonword repetition task distinguished children enrolled in language intervention from language-normal children with a higher degree of accuracy than a norm-referenced language test. Edwards and Lahey (1998) also found that children with SLI were less accurate than their normal-achieving counterparts on nonword repetition tasks. On the other hand, they found little correlation between nonword repetition and speech motor skills, implying that the salient connection between SLI and nonword repetition is language-based rather than motor-based. Based on the types of errors noted in their sample, they hypothesized that nonword repetition is correlated with ability to form or to hold phonological representations in working memory; they also found a greater correlation with expressive language abilities than with language comprehension. Since that time, Ellis Weismer et al. (2000) have further extended the finding of poorer nonword repetition ability in children with SLI, and have in fact suggested that it might serve as a language- and culture-free assessment measure for identifying children with SLI. Recently, in an assessment of multiple potential markers for SLI, Conti-Ramsden (2003) found nonword repetition and past tense marking to be the best markers for identifying the condition. There also are some data linking phonological memory deficits and stuttering. Bosshardt (1993) found that adults who stutter performed more poorly on a serial short-term memory task than normally fluent adults, and interpreted his results as suggesting that adults who stutter have slower phonological encoding and rehearsal times. Ludlow, Siren, and Zikria (1997) found that adults who stutter demonstrate more difficulty learning novel phonological sequences than fluent speakers. In addition, although the relationship between articulation proficiency and dynamic measures of speech encoding is unclear, Melnick, Conture, and Ohde (2003) reported that preschool CWNS exhibited a significantly negative correlation between their speed of speech reaction time and score on the Goldman-Fristoe, but no such relation was observed with CWS. Thus, there is an apparent and established link between stuttering and diminished language ability, as well as a smaller body of evidence suggesting a link between stuttering and phonological encoding deficits. Given this, the nonword repetition paradigm provides a particularly suitable way to probe further whether or not children who stutter have weaker-than-normal language systems, and, if so, whether or not the systems that support nonword repetition are involved. 1.2. Linguistic task and fluency performance As noted earlier, beyond using linguistic measures to discriminate between the abilities of stuttering and fluent populations, language measures appear to predict the frequency and location of stuttered events. Within individuals who stutter, more specifically, there is evidence that as length and syntactic complexity of an utterance increase, so too does the amount of stuttering (Gaines, Runyan, & Meyers, 1991; Zackheim & Conture, 2003). It also has been shown that increasing syntactic complexity alone, independent of length of an utterance, increases the number of dysfluencies (Bernstein Ratner & Sih, 1987). Furthermore, adults who stutter show decreased speech motor stability (i.e., stability of the lower lip during articulation) when syntactic complexity, but not length, of utterances increases (Kleinow & Smith, 2000). In fact, decreased motor speech stability is seen with increased syntactic complexity in normally fluent adults and children as well (Maner, Smith, & Grayson, 2000). The decrease in stability with increased syntactic complexity suggests that the extra demand placed on a speaker by increased syntactic complexity imposes demand on the speech system; the hypothesis is that in PWS, this can also precipitate dysfluencies. A number of recent models hypothesize linguistic impairment in some domain, including those of phonology and prosody, as a portion of the underlying factors that precipitate and/or maintain stuttering (Au-Yeung & Howell, 1998; Karniol, 1995 and Packman et al., 1996; Postma & Kolk, 1993). The role of syllabic stress in precipitating stuttering in both adults (Prins, Hubbard, & Krause, 1991; Wingate, 1984) and children (Natke, Sandrieser, van Ark, Pietrowsky, & Kalveram, 2004) has been a matter of both empirical debate and theoretical speculation. However, little testing of such hypotheses has been systematically carried out in children using anything other than published standardized test results or spontaneous language data. Thus, there is evidence both of depressed language function in PWS and of a correlation of stuttered dysfluencies with certain linguistic, particularly syntactic, variables. In terms of language capabilities, although results of standardized language tests show some differences in stuttering populations, results are neither consistent nor very illuminating about the nature, or even the degree, of the deficits. Moreover, although the evidence from studies examining specific linguistic tasks points to some relationship with stuttering, it is not at all clear what specific functions are impaired in PWS. That is, there is a notable lack of evidence pointing to a specific cognitive/linguistic deficit in PWS which might be at the root of their depressed language, not to mention their stuttering (Bernstein Ratner, 1997). Hence, the identification of a specific ability which could be shown to be depressed in those who stutter might shed light on the specific difficulties underlying stuttering, and therefore on the viability of certain models of stuttering. This research examined the performance of stuttering children on nonword repetition tasks to answer the following questions: Do children who stutter exhibit more errors than do children who do not stutter on a nonword repetition task? Does stuttering increase as length of the nonword increases? Given some recent models of stuttering that posit an underlying prosodic encoding deficit in stuttering children (Karniol, 1995 and Packman et al., 1996), does the imposition of non-English lexical stress differentiate repetition accuracy between children who stutter and normally fluent children? Is variation in lexical stress more likely than English-like stress to cause stuttering?

If there is an underlying linguistic deficit that plays a role in precipitating or maintaining stuttering, we would suggest that experimental tasks, rather than standardized diagnostic test batteries, will be needed to discover the nature and extent of the deficit. The use of standardized diagnostic instruments is more likely to uncover true co-morbid disability of fluency and either language or articulation. Furthermore, such tests are unlikely to identify subtle and perhaps quite task-limited areas of relative impairment in linguistic encoding and/or retrieval. This exploratory study found a trend for children who stutter to perform more poorly on a nonword repetition task than normally fluent children. Moreover, for all children, accurate nonword repetition performance decreased when the stress pattern of the nonwords did not conform to the usual English stress patterns, but fluency was not affected systematically. Future research should seek to confirm the generalizability of these preliminary findings as well as to explore how phonological encoding deficits may be related to the phenomenology that characterizes instances of stuttering.