In line with the original presentation of nonword repetition as a measure of phonological short-term memory (Gathercole & Baddeley, 1989), the theoretical account Gathercole (2006) puts forward in her Keynote Article focuses on phonological storage as the key capacity common to nonword repetition and vocabulary acquisition. However, evidence that nonword repetition is influenced by a variety of factors other than item length has led Gathercole to qualify this account. In line with arguments put forward by Snowling, Chiat, and Hulme (1991), one of Gathercole's current claims is that nonword repetition and word learning are constrained by “the quality of temporary storage of phonological representations, and this quality is multiply determined.” Phonological storage is not just a quantity-limited capacity.
In this Commentary, I take Gathercole's (2006) account a step further. I propose that the factors contributing to the quality of temporary storage and their relative contribution change in the course of development. After making the case for a developmental trajectory of nonword repetition, I argue that this leads to a more convincing explanation for findings that are challenging for the storage-based account.
EARLY PHONOLOGICAL SENSITIVITY
Long before children produce words, and can be asked to repeat nonwords or to perform phonological awareness tasks, they deploy skills in recognizing and storing details of phonological input. In the first year of life they show acute sensitivity to the properties and frequency of rhythmic patterns and segmental combinations in phonological input, and use these as cues to segmenting words for storage (see, e.g., Gerken, 2001; Jusczyk, 1997; Jusczyk, Cutler, & Redanz, 1993; Jusczyk, Houston, & Newsome, 1999; Mattys & Jusczyk, 2001; Morgan & Saffran, 1995; Saffran, Aslin, & Newport, 1996; Saffran & Thiessen, 2003; Thiessen & Saffran, 2003). The implication is that children come to nonword repetition with previously established phonological representations and the phonological processing skills that underpin these. This gives rise to what I call the “early phonological sensitivity hypothesis.” According to this hypothesis, variations in children's early sensitivity to phonology and their resulting phonological representations will influence their ability to repeat nonwords. Children with significantly reduced sensitivity to phonological details will have deficits in nonword repetition. They will also have deficits in lexical segmentation and recognition, and may ultimately present with language impairments (see Chiat, 2001, for elaboration of this claim).
Direct evaluation of these predictions requires longitudinal studies investigating relations between infants' responses to speech input at age 6–12 months and their performance on nonword repetition at age 2–3 years. Such studies are in their early stages, but initial results reported by researchers at the University of Wales, Bangor, are promising (Bywater, 2004; Vihman, Keren-Portnoy, & Armstrong, personal communication; see also Friedrich & Friederici, 2005). A different source of evidence for the early phonological sensitivity hypothesis is the finding that factors influential in early phonological development are also influential in nonword repetition. As pointed out above, infant studies have revealed very early sensitivity to prosodic structure, and use of prosodic structure to segment the stream of speech. In our studies of word and nonword repetition in typically developing children at age 2–4, and children referred with concerns about language development at age 2.5–4 (Chiat & Roy, in press; Roy & Chiat, 2004), we controlled for prosodic structure of items and included analysis of prosodic effects. We found that the overall performance of the clinically referred group was very much poorer than that of the typically developing group. Errors of syllable omission were also significantly more frequent in the clinically referred group. However, both groups of children showed effects of prosodic structure on syllable omission that could override effects of length. They were many times more likely to omit unstressed syllables that precede the stress than those that follow the stress in two-syllable items. Furthermore, loss of prestress syllables in two-syllable items was nearly double that of poststress syllables in longer, three-syllable items. In line with our results, a study of Swedish 5-year-olds with language impairment (Sahlén, Reuterskiöld-Wagner, Nettelbladt, & Radeborg, 1999) found that these children omitted six times more prestress syllables than poststress syllables. These findings show that differential sensitivity to elements of prosodic structure is robust in typically and atypically developing children, and plays a role in their nonword repetition. It is certain weak syllables and segmental details within prosodic structures that are most vulnerable, and that give rise to variation in performance.
GROWING CONTRIBUTIONS OF VOCABULARY
It is now generally accepted that children's growing vocabulary is an emergent source of support for nonword repetition, supplementing the support of earlier emerging phonological representations discussed above. In her paper, Gathercole points out that the relationship between nonword repetition and vocabulary changes with age. Previous studies have also indicated that the direction of influence changes, with abilities tapped by nonword repetition only predicting vocabulary in the early stages of development (Gathercole, Willis, Emslie, & Baddeley, 1992; Jarrold, Baddeley, Hewes, Leeke, & Phillips, 2004). The nature of relations between vocabulary, nonword repetition, and phonological awareness at later stages of development is the subject of considerable debate, which Gathercole addresses. However, little attention has been paid to other influences on vocabulary development that will indirectly influence nonword repetition. Vocabulary acquisition relies not only on the establishment of phonological forms. Word forms must be attached to word meanings. Hence, strengths in identifying meanings and connecting these to phonology will influence vocabulary growth. So will exposure to vocabulary. These child internal and external influences on vocabulary growth will indirectly influence the potential facilitatory effects of vocabulary on nonword repetition. This increasingly complex interplay of influences on nonword repetition opens up the possibility of increasingly divergent trajectories and outcomes.
INSIGHTS FROM A DEVELOPMENTAL PERSPECTIVE
This analysis of the developmental trajectory may account better for the two findings that are puzzling for the storage-based account of nonword repetition and that lead Gathercole (2006) to supplement this account. The first is the finding that children with specific language impairment (SLI) are considerably more impaired on nonword repetition than serial recall tasks, even when stimuli in the two tasks contain identical syllables. The finding of differential effects on items of the same length cannot be explained by a purely quantitative constraint on storage. Gathercole suggests that the selective deficit in nonword repetition may be due to fast rates of transmission of the acoustic signal, but does not elaborate on this possibility. What distinguishes the stimuli in the two tasks is prosodic structure: a nonword, unlike a string of syllables, forms a rhythmic unit. According to the early phonological sensitivity account, children with SLI have reduced ability to register weak syllables and segmental details within a prosodic structure. This is in keeping with the disproportionate difficulty they show repeating polysyllabic nonwords, whose constituent syllables vary in stress, compared with strings of monosyllabic nonwords whose constituent syllables are all stressed.
The second troubling finding for the storage-based account is that some children with marked impairments in nonword repetition and serial recall at 5 years present with age-appropriate vocabulary and language abilities at 8 years. If nonword repetition and vocabulary acquisition crucially depend on the same phonological storage capacity, this mismatch between an early measure of phonological memory and later vocabulary is puzzling. In her Keynote (2006), Gathercole infers that a phonological storage deficit may not be sufficient to disrupt language development, and suggests that there must be a further deficit, proposing “working memory” as a candidate.
Analysis of the developmental trajectory may yield alternative accounts of the observed differences in outcome. The above analysis leads to at least three reasons why some children with poor nonword repetition at age 5 may have intact vocabulary and language at age 8. First, even though they attained comparable scores to children with SLI on a nonword repetition test at 5 years, they may have differed in the severity and/or pervasiveness of their phonological deficit. Second, these children may differ in relevant nonverbal abilities. Strengths in nonverbal development could help children discover word meanings and thereby support their vocabulary development. Third, they may differ in their capacity to register links between phonology and semantics: stronger connections between phonology and semantics may mitigate the effects of weak phonological processing on vocabulary and syntax.
These reinterpretations illustrate why it is important to recognize that nonword repetition has a developmental trajectory, and to identify the forces operating at different stages in that trajectory. Unusual patterns of performance will then lead us to explore which hypothesized skills and resulting representations are out of step with each other relative to their typical trajectory.
Furthermore, a developmental perspective opens up the possibility that nonword repetition may be “the most effective predictor of language learning ability” (Gathercole, 2006) at some but not all stages in the trajectory. Other measures, such as sentence repetition, may become better predictors. However, that is another story.
