It is well established that the span of immediate memory for verbal items is lower in children than in adults. A range of factors has been proposed to account for this developmental increase, including rehearsal strategies, chunking, and speed of encoding (see, e.g., Dempster, 1981, for a review). More recently, insights have emerged from applying the concept of the phonological loop. The principal findings concern the word-length effect.
As described earlier, studies of adults indicated that immediate serial recall of words of different length can be predicted from the time it takes to articulate the items (Baddeley et al., 1975). This empirical relationship suggested that recall is limited by the rate at which subvocal rehearsal can be used to refresh decaying phonological traces. Consistent with such an interpretation, faster speakers tend to have higher spans (Bad-deley et al., 1975). Developmental studies of the word-length effect produced results that conform to the same empirical relationship as in adults (Nicolson, 1981; Hitch & Halliday, 1983; Hitch, Halliday & Littler, 1993; Hulme, Thomson, Muir & Lawrence, 1984). For example, Hulme et al. (1984) found that a single linear function described the relationship between recall of auditorily presented lists and articulation rate from 4 years of age through to adulthood. That is, the improvement in amount recalled during development can be predicted from the increase in speed of articulation. This empirical relationship is consistent with the concept of a time-based phonological loop and suggests continuity in its functional characteristics over a large part of development.
However, developmental data on the word-length effect raise a problem for interpreting the relationship between span and articulation rate in terms of speed of rehearsal. This is because other evidence has established that rehearsal develops around the ages of 6-9, implying that 4-year-olds are unlikely to be rehearsing (Flavell, Beach & Chinsky, 1966; see Kail, 1984, for an overview). Consistent with the late development of rehearsal, the correlation between individual differences in span and speech rate observed in adults is not seen in 4-year-olds (Gathercole, Adams & Hitch, 1994) and articulatory suppression does not disrupt immediate recall in 5-year-olds (Henry, 1991a). However, an ingenious study by Cowan, Day, Saults, Keller, Johnson, and Flores (1992) suggested that differences in rehearsal time are not the only cause of differences in the recall of long and short words. By varying the arrangement of short and long words within lists, Cowan et al. showed that the greater time taken to output long words contributes to their poorer recall. This is consistent with the decay of information in the phonological loop as longer output delays allow more time for forgetting. Indeed, in their connectionist implementation of the phonological loop, Burgess and Hitch (1999) assumed that articulation time has the same effect during recall as in rehearsal.
In a developmental study, Henry (1991b) reported data that point to the importance of output delays for observing the word-length effect in young children. She found that 5-year-olds were sensitive to word length in serial recall but not probed recall. Henry argued that as her probe task involved recalling only a single item, differences in output delays for long and short words were minimal. In contrast, serial recall resulted in large differences in output delays for lists of long and short words. When older children were investigated, Henry found that word length affected performance in both probed and serial recall, as expected from their development of rehearsal strategies. In the light of findings such as these,
Gathercole and Hitch (1993) suggested that the developmental growth of auditory-verbal short-term memory (STM) reflects the increased speed with which information can be read out from the phonological store during rehearsal or recall. Thus, the concept of the phonological loop has helped to describe developmental change in verbal short-term memory, and in turn, developmental data have helped clarify the operation of the phonological loop.
An important further question concerns the function of the phonological loop in development. Much research has focused on its role (or equivalently, that of auditory-verbal STM) in learning to read (see, e.g., Wagner & Torgerson, 1987). As the evidence for such a role is well-established, it will not be summarized here. Other work has found evidence for the involvement of the loop in cognitive skills such as arithmetic calculation (Hitch et al., 1987; see also Logie, Gilhooly & Wynn, 1984). However, it appears that a major function of the phonological loop is in acquiring native vocabulary (Baddeley, Gathercole & Papagno, 1998).
Gathercole and Baddeley (1989a, 1989b; see also Gathercole et al., 1992) reported a longitudinal study of the role of the phonological loop in vocabulary acquisition in children aged from 4 to 8. They took performance on a nonword repetition task as their principal measure of phonological STM. This task measures the ability to repeat individual nonwords containing different numbers of syllables and is simpler for children to comprehend than word or digit span. Cross-lagged correlational analyses in which differences in age and nonverbal intelligence were partialled out showed that nonword repetition at age 4 was a good predictor of vocabulary scores at age 5. Furthermore, this correlation was stronger than the backward association between nonword repetition at age 5 and vocabulary at age 4. These observations suggest that the phonological loop underpins vocabulary learning during this phase of development. When the children were older, similar correlational analyses indicated that nonword repetition is itself influenced by vocabulary skills, which suggests a complex developmental pattern. Research on language-impaired children provides further evidence for a link between the phonological loop and vocabulary. For example, Gathercole and Baddeley (1990) assessed nonword repetition and vocabulary skills in 8-to-9-year-old language-impaired children. As expected, the language-impaired children had poorer vocabulary scores than controls matched on nonverbal intelligence. More interesting, the language-impaired children were significantly poorer at nonword repetition than a group of younger children matched on language ability, which suggests that a deficit in the phonological loop was responsible for their language-learning difficulties. Thus, taken together, evidence from both normal and abnormal development indicates a close relationship between the phonological loop and native-vocabulary acquisition.
It seems that the loop is also involved in acquiring vocabulary in a second language. This has been demonstrated most dramatically in a neuropsychological patient with a selective impairment of auditory-verbal digit span. The patient showed no word-length effect in immediate serial recall and a phonemic-similarity effect only for spoken stimuli, which suggests an impairment of the phonological store (Vallar & Bad-deley, 1984a). She showed normal learning of a list of word-word pairs that was repeated several times. However, unlike normal controls, she was unable to learn a list consisting of word-nonword pairs (Baddeley, Papagno & Vallar, 1988). Thus the patient was able to learn new combinations of familiar items, but damage to the phonological store prevented her from learning novel phonological forms.
Because of the obvious importance of vocabulary learning in the development of language and cognition, it is tempting to conclude that this is the primary role of the phonological loop. Interestingly, however, the original concept of the phonological loop made no reference to learning. A step toward rectifying this omission is the inclusion of a learning mechanism in Burgess and Hitch's (1999) network model of the phonological loop, though it remains to be seen whether this model is adequate for simulating vocabulary learning. The loop may also serve other functions besides vocabulary acquisition. We have already mentioned research on reading and arithmetic skills, and we note also evidence that the loop may provide a backup store in comprehension, for example when "first pass'' parsing fails (see, e.g., Baddeley & Lewis, 1981; Caplan & Waters, 1999).
Was this article helpful?