Gather Cole 1993

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

European Journal of Psychology oj Education

1993, Vol. VIII, n? 3, 259-272


© 1993, I.S.P.A.

Phonological Working Memory: A Critical


Building Block for Reading Development
and Vocabulary Acquisition?
Susan E. Gathercole
University of Bristol, UK

Alan D. Baddeley
MRC Applied Psychology Unit, Cambridge, UK

In this article we review findings from a recent longitudinal study


of the contribution of phonological working memory to vocabulary
acquisition and reading development. A total of 80 children were tested
initialiy at school entry at the age of four years, and were tested in three
further waves at ages 5, 6, and 8 years. The results indicate that
phonological memory skills constrain vocabulary growth during the first
year or so in school but that subsequently, vocabulary knowledge is a
pacemaker in the development relationship with memory. Phonological
memory skill in prereading children was found to be significantly linked
with scores on a reading test at age 8 which encourages the use of a
phonological recoding strategy. The theoretical and practical implications
of the findings, and important areas for future research, are discussed.

Introduction

In this article we take the opportunity to review some of our recent work on the contribution
of phonological working memory to the development during the early school years of two

The research reported in this article was supported by the Medical Research Council and the Economic and Social
Research Council. Requests for reprints should be sent to Susan E. Gathercole, University of Bristol, 8 Woodland Road,
Bristol BS8 1TN, UK.
260 S. E. GATHERCOLE & A. D. BADDELEY

language processing skills: reading and vocabulary acquisition. As measures of these skills
are often used by both educationalists and psychologists as convenient indices of a child's
intellectual development, the nature of the developmental constraints on vocabulary and reading
development is an issue of exceptional theoretical and practical importance. Our work suggests
that children's phonological memory skills as they start school represent one such constraint.
Convergent findings from longitudinal and cross-sectional studies, and from experimental
comparisons of children with specific language-related difficulties, indicate that young children's
skills at temporarily storing phonological material may place significant limits on the ease
and rate with which they both develop their vocabulary knowledge and learn to read.
We discuss four specific issues concerning the developmental role of phonological memory
in language development that have guided our recent research. These issues concerns links
between phonological memory and developmental language disorders, vocabulary acquisition,
reading development and phonological awareness, and each issue is addressed in a separate
section below. Before considering in detail our findings and their theoretical and practical
implications, however, our theoretical framework for conceptualising children's phonological
memory skills is outlined.

The development of phonological working memory

According to the working memory model introduced by Baddeley and Hitch (1974) and
developed by Baddeley (1986), there is one component of memory that is specialised for the
temporary maintenance and processing of verbal material. This component is termed the
'phonological loop'. The phonological loop system consists of two sub-components: the short-
term phonological store. and an articulatory rehearsal process. Information is held in
phonological form in the short-term store, from which it decays with time unless the material
is rehearsed. The effect of rehearsal is to refresh the phonological traces in the short-term
store, and so to offset decay. In this way, the short-term store and the rehearsal process are
semi-independent processes which function in concert with one another in memory tasks.
However, the work discussed in the present article does not attempt to distinguish between
the phonological short-term store and rehearsal as sources of individual differences in children's
phonological memory skills. We therefore prefer to use the more neutral term phonological
working memory to refer to the parts of working memory, such as the phonological loop,
that are involved in the temporary storage of verbal material.
Phonological working memory undergoes dramatic development during childhood, with
memory span for words increasing almost three-fold between the ages of 4 and 11 years
(Nicolson, 1981; Hulme, Thomson, Muir, & Lawrence, 1984). Much of this developmental
increase appears to be attributable to the increasing rapidity of subvocal articulatory rehearsal
as children become more experienced language users. As a consequence of this increase in
rehearsal rate, more items can be actively maintained in the phonological store and so the
functional capacity of the phonological loop expands.
Although the phonological working memory skills of preschool children are limited in
comparison with older children and adults, they show a considerable degree of individual
variation. In fact we have recently shown that in children as young as 2 years of age, performance
on short-term memory tasks such as digit span (involving the immediate recall of a series
of spoken digits) and nonword repetition (in which the task is simply to repeat an unfamiliar
phonological form such as 'biffle') are closely associated with one another, indicating systematic
differences in phonological memory ability (Gathercole & Adams, 1993). The roots of differences
in phonological memory skill in very young children such as these are as yet unknown, although
recent evidence reveals a close link with speech production abilities (Adams & Gathercole,
in prep.). For the moment, though, the important point is that some preschool children are
much better at temporarily storing phonological material than others.
PHONOLOGICAL WORKING MEMORY 261

Developmental language disorders: The consequences of poor phonological working


memory skills?

The first indication for us that these individual differences in children's phonological
memory skills may directly influence both reading development and vocabulary acquisition
arose from working with a group of children with disordered language development (Gathercole
& Baddeley, I990a). Also known as specific language-impaired, children with this developmental
disorder have deficits in a wide range of language abilities in the absence of any detectable
physical, social or emotional disturbances. A large body of research has documented the multiple
language deficits of language disordered children (e.g., Aram & Nation, 1975; Benton, 1978;
Leonard, 1982; Stark & Tallal, 1981). Hallmarks of the disorder include poor vocabulary growth,
language comprehension difficulties, and impaired literacy acquisition.
Our principal interest was in the phonological working memory skills of language disordered
children. Deficits of verbal short-term memory in such populations had already been established
(Graham, 1980; Kirchner & Klatzky, 1985; Locke & Scott, 1979; Wiig & Semel, 1976), although
previous work had not been guided by a detailed model of short-term memory. We wanted
to know what particular aspect of working memory was deficient in children with developmental
disorders of language. Accorclingly, a group of language clisordered children attencling a Language
Unit for remedial instruction in language skills were selected for the study. The children were
aged between 6 and 10 years, and on standardised tests each child performed well below the
appropriate level for their age on measures of vocabulary knowledge, reading and language
comprehension. This group were individually matched with two control groups: a group matched
on nonverbal intelligence of the same age, and a group of children on average aged 2 years
younger than the language disordered children but who were matched on vocabulary and reading
measures.
A range of standard tests of phonological working memory was given to each group of
children. These included nonword repetition, and serial recall of lists of short words, long
words and phonologically similar words. On each measure, the language disordered children
were significantly impaired compared with both control groups. The poor memory abilities
of the disordered group were therefore not simply a consequence of their low level of language
development , as this was matched with the younger control children. Interestingly, though,
the language disordered children were no less sensitive to variables known to influence the
phonological loop such as word length and phonological similarity, despite their lower level
of overall recall. It therefore appears that their phonological loop systems were intact, but
were of reduced capacity.
The most striking results arose with a test of nonword repetition , in which the child hears
an unfamiliar phonological form (such as 'blonterstaping') and has to repeat it immediately.
The nonwords ranged in length from one to four syllables, and the accuracy of each repetition
was scored. This test appears to be particularly sensitive to phonological memory skills
(Gathercole, Willis, Emslie, & Baddeley, 1991). The language disordered children were much
less accurate at nonword repetition, scoring on average only 52010 correct, whereas both control
groups correctly repeated on average 84010 of the nonwords. Moreover, scores on this test perfectly
discriminated the disordered children from the controls. Figure 1 summarises the repetition
performance of the three groups as a function of the number of syllables in the nonwords.
The language clisordered children were as accurate at repeating the one- and two-syllables items
as the controls, but their performance declined dramatically for the lengthier stimuli. The
phonological loop system is known to be sensitive to the temporal duration of memory items,
with recall declining as a function of the number of syllables in memory items in both adults
and children (Baddeley, Thomson, & Buchanan, 1975; Hulme et al., 1984). The differential
sensitivity of the language disordered children to the length of the nonwords is therefore entirely
consistent with the view that their phonological loop systems have a reduced capacity.
The non word repetition test administered to the language disordered children and their
control was also given to a large unselected cohort of children in the Cambridge longitudinal
262 S. E. GATHERCOLE & A. D. BADDELEY

study described in the section below (Gathercole, Willis, Emslie, & Baddeley, 1992). We therefore
have normative data on this test. Applying these data, the nonword repetition scores of the
language disordered children (mean score 21.8, with an average age of eight years in the group)
corresponded to those obtained for normal four-year old children (mean score 22.0) from the
longitudinal study. In nonword repetition performance, therefore, the language disordered group
were developmentally delayed on average by about four years. This finding indicates that the
phonological memory deficits of the disordered group were even greater in magnitude than
their developmental language problems: on standardised tests of vocabulary, reading and
comprehension, the mean delays of the group ranged between 18 and 24 months.

100

c
0
+=:
o+:;
80
Q)
a.
Q)

-~

0
Q)
~
~
60
0
0
Q)

-
--9--
0>
~
40 Nonverbal controls
c
Q) • Verbal controls
~
Q)
a. 20
• Language disordered
group
c
~
Q)
:2
0
1 2 3 4
Number of syllables in nonword

Figure 1. Mean accuracy of nonword repetition of the language disordered group (LDG) and
the two groups of control children (Verbal Controls), as a function of the number of syllables

Nonword repetition deficits of similar magnitude in samples of language disordered children


have also been reported by Bird (1991) and Taylor, Lean, and Schwartz (1989). In our sample,
there was no indication that the disordered group had either phonological discrimination
problems or slow articulatory output rates. It therefore appears that a deficit in phonological
storage, rather than perceptual or articulatory output skills, formed the basis of the repetition
deficits of these children (see, also, Taylor et al., 1989).
One possible explanation for the profile of deficits of the language disordered children
is that their poor phonological memory skills are instrumental to their failure to develop language
normally. We were interested specifically in whether the memory impairments had contributed
PHONOLOGICAL WORKING MEMORY 263

to these children's difficulties in vocabulary acquisition and in learning to read. In order to


explore these hypotheses directly, we conducted a longitudinal study over a 4-year period which
allowed us to evaluate developmental associations between phonological memory skills and
both vocabulary and reading development.

Phonological working memory as a longitudinal predictor of vocabulary acquisition

We have now completed a longitudinal study of 118 children who were tested initially
at the age of 4, within two months of joining infant school in the Cambridge area of England.
Subsequently, the children were tested at ages 5 and 6 years, and then finally at age 8. At
each age, measures of phonological memory skill, nonverbal intelligence, receptive vocabulary
and reading were taken. The principal measure of phonological memory skill was a test of
nonword repetition (Gathercole & Baddeley, 1990a; Gathercole et al., 1991). Complete data
from each of the four waves of the study is available for 80 children (see Gathercole & Baddeley,
1989, and Gathercole, Willis, Emslie, & Baddeley, 1992, for further details).
It seemed very likely to us that the children's skills at temporarily maintaining phonological
material, assessed by the nonword repetition test, might act as a significant constraint on their
abilities to construct a long-term representation of the phonological form of a new word.
Quite simply, if a child cannot retain an item sufficiently well to be able to repeat it immediately,
he or she will stand little chance of being able to retrieve that item from long-term memory
some time later. And certainly, the findings that language disordered children have impaired
phonological memory skills and poor vocabulary growth is entirely consistent with this view.
So too was the finding that a neuropsychological patient with a highly selective deficit of
the phonological loop was unable to learn novel phonological forms such as the vocabulary
of a foreign language (Baddeley, Papagno, & Vallar, 1988).
Correlations between the nonword repetition and vocabulary tests were significant at each
of the 4 waves of the longitudinal study: the correlation coefficients were .559 at age 4, .524
at age 5, .562 at age 6 and .284 at age 8. These associations were consistent with the hypothesis
that phonological memory ability constrains vocabulary learning, although it is noteworthy
that the association between repetition and vocabulary scores is significantly reduced at age
8 relative to the three earlier waves. Correlations cannot, however, establish the direction of
causality between two abilities. In order to assess whether there was a causal association between
phonological memory skills and vocabulary acquisition, we turned to the technique of cross-
lagged correlations (Crano & Mellon, 1978). Across each one-year time interval within the
four years of the project, the forward partial correlations between nonword repetition and
vocabulary scores, and between vocabulary and nonword repetition scores (adjusted for age,
nonverbal intelligence and scores on the later measure at the earlier point in time, in each
case), were compared. For example, the partial correlation between nonword repetition at 4
and vocabulary at age 5 was compared with the partial correlation between vocabulary at
age 4 and repetition at age 5. According to the logic of cross-lagged correlations), causal
developmental relationships will be reflected in a stronger forward than backward association
in time between the causal factor and its consequent (see Figure 2).
The patterns of associations across the time panels are summarised in Figure 2. The causal
underpinning of the developmental association between nonword repetition and vocabulary
appears to shift dramatically at age 5. Between 4 and 5 years, the pacemaker in the developmental
relationship is clearly the phonological memory measure. Whereas the partial correlation between
repetition scores at 4 and the vocabulary measure at 5 is highly significant, the converse
association between vocabulary at 4 and repetition at 5 is nonsignificant. Between 5 and 6,
however, it is the partial correlation between vocabulary and later repetition which is significant,
while the converse association between nonword repetition at 5 and vocabulary at 6 is near
zero. The same pattern of a forward association between vocabulary and later nonword repetition
scores and a nonsignificant converse relationship is reflected between the ages of 6 and 8.
264 S. E. GATHERCOLE & A. D. BADDELEY

Note too that one other measure of phonological memory skill was taken at each wave, and
that these data showed the same shift in the causal relationship with vocabulary during the
course of the longitudinal study (Gathercole et at, 1992).

Age 6 Age 8
Nonword ,,
repetition -, , .018
,,
,,
,,
,

Vocabulary .089
Figure 2. Cross-lagged partial correlations (with age, nonverbal intelligence scores on the outcome
variable one year earlier controlled) between nonword repetition and vocabulary scores
(Gathercole et at, 1992). Lines shown in bold denote partial correlations that are significant
greater than the cross-lagged

The results from the first two waves of the study are entirely consistent with the notion
that phonological working memory plays a critical role in establishing long-term memory
representations of the phonological forms of new words. They also fit well with experimental
findings that normal 5-year old children with good phonological memory skills learn new
names of toy animals more rapidly than children of low memory skills but comparable nonverbal
intelligence (Gathercole & Baddeley, 1990b). Recent findings by Service (1992) that the nonword
repetition abilities of Finnish children predicted their achievement in second language learning
over two years later similarly point to phonological memory skills representing an important
developmental constraint on word learning. Although the nature of normal individual variation
in these skills is not yet entirely clear, it does appear that the discriminability and persistence
of the phonological traces in memory will directly influence whether those traces will be held
in more stable long-term storage and so link semantically with stored representations of the
new word's conceptual attributes,
Beyond the age of 5, though, vocabulary knowledge appears to become the driving force,
becoming a very good predictor of subsequent repetition ability. There are a number of reasons
why this developmental change may occur. It may be that children learn to use their existing
vocabulary knowledge to mediate the repetition of nonwords, perhaps by accessing words which
share some of the phonological segments of the nonword to alleviate the burden on temporary
phonological storage (Gathercole et aI., 1991). This may explain the emergence of a causal
link between vocabulary knowledge and subsequent nonword repetition performance. Why,
though, does the phonological memory measure not maintain its strong positive association
with future vocabulary scores? Perhaps other constraints on vocabulary development become
more important during middle childhood. Children's phonological memory skills develop very
rapidly during the first years at school, so that for most children temporary storage factors
may no longer limit their word learning. Also, it seems likely that as children mature, the
words that they learn become more abstract in nature, showing less direct correspondence
to tangible physical events. The child's abilities to understand the conceptual attributes of
novel words may therefore become a more critical word learning constraint. A further emerging
factor may be the child's experience of reading. Recent work by Cunningham and Stanovich
(1991) suggests that vocabulary acquisition in late childhood is influenced by the extent of
the child's exposure to print. There is indirect evidence for this factor in our longitudinal
PHONOLOGICAL WORKING MEMORY 265

study, too: reading ability at age 6 correlated highly (r = .681) with vocabulary scores two
years later.
So, phonological memory constraints appear to be important in natural vocabulary learning
in the early school years, although they become less critical during middle childhood as memory
skills improve and other factors come into play. This does not, however, mean that phonological
working memory has no role to play in word learning in late childhood and adults. Work
with both older children and adults has shown that phonological memory is important in
situations where the long-term learning of the novel phonological form cannot be readily
mediated by more semantically-based strategies, for example in foreign vocabulary learning
where the foreign word is very distinctive in sound structure from words within the first language
(Papagno, Valentine, & Baddeley, 1991; Service, 1992).

Phonological working memory as a longitudinal predictor of reading development

Links between reading ability and phonological working memory skills during the primary
school years are already well-established (see Jorm, 1983; and Wagner & Torgesen, 1987, for
reviews). Most of the evidence, though, is based on cross-sectional correlational associations
between reading and memory measures; in a typical experiment, lower scores of a test of
serial recall of verbal material are found in 8-year old children of poor than good reading
ability (e.g., Shankweiler, Liberman, Mark, Fowler, & Fischer, 1979). Such studies are of value
in establishing the deficits associated with poor reading achievement, but cannot distinguish
between alternative causal hypotheses. For example, it could either be that phonological memory
problems retard normal reading development, or that reading achievement itself stimulates
phonological memory development.
In order to distinguish between these alternative theoretical accounts of the positive link
between phonological memory skills and reading development, we have investigated whether
phonological memory skill before children started learning to read can be used as a predictor
of later reading achievement, by studying the children from the Cambridge longitudinal study
who could not read when they were first tested. Although there have been a small number
of longitudinal studies which have looked at the developmental associations between phonological
memory and reading, there are methodological problems with each. In one case the majority
of children at initial test had started to read (Ellis & Large, 1988) and in another, no reading
test was reported to have been administered at selection making it impossible to determine
which children had not yet started to read (Mann & Liberman, 1984). Wagner & Torgesen
(1987) point out that unless the children are prereaders at initial test, the possibility that a
significant link between phonological memory skill and later reading achievement is mediated
by an early causal contribution of reading success to memory development cannot be rejected.
Therefore in our study, data are only reported for children who failed to correctly identify
any of the words in the reading test administered in the first wave (Gathercole, Willis, Emslie,
& Baddeley, in prep.) (see Table I).
The children were tested at regular intervals until the age of 8, when they were given
the range of reading tests illustrated in Table 1. Other measures taken from the cohort included
a test of arithmetic ability and a language comprehension test. Our primary interest was in
whether the measure of phonological memory skill taken before the children started to read,
which was score on the nonword repetition test, was a significant predictor of reading
achievements at age 8. This was estimated using a fixed-order multiple regression procedure
which allowed us to control for differences in age and nonverbal intelligence at age 4. Further
regression analyses were also performed in which differences in prereading vocabulary scores
were also controlled. This second set of analyses was performed in order to maintain
comparability of statistical procedure with an earlier influential longitudinal study reported
by Bradley and Bryant (1983) which showed that awareness of rhyme in prereading children
was significantly associated with reading ability at age 8. However, as discussed in the previous
266 S. E. GATHERCOLE & A. D. BADDELEY

section, phonological memory skill appears to play an important role in vocabulary development
in the early school years (Gathercole, Willis, Emslie, & Baddeley, 1992). Partialling out the
variance in later reading scores associated with the early vocabulary measure will therefore
lead to significant underestimations in the strength of the developmental relationship between
prereading phonological memory skill and reading achievement.

Table 1
Partial correlations between nonword repetition scores at age 4 and outcome measures at age
8 (additional r shown)

Test Example of a line from the test

British Abilities Scales! said water bird wood running


Neale Tese The bird hopped up to my window
Primary Reading Test3 He opened his _ _ to shout (asleep chest mouth ears pocket)

Note. I Elliot (1983); 2 Neale (1989); 3 France (1981).

Table 2
Examples of items from the Reading Tests used in the longitudinal study

With age and nonverbal With age, nonverbal intelligence


Measure intelligence controlled and vocabulary controlled

Reading measures
BAS .014 .001
Neale accuracy .015 .001
Neale comprehension .017 .000
Primary reading .088** .049*
Other measures
Arithmetic! .000 .001
Comprehension' .010 .005

Note. ~ Weschler (1974); 2 Bishop (1982); • p < .05; •• p < .01.

The results are summarised in Table 2. When obvious confounding factors were taken
into account, the nonword repetition measure at age 4 was a significant predictor of only
one measure of age 8 reading ability, the Primary Reading test. Performance on the other
reading measures, and on the tests of arithmetic and language comprehension abilities, showed
no indication of a separable positive link with earlier scores on this phonological memory
test. The predictive relationship between nonword repetition and reading development therefore
shows a surprising degree of specificity. Not only was the forward association restricted to
reading performance, ruling out the possibility that the repetition measure is an intellectual
facilitator of a more general nature, but its influence was restricted to one particular measure
of reading.
The important question is, of course, why there is this high degree of specificity to the
developmental relationship between prereading phonological memory skills and later reading
achievement. In attempting to answer this question, it is necessary to consider in detail the
nature of the three reading tests, illustrated in Table 1. The factor that we consider most clearly
distinguishes the Primary Reading test from the other measures is the extent to which a strategy
PHONOl.DGICAL WORKING MEMORY 267

of phonological recoding unfamiliar words is likely to be successful for the child. Consider
first the single-word reading test, taken from the British Abilities Scales. No linguistic context
is provided for the words the child has to read, so the test probably provides a relatively pure
measure of the child's word decoding skills. Moreover, many of the words in the test involve
irregular or inconsistent spelling patterns: by our calculations, at least 500/0 of the first 30
words. A strategy of phonological recoding from print into sound will therefore not be very
successful in this test. And even for the regular words, the child would need to successfully
recode all of the letters in order to generate the correct pronunciation in the absence of any
contextual cues. The partial recoding strategy which many young children adopt, focussing
on the initial one or two letters of a word only, is not very likely to yield the right word
in this test. Performance on this reading measure therefore seems likely to largely reflect the
child's sight vocabulary; that is, the words whose visual configurations the child had already
learned prior to the test.
The Neale test is rather different. It involves the child reading aloud short stories which
are illustrated in pictures. This test therefore allows context to be used to guide the recognition
of unfamiliar words. However, as many of the words in the sentences are not highly predictable
from the preceding context, a sophisticated guessing strategy in which the child matches a
partial phonological code with plausible candidates for the word may not be particularly effective
unless either the word is regular or the child uses the strategy very efficiently. Performance
on this test therefore seems likelyto reflect both the child's sight vocabulary and use of context,
and to a lesser extent phonological recoding skills.
In contrast, the design of the Primary Reading test is highly compatible with the use
of a phonological recoding strategy. The phonological form of the target word is provided
either by the accompanying picture in the early test items, or by the rest of the sentence in
the later items. The child's task is to choose which of the five alternative words corresponds
to the target word. For half of the items in the test, none of the foils share the same initial
letter (or phoneme) of the target words. The degree of letter overlap between the target and
foils in the remaining items is variable, but in no cases would the child have to accurately
graphemically recode more than the first three phonemes of the target (or, phonologically
recode more than the first three letters in the five words) in order to make the correct selection.
Thus a partial recoding strategy for unfamiliar words is likely to be very successful in this test.
We therefore consider it likely that a strategy of partial phonological recoding of unfamiliar
words will be much more successful in the Primary Reading test than the BAS test, a view
that fits well with the relatively low correlations between the tests. This analysis raises the
interesting possibility that phonological memory skills constrain the development of a
phonological recoding strategy. We have considered a number of ways in which this
developmental constraint may operate. Phonological memory may be used to temporarily store
the phonological segments generated when a child maps successive graphemes onto their
phonological equivalents, prior to output of the blended phonological sequence (Baddeley,
1978). So, children with small phonological memory capacities may be relatively less successful
at developing this recoding strategy. Alternatively, the ease of long-term learning of
correspondenc:es between letters and sounds may depend on the child being able to retain
adequate temporary representations of the sound segments, in the same way that short-term
phonological memory appears to limit the long-term learning of novel words (Baddeley &
Gathercole, 1992; Gathercole & Baddeley, 1993). In this way, poor phonological memory skills
may retard the: acquisition of the rule-based knowledge needed to exploit the regularities between
the spoken and written forms of language.

Phonological working memory and phonological awareness

These results suggest that children's phonological memory skills as they start school have
an important influence both on their vocabulary acquisition in the near future and possibly
268 s. E. GATHERCOLE & A. D. BADDELEY

on the development of a phonological recoding strategy for reading some years later. An
important issue that we have not considered so far concerns the relationship between
phonological working memory and another skill known to be closely related to reading success:
phonological awareness. This term refers to the ability to respond to the explicit sound structure
of spoken language, and is reflected in a wide range of measure which include asking the
child to tap out the number of component sounds in words (Liberman, Shankweiler, Fischer,
& Carter, 1974), to add or delete sounds at the beginning or end of words (Morais, Cary,
Alegria, & Bertelson, 1979) and to judge which words in a list do not share a common sound
sequence (Bradley & Bryant, 1978).
Bradley and Bryant (1983) found that awareness of rhyme in prereading children, assessed
by asking the child to identify which of three words does not share the same sequence of
sound (e.g., 'peg, mat, cat'), is a significant predictor of later success in reading and spelling
(see, too, Bryant, Maclean, Bradley, & Crossland, 1990). Scores on the rhyme task before the
children started to read accounted for between 4 and 100/0 of unique variance in reading and
spelling performance when the children were aged 8, and so were of a similar predictive strength
to the nonword repetition measure used in the Cambridge longitudinal study over the same
interval of time. On the basis of these findings, it has been suggested that children need to
have awareness of the componential structure of spoken language in order to capitalise upon
the regularity of spelling-sound relationships.
A pressing question which is not yet fully answered is whether the two types of measure,
of phonological awareness and phonological working memory, have the same or dissociable
influences on reading development. It certainly appears likely that good phonological memory
skills are necessary if a 4-year old child is to succeed on Bradley and Bryant's (1983) rhyme
awareness task, in which the child chooses which word of three does not rhyme with either
of the other two. The phonological structures of each word need to be held while the child
makes at least two pairwise comparisons. As the memory span for the average 4-year old
child is somewhere between two and three words (Hulme et al., 1984), it follows that at least
some of the children will be unable to perform the task reliably because their phonological
memory capacity will be exceeded.
This analysis certainly fits well with the positive correlations between measures of
phonological awareness and phonological working memory skill in children during the early
years of school. In a cross-sectional study (Gathercole, Willis, & Baddeley, 1991), we found
that rhyme awareness scores in four-year old children were positively associated with nonword
repetition but not with digit span scores (r == .274, P < .05, and r == .228, n.s., respectively).
Most of these children failed to read any words on the single-word reading test of the British
Abilities Scales (Elliot, 1983). For the five-year old children, the links between rhyme awareness
and phonological memory measures were much stronger: r == .597, P < .001, with repetition,
and r == .519, p < .001, with digit span. The majority of this older subject group (34 of
a total of 51) were able to read one or more words in the single-word reading test. However,
despite the associations between the awareness and memory scores in the five-year olds, the
measures of rhyme awareness and phonological memory (nonword repetition and digit span)
had dissociable links with reading and vocabulary ability. In particular, whereas scores on
the phonological memory tasks accounted for significant unique portions of variance in
vocabulary knowledge, awareness of rhyme did not. Also, whereas the rhyme awareness measure
was closely linked with reading scores in both age groups, the association between phonological
memory skill and reading emerged only for the older, five-year old, children. Similarly, Ellis,
and Large (1988) found differentiable time courses to the developmental associations between
reading and the two types of phonological processing ability.
On present evidence, it looks as though phonological awareness and phonological working
memory abilities do share a common phonological processing component, perhaps because
phonological memory is needed in tasks designed to assess awareness of the sound structure
of language. With respect to reading and vocabulary development, however, it looks as though
the contributions of the two types of skills are, to some degree, unique (see Gathercole &
PHONOWGICAL WORKING MEMORY 269

Baddeley, 1993). One possibility is that phonological awareness is a prerequisite for distinguishing
the component sounds in spoken words, whereas phonological working memory is used both
in learning the associations between these phonemes and the letters, and in storing sound
segments generated by the child when attempting to phonological recode an unfamiliar printed
word. In other words, awareness and working memory may make different contributions and
place differentiable constraints upon the component processes involved in phonological recoding.
This view is consistent in principal at least with recent analyses which have identified several
critical stages in the development of an efficient phonological recoding strategy in reading
(e.g., Byrne & Fielding-Bamsley, 1989).
Proper resolution of the debate over whether phonological working. memory and
phonological awareness make common or differentiable contributions to reading development,
though, can probably only be achieved by a direct empirical evaluation of the longitudinal
contribution of the two phonological processing skills to literacy acquisition. To this end, a
project currently in progress in our laboratory is tracing the development of rhyme awareness,
phonological working memory, and many other verbal and nonverbal skills during the preschool
and early school years. We hope that the results of this study will answer at least some of
the remaining questions about the nature of the developmental contributions of phonological
awareness and memory to literacy acquisition.

Practical implications
Evidence from both normal and language disordered children suggests that phonological
memory skills are directly related to the child's abilities to learn new words, and to develop
an important reading strategy. A child whose ability to temporarily retain phonological material
is inadequate therefore seems likely to be at risk of poor vocabulary growth and developmental
reading difficulties. The practical implications seem straight forward: promoting phonological
memory skills in children with below-average abilities in this domain should yield important
benefits and may even offer immunity from serious disorders of language development. The
approach could be used either remedially, to improve the language abilities of children with
developmental language problems that have already been identified, or preventatively, in younger
children of low phonological working memory ability.
The idea of training phonological memory in children as a way of boosting their language
development has not so far received much attention, and as yet little is known about the nature
of suitable training techniques. Traditional methods of promoting verbal short-term memory
skills have involved recruiting resources from other parts of memory, and using them to avoid
reliance upon the limited phonological memory system (e.g., Paivio, 1971; Ericsson & Chase,
1982). These techniques are therefore not designed to directly enhance phonological working
memory.
One possibility is that phonological working memory may be boosted by enriching the
linguistic environment of the child. Mann (1984) suggested that methods such as providing
practice in naming letters and objects, in remembering spoken sequences and nursery rhymes,
and at listening to stories, may all be useful for children at risk of developing reading problems.
Indeed, they may directly enhance the efficiency of phonological encoding, and hence of retaining
phonlogical material in working memory. Another way of promoting phonological memory
skills may be to provide children with practice in repeating unfamiliar phonological forms,
which may encourage them both to perceive the essentially combinatorial basis of the phonetic
structure of language (Lindblom, 1989), and to develop articulatory gestural skills which may
themselves mediate speech perception (Liberman & Mattingley, 1985).

Summary
Our longitudinal study of phonological working memory skills and their associations with
270 S. E. GATHERCOLE & A. D. BADDELEY

important aspects of language development such as vocabulary and reading acquisition has
yielded a rich database. There do appear to be close links between working memory and both
vocabulary learning and reading development, but the nature of the relationships is more
dynamic than we expected (in the case of vocabulary growth) and also more specific (in the
case of reading). This, we suspect, is one of the most powerful features of the longitudinal
study: the opportunities for discovering hitherto unsuspected developmental relationships, and
hence for provoking theoretical development, appear to be notably greater than in cross-sectional
studies.

References

Adams, A -M., & Gathercole, S. E. (in prep.), Developmental associations between speech production and ph onological
working memory.
Aram, D. M., & Nation, J. E . (1975). Patterns of language behaviour in children with developmental language disorders.
Journal of Speech & Hearing Research, 18, 229-2411.
Baddel ey, A D. (1978). Working memory and reading. In P. A Kolers, M. E. Wrolstad, & H. Bouma (Eds.), Processing
of Visible Language, Vol. 1, (pp. 355-370). New York: Plenum Press.
Badd eley, A. D. (1986). Working Memory. Oxfo rd: Oxford University Press.
Baddeley, A D., & Gathercole, S. E. (1992). Learning to read: Th e role of th e phono logical loop. In J. Alegr ia, D.
H olander, 1. J. de Mo rais, & M. Radeau (Ed s.), A naly tic Approach es to Human Cognition, (pp, 153-168). Elsevier.
Baddeley, A. D., & Hitch, G. J. (1974). Worki ng mem ory. In G. Bower (Ed .), The Psychology of Learning and Motivation,
Vol. 8, 47-90.
Baddel ey, A. D., Papagno, C, & Vallar, G. (1988). When long-term learning depend s on sho rt-term sto rage, Journal
of Memory & Language, 2 7, 586-5%.
Baddeley, A D., Thomson , N., & Bucha nan, M. (1975). Wo td length a nd the structure of short-term mem ory. Journal
of Verbat Learning & Verbal Behavior, 14. 575-589 .
Benton , A. (1978). The cognitive functi on ing of children with develop men tal a phasia. In M. Wyke (Ed.), Developmental
Dysphasia. New York : Aca demic Press.
Bishop , D. (1982). Test fo r the Reception of Grammar. Available fro m the auth or at M RC A pp lied Psychology Unit,
Cam bridge, UK.
Bird, 1. (1981). The auditory percepti on, phonem ic awareness and literacy sk ills of phonologically disordered children.
Unpublish ed PhD th esis from the Universi ty of Ma nchester.
Bradley, L., & Bryant , P_ E _ (1978). Diffi : ulties in auditory organization as a pos sible cau se of reading backwardn ess.
Nature, 2 71, 746-747.
Bradley, L., & Bryant , P. E. (1983). Categorizing sounds and learnin g to read - a causal co nnection. Nature, 301, 419-420.
Bryant , P. E., Maclean, M., Bradl ey, L., & Crossland, J. (1990). Rhyme and alliteration, phoneme detection , and learni ng
to read. Developmental Psychology. 26, 429-438.
Byrne, B., & Fielding-Barnsley, R. (1989). Phonemic awareness and letter knowledge in the child' s acqu isition of the
alphabetic principle. Journal of Educational Psychology, 81, 313-321.
Cra no, W. D., & Mellon, P. M. (1978). Ca usa l influence of teachers' expecta tio ns on childre n' s academic performance:
A cross-lagged panel analy sis. Journal of Educational Psychology, 70, 39-49.
Cunningham, A. E., & Stanov ich, K. E. (1991). 'Iracking the unique effects of print exposure in children: Associations
with vocabulary, general knowl edge and spelling. Journal of Edu cational Psychology, 83, 264-274.
Elliott , C. D. (1983). British Abilities Scales. Windsor : NFER-Nelson Publi shing Com pa ny Ltd,
Ellis, N., & Lar ge, B. (1988). The ear ly stages of reading: A longitudinal stu dy. Applied Cognitive Psycho logy, 2, 47-76.
Ericsson, K. A, & Chase, W. G. (1982). Exceptional memory. A merican Scientist, 70, 607-615.
Fran ce, N. (1981). Primary Reading Test (Revised Edition). NFER-Nelson P ublishing Compa ny Ud: Windso r.
Gat hereo le, S. E., & Adam s, A -M. (1993). Ph onological working memo ry in very young chi ldren. Developmental
Psychology, 29.
Gat hercole, S. E., & Badd eley, A D. (1989). Evaluation o f the role of phon ological STM in the development o f vocabulary
in children: A lon gitudinal study. Journal of Memory & Language, 28, 200-213.
Gat herco le, S. E., & Badd eley, A. D. (19903). Ph onological m emory deficits in language disordered childre n: Is the re
a causa l con nect ion? Journal of Memo ry & Language, 29, 336-360.
Gathereole, S. E., & Baddel ey, A D. (1990b). The role of phonological m emor y in vocabulary acqu isition : A stud y of
you ng children learning arb itrary nam es of toys. British Journal of Psychology, 81, 439-454.
PHONOLOGICAL WORKING MEMORY 271

Gathercole, S. E., & Baddeley, A. D. (1993). Working Memory and Language Processing. Hove: Lawrence Erlbaum
Associates.
Gathercole, S. E., Willis, C. S., & Baddeley, A. D. (1991). Differentiating phonological memory and awareness of rhyme:
Reading and vocabulary development in children. British Journal of Psychology, 82, 387-406.
Gathercole, S. E., Willis, C., Emslie, H., & Baddeley, A. (1991). The influences of number of syllables and word-likeness
on children's repetition of nonwords. Applied Psycholinguistics, 12, 349-367.
Gathercole, S. E., Willis, c., Emslie, H., & Baddeley, A. (1992). Phonological memory and vocabulary development
during the early school years: A longitudinal study. Developmental Psychology, 28, 887-898.
Gathercole, S. E., Willis, C. S., Emslie, H., & Baddeley, A. D. (in prep.), Phonological memory skill in prereading children
as a predictor of later reading achievement.
Graham, N. C. (1980). Memory constraints in language deficiency. In F. Margaret Jones (Ed.), Language Disability in
Children, (pp. 69-94). Baltimore: University Park Press.
Hulme, C., Thomson, N., Muir, C., & Lawrence, A. (1984). Speech rate and the development of short-term memory
span. Journal of Experimental Child Psychology, 38, 241-253.
Jorm, A. (1983). Specific reading retardation and working memory: A review. British Journal of Psychology, 74, 311-342.
Kirchner, D., & Klatzky, R. L. (1985). Verbal rehearsal and memory in language-disordered children. Journal of Speech
& Hearing Research, 28, 556-565.
Leonard, L. B. (1982). Phonological deficits in children with developmental language impairment. Brain & Language,
16, 73-86.
Liberman, A. M., & Mattingley, I. G. (1985). The motor theory of speech perception revisited. Cognition, 21, 1-36.
Liberman, I. Y., Shankweiler, D., Fischer, F. w., & Carter, B. (1974). Explicit syllable and phoneme segmentation in
the young child. Journal of Experimental Child Psychology, 18, 201-212.
Lindblom, B. (1989). Some remarks on the origin of the phonetic code. In A. von Euler, I. Lundberg & G. Lennerstrand
(Eds.), Brain and Reading, (pp, 2744). Stockton Press.
Locke, J. L., & Scott, K. K. (1979). Phonetically mediated recall in the phonetically disordered child. Journal of
Communication Disorders, 12, 125-131.
Mann, V. A. (1984). Longitudinal prediction and prevention of early reading difficulty. Annals of Dyslexia, 34, 117-136.
Mann, V. A., & Liberman, I. Y. (1984). Phonological awareness and verbal short-term memory. Journal of Learning
Disabilities, 17, 592-599.
Morais, J., Cary, L., Alegria, J., & Bertelson, P. (1979). Does awareness of speech as a sequence of phones arise
spontaneously? Cognition, 7, 323-331.
Neale, M. (1989). Neale Analysis of Reading Abilities (Revised British Edition). Windsor: NFER Nelson.
Nicolson, R. (1981). The relationship between memory span and processing speed. In M. Friedman, J. P. Das, & N.
O'Connor (Eds.), Intelligence and Learning, (pp. 179-184). Plenum Press.
Paivio, A. (1971). Imagery and Verbal Processes. New York: Holt, Rinehart and Winston.
Papagno, c., Valentine, T., & Baddeley, A. (1991). Phonological short-term memory and foreign-language vocabulary
learning. Journal of Memory & Language, 30, 331-347.
Service, E, (1992). Phonology, working memory and foreign-language learning. Quarterly Journal of Experimental
PSYChology, 45A, 21·50.
Shankweiler, D., Liberman, I. Y., Mark, L. S., Fowler, C. A., & Fischer, F. W. (1979). The speech code and learning
to read. Journal of Experimental Psychology: Human Learning & Memory, 5, 531-545.
Stark, R., & 'Iallal, P. (1981). Selection of children with specific language deficits. Journal of Speech and Hearing Disorders,
46, 114-122.
Taylor, H. G., Lean, D., & Schwartz, S. (1989). Pseudoword repetition ability in learning-disabled children. Applied
Psycholinguistics, 10, 203-219.
Wagner, R. K., &: Torgesen, J. K. (1987). The nature of phonological processing and its causal role in the acquisition
of reading skills. Psychological Bulletin, 101, 192-212.
Wechsler, D. (~974). Manual for the Wechsler Intelligence Scale for Children - Revised. New York: Psychological
Corporation,

Wiig, E., & Semel, E. M. (1976). Language Disabilities in Children and Adolescents. Columbus, OH: Charles E.
Merrill.

Key words: Phonological memory, Reading, Vocabulary

Received: April 1993


272 S. E. GATHERCOLE & A. D. BADDELEY

Susan Elisabeth Gathercole. University of Bristol, 8, Woodland road, Bristol BS8 ITN, United Kingdom.

Current theme of research:

Working memory and language development

Most relevant publications in the field of Educational Psychology:

Gathereole, S. E., & Baddeley, A. D. (1989). Evaluation of the role of phonological STM in the development of vocabulary
in children: A longitudinal study. Journal of Memory & Language, 28, 200-213.
Gathercole, S. E., & Baddeley, A. D. (I99Oa). Phonological memory deficits in language disordered children: Is there
a causal connection? Journal of Memory & Language, 29, 336-360.
Gathercole, S. E., & Baddeley, A. D. (199Ob). The role of phonological memory in vocabulary acquisition: A study of
young children learning arbitrary names of toys. British Journal of Psychology, 8/ , 439-454.
Gathercole, S. E., Willis, C., Emslie, H., & Baddeley, A. (1992). Phonological memory and vocabulary development
during the early school years: A longitudinal study. Developmental Psychology, 28, 887-898.
Gathercole, S. E., & Baddeley, A. D. (1993). Working Memory and Language Processing. Hove: Lawrence Erlbaum
Associates.

Alan David Baddeley. MRC, Applied Psychology Unit, 15 Chaucer Road, Cambridge CB2 2EF, United Kingdom.

Current theme of research:

Working memory, reading and speaking acquisition

Most relevant publications in the field oj Educational Psychology:


Baddeley, A. D. (1986). Working memory. Oxford : Oxford University Press.
Baddeley, A. D. (1990). Human Memory: Theory and Practice. Hove, Sussex: Lawrence Erlbaum Associates.
Baddeley, A. D. (in press). Your Memory: A Users' Guide (second edition). Harmondsworth, Middx: Penguin Books.

You might also like