Persistence of phonological awareness deficits in older children with dyslexia

Persistence of phonological awareness deficits in older children with dyslexia

ANGELA J. FAWCETT and RODERICK I. NICOLSON Department of Psychology, University of Sheffield, UK

ABSTRACT. Three groups of children with dyslexia, with mean age 8, 13 and 17 years, together with three groups of normally achieving children matched for age and IQ with the dyslexic groups, undertook tests of sound categorization and phoneme deletion. The design allowed comparison not only across chronological age but also across reading age. The children with dyslexia performed significantly worse even than their reading age controls on both tasks. Indeed, overall performance of the t 7 year old children with dyslexia was closest, but inferior, to that of the 8 year old controls. Since the sound categorization task was designed to minimize working memory load, the results extend previous findings on the phonological awareness deficits in dyslexia by dissociating the deficit from memory load and by showing that it persists at least into late adolescence.

KEY WORDS: Adolescence, Dyslexia, Phonological awareness deficit

INTRODUCTION

Developmental dyslexia is characterised by an otherwise inexplicable failure to learn to read in children of at least average intelligence. One of the major theoretical achievements of dyslexia research in the past decade was the demonstration that many of these reading-related deficits are attributable to a disorder of phonological processing (Vellutino 1979; Bradley & Bryant 1983; Snowting et al. 1986; Stanovich 1988). Much of the most compelling data points to a deficit in phonological awareness.

Some of the strongest evidence for a phonological awareness deficit derives from a seminal study of sound categorization deficits in children with dyslexia (Bradley & Bryant 1978), in which the experimenter presented a series of words, such as sun, sock, see and rag (with rag the odd one out on the basis of the first letter), and the subject had to say which is the odd one out. On all the tasks presented, the children with dyslexia were significantly worse at judging which was the odd one out than were younger children who had reached the same level in their reading. In a later training study (Bradley & Bryant t983), which involved pre-readers with deficits in rhyming skills, they investigated the comparative effectiveness of training in rhyming and alliteration skills ( 'sound categorization') versus training in semantic categorization. The group who were trained in sound categorization made significantly more progress in reading, but equivalent progress in mathematics, providing evidence for a causal link between early phonological awareness skills and

Reading and Writing: An Interdisciplinary Journal 7: 361-376, 1995. © 1995 Kluwer Academic Publishers. Printed in the Netherlands.

362 A. J. FAWCETT AND R. I. N I CO L SO N

acquisition of reading. A series of further studies have confirmed and extended Bradley and Bryant's results both for normal readers (e.g., Hatcher, Hulme & Ellis 1994) and for children with dyslexia (e.g., Rack 1985; Ackerman, Dykman & Gardner 1990) and have linked the ability to rhyme to knowledge of nursery rhymes (e.g., MacLean, Bryant & Bradley 1987; Bryant et al. 1990), and developed procedures to train up phonological awareness skills (e.g., Treiman & Baron 1983; Fox & Routh 1984; Sanchez & Rueda 1991; Hurford 1990; Ball & Blachman 1991; Tangel & Blackman 1992; Felton 1994).

Phonological awareness is a metalinguistic skill involving knowledge about the sounds that make up words. Two levels of phonological awareness may be distinguished - syllabic knowledge and phonemic knowledge, At the syllabic level, which is the simpler, awareness is measured by a variety of tasks, including tapping out the number of syllables, counting syllables, and deleting syllables. The development of awareness at the phonemic level (e.g., that ca t i s / c / / a / / t / ) is far more difficult to acquire (Adams 1990), and is measured by counting phonemes, dividing words up into a series of phonemes, deleting phonemes, and substituting phonemes. The ability to divide words into onsets and rims (e.g., that ca t may be broken down into/c/, the onset, and/at/, the rime) falls midway in difficulty" between syllabic and phoneme awareness.

In terms of the acquisition of phonological awareness skills, the ability to count the phonemes in a word develops around first grade for normal readers, but the ability to manipulate these phonemes is developing up to secondary school level (Adams 1990). A typical progression would be, first, syllable recognition at around three or four years; then an intermediate stage based on recognition of onsets and rimes; and finally recognition of individual phonemes after the age of 6 (Goswami & Bryant 1990). It is no coincidence that these skills develop at this time, in that early phonological awareness skills provide the foundations for the acquisition of higher levels of metaphonological skill. Data from illiterate adults (e.g., Morais et al. 1986; Bertelson et al. 1989) suggests that these higher level skills are to some extent acquired through learning to read, and themselves form the foundation of spelling skills.

Within the general category of phonemic awareness, there are considerable differences in the level of ability each task demands. The sound categorization tasks, particularly rhyming and alliteration, are amongst the simpler phonemic awareness tasks (Bradley & Bryant 1978; Stanovich et al. 1984), because the children do not require much knowledge of how to segment phonemes, and the task is simply dependent on the ability to compare and contrast words in terms of similarities and differences in their onsets and rimes. The easiest phoneme deletion task (Stanovich et al. 1984) is deletion of the initial phoneme (for example, say ca t without the/c/), with the most complex being phoneme substitution (for instance, replacing the/c/ of ca t

with/s/ to make sat) . The slow development of phonemic awareness has been linked to the memory organisation, perception and lexical representation

P H O N O L O G I C A L SKILL AND D Y S L E X I A 363

necessary to support phonemic segmentation, which is developing for normally achieving children up to the age of seven or eight (Fowler 1991). From this age onwards, Fowler (1991) notes that highly familiar items should have become fully specified in the lexicon and difficulty for poor readers will therefore be more evident with novel or nonsense words.

There is still considerable controversy over the factors responsible for success or failure in phonological awareness tasks of all types (for a review, see Yopp 1988). It has been argued that phonological awareness is either a precursor (Bradley & Bryant 1983), a co-requisite (Perfetti et al. 1987) or a by-product (Morais et al. 1979) of successful reading. Most recently, it has been suggested that phonological analysis may reflect stable individual differences linked to intelligence, as opposed to phonological synthesis (blending) which measures reading-related knowledge (Wagner et al. 1993). Analysis tasks include phoneme segmentation (break cat down into/c//a//t /) , phoneme deletion (say cat without the/c/), sound isolation (find the middle letter in cat), and sound categorization (the oddity task). Synthesis tasks all involve blending - onset and rime (c and at makes cat), phonemes into words (/c//a/ / t /makes cat) and phonemes into nonwords (/g/ /a/ /c/ makes gac). Regardless of the interpretation of phonological awareness skills, the ability to tap out or count the number of syllables correlates well with progress for beginner readers (Blachman 1984; Morals et al. 1986; Treiman & Baron 1983); it predicts future reading (Blachman 1984; Lundberg, Olofsson & Wall 1980); and it differentiates children with dyslexia from normal first graders (Mann 1984, 1986; Morais et al. 1986).

In summary, phonological awareness skills play a critical and reciprocal role at all levels of reading acquisition - alliteration and rhyming ability are implicated in early reading, and phonemic segmentation and deletion are required for becoming a fluent reader. Furthermore, training in segmentation and spelling in first grade has been extremely successful in improving reading of both real and nonsense words (Uhry & Shepherd 1993). It is likely that reading and spelling skill becomes increasingly more important in phoneme deletion and other more complex phonological awareness tasks (Mann 1986).

The ease or difficulty with which phonological awareness is induced by pre-reading experience may well be the critical variable underlying correla- tions between early phonological awareness and progress in reading (Wimmer et al. 1991). Perhaps the hallmark of the dyslexic child is the difficulty in attaining phoneme awareness under normal instructional conditions. As noted earlier, there is clear evidence for a phonological awareness deficit in younger children with dyslexia in the tasks described above (e.g., Bradley & Bryant 1978; Rack I985; Bruck & Treiman 1990; Sanchez & Rueda 1991; Hurford 1990). Unfortunately, direct training of phonological awareness skills has been less successful in both children with dyslexia and disadvantaged children with poor pre-school language experience. An early study by Rosner (1974) attempted to train disadvantaged pre-readers in phoneme deletion skills, but whereas the children quickly learned to break words down into syllables, even

364 A. J. FAWCETT AND R. I. NICOLSON

a year's intensive training had no effect on phoneme deletion skills for children below 5 years. Adams (1990) concluded that, in view of the central impor- tance of phoneme deletion skills to reading, the difficulty in teaching them to children with poor skills is particularly disturbing. More recently, it has been found that indirect training of phonological awareness skills may also be unsuccessful, in that training in reading skills for children with dyslexia does not transfer to phonological awareness skills (Manis et al. 1993) even in older children. An important issue therefore, is the extent to which phonological awareness skills improve with age and reading experience for children with dyslexia.

There is also strong evidence for deficits in phonological awareness in adults with dyslexia. However, the majority of this evidence is for problems in complex tasks, which demand higher levels of metalinguistic skill. Olson (1985) suggested that children with dyslexia may differ from normal readers in terms of the level of segmentation that their phonological awareness skills support, so that for instance they might cope with segmentation at the syllable but not the phoneme level. Olson and his colleagues (1989, 1991) showed that the phonemic substitution skills of children with dyslexia (mean age 15 years) were worse than 10 year old controls matched for reading age. The tasks they presented were complex, as appropriate for children of this age range, and included phoneme deletion in nonsense words and 'pig Latin' (in which the first letter of the word is moved to the end, and the suffix 'ay' added, turning the word 'pig' into "igpay', for example). These results were replicated using the same tasks with adult dyslexics in the Colorado family study (Pennington et al. 1990). There is evidence of deficits in adolescents and adults in similar complex tasks (e.g., Pratt & Brady 1988; Liberman et al. 1985, see also Fowler & Scarborough 1993) and also clinical evidence of deficit from case studies (e.g., Temple 1988).

The pig Latin task highlights an important issue in skilled performance, namely that phonological awareness alone is not sufficient to support skilled phonological performance. The pig Latin task requires several stages - stripping off and storing the onset (p) and the rime (ig), combining the onset with ay (pay), then blending the rime with the augmented offset igpay; loading the corresponding speech sounds into the articulatory buffer, and then artic- ulating them. It is a complex task, dependent not only on phonological awareness, but also processing speed (each subtask must be accomplished in the context of a rapidly fading memory trace), working memory capacity and/or general processing efficiency. A child may, with conscious effort, be able to decide that the onset is 'p' but the effort required would interfere with the efficient storage of that information. It is therefore important that the phonological skill be fluent and without effort, that is, automatic (Shiffrin & Schneider 1977).

It is now clearly established that the children with dyslexia suffer difficulties in processing speed (Denckla & Rudel 1976; Nicolson & Fawcett 1994), in working memory (Jorm 1983; Gathercole & Baddeley 1990;

PHONOLOGICAL SKILL AND DYSLEXIA 365

Nicolson, Fawcett & Baddeley 1992), and in automatization of skill (Nicolson & Fawcett t990). It is therefore not clear whether the pig Latin deficit does imply a phono-logical awareness skill deficit over and above a working memory deficit. Similarly, the problem in the deletion of phonemes in nonsense words identified by both Olson and Pennington (see also de Gelder & Vroomen 199t; Liberman et al. t985) is also difficult to interpret, in view of the established deficits for children with dyslexia even in the repetition of nonsense words (e.g., Snowling et al. 1986). Any of these difficulties could lead to problems in complex phonological skills, irrespective of phonological awareness competence.

This analysis led to the (2) issues motivating the research reported here: (1) to what extent is the deficit in phonological awareness skill specific to tasks which also involve significant memory and/or processing load; and (2) what is the comparative progression of phonological awareness skill acquisition, in particular do the early phonological awareness skill deficits of children with dyslexia disappear with extra years of reading training and the associ- ated improvement in reading age? If, say, by age 17 years children with dyslexia showed entirely normal phonological awareness skill in simple tasks, coexistent with deficits on complex tasks, one might reasonably conclude that there was only a lag in their acquisition of phonological awareness skills (albeit very severe in the early years), whereas a continuing deficit compared with reading age controls would further support the case for a specific disorder in the acquisition of phonological awareness skill (Bryant & Goswami 1986).

METHOD

Participants

In order to monitor change with years of remediation, three age groups of children with dyslexia (mean age 8, 13 and 17 years) were studied together with three groups of normally achieving children matched for age and IQ. We shall refer to the dyslexic and control groups as D8, Dt3, D17 and C8, C13, Ct7 respectively, with the suffix denoting the mean age. There were around a dozen children in each group (see Appendix 1 for full details). This design allows not only the standard chronological age match comparison, but also (by comparing the children with dyslexia with the younger controls) a reading age match comparison, and even (by comparing the oldest children with dyslexia with the youngest controls) a 'twice chronological age' match comparison. The reading age match comparison is important theoretically, since a deficit compared with chronological age controls may indicate only delay, whereas a deficit compared with reading age controls provides evidence of a disorder in that skill (Bryant & Goswami 1986).

All the children with dyslexia had been diagnosed as dyslexic between the ages of 7 and 10, based on discrepancies of at least 18 months between chronological and reading age, together with full scale IQ of at least 90 on the WISC-

366 A. J. F A W C E T T A N D R. I. N I C O L S O N

R (Wechsler 1976). They were recruited from the Dyslexia Institute and the British Dyslexia Association, and included all those children referred to us who satisfied the standard exclusionary criteria (that is, with no evidence of neurological damage, primary emotional difficulties or socioeconomic deprivation), and who were prepared to participate in our longitudinal study on a regular basis. They were paid the sterling equivalent of around $5 per hour. The normally achieving children were selected from local schools (with reading age no more than six months below their chronological age and no history of reading problems) in order to achieve a reasonably good IQ match for the children with dyslexia. The majority of the children were male, and drawn predominantly from middle class or skilled working class backgrounds. Psychometric details of the subjects are given in Appendix 1.

Procedure

Our intention was to minimize aspects of these tasks which make heavy demands on speed of processing or working memory capacity, because such factors confound interpretation of any deficit. In order to reduce scope for strategic variation we used tests of simple phonological awareness skill, ranging in difficulty fiom sound categorization (detection of rhyme and alliteration) to the more complex phoneme deletion. Two tasks were admin- istered, a sound categorization task, based on the procedure used by Bradley and Bryant (1983), and a phonemic deletion task, based on the procedure used by Rosner (1971). All testing was completed within a single one-hour session, with breaks between each subtask. Testing took place in an experimental room in the Psychology department (for the children with dyslexia and the older controls), or in a quiet room in school (for the younger controls). Each child was tested individually.

Sound categorization. In pilot work using the three-stimulus 'Odd man out' format devised by Bradley & Bryant (1983), we established that several of our oldest dyslexic group had difficulty in remembering all three stimuli long enough to make the necessary comparison. Typically, they would repeat the first two stimuli, and then question the experimenter as to the third stimulus. Since we wished to dissociate memory function from pure phonological awareness skills, we reduced the memory load by presenting only two stimuli at each trial, and asking whether or not they had the same beginning, middle or end, depending on the condition.

This experiment was carried out on an Apple Macintosh computer, using Apple's Hypercard multimedia environment. The stimuli used were the 90 single syllable three or four letter words derived by Bradley and Bryant (1983) (see Appendix 2). Each stimulus was spoken by the first author, digitised at a 22KHz sampling rate, and stored as a HyperCard sound resource which could then be played under experimental control. In pilot work we confirmed that the stimuli were clearly discriminable.


Three different experimental conditions were presented, in a fixed order, with ten trials in each condition, following 2 practices. For each trial, a word triple was selected (without replacement) pseudorandomly from the pool of 30 word triples, and the first word was chosen, together with either the second or the third word, so that the two stimuli shared a phoneme on exactly 50% of the trials for each condition. The three conditions, based on Bryant and Bradley (1983) were as follows:

Condition 1: Rhyming. This condition was introduced by asking the child to recite a nursery rhyme, and emphasising the ends of lines that rhymed to ensure that the children were familiar with the concept. The child was then asked to generate a word first that rhymed with 'cat'. This was followed by a computer-presented practice with feedback, and then the experimental condition. The experimental instructions were presented by the computer in synthesized speech and were 'I will say two words. Sometimes they rhyme. Listen carefully and tell me whether they rhyme or not ' . Synthesized speech was used to indicate that the children were not expected to read the instructions, and also because in previous research we found that the children responded well to the robot-like voice which added an element of fun. In order to check that the participants had followed the instructions, these were repeated by the experimenter, and subjects were instructed to say 'yes ' if the word ended with the same sound, or 'no ' if it did not.

Condition 2: The same sound in the middle. For this condition the child was again asked first to generate a word, this time with the same sound in the middle as cat. The instructions were 'I will say two words. Sometimes they have the same sound in the middle. Listen carefully and tell me whether they have the same sound in the middle or not ' . The instructions were repeated by the experimenter, and subjects were instructed to say 'yes ' if the word had the same sound in the middle, or 'no ' if it did not. I

Condition 3: Alliteration. The alliteration condition was introduced with the idea of ' I -Spy' , and the child asked to generate words starting with the selected letter. The instructions were 'I will say two words. Sometimes they start with the same sound. Listen carefully and tell me whether they start with the same sound or not. ' The instructions were repeated by the experimenter, and subjects were instructed to say 'yes ' if the word started with the same sound or 'no ' if it did not.

This experiment had no speed component, and the experimenter recorded the child's decision for each trial. The experimenter ensured that the partici- pant repeated the two words to check that the words had been correctly heard and remembered. Results were analysed automatically, giving overall accuracy in each condition, with a maximum score of 10 in each condition, and 30 overall.

Test of Auditory Analysis Skills (Rosner & Simon 1971). The TAAS test is a spoken test of the ability to segment words into syllables and delete phonemes (see Appendix 3). It starts with two simple practice items, in the

368 A. J. FAWCETT AND R, I, NICOLSON

first of which the child is asked to analyse a two syllable compound word into syllables (say 'cowboy' and then say it again without the 'boy'). The experimenter records the response. A series of 13 items of increasing complexity is then presented, suitable for use from kindergarten onwards. At the phoneme level, the position of the sound is controlled for difficulty, starting with the easiest, substituting the first phoneme, then the final sound, and finally part of a consonant blend. The discontinuation rules suggest that presentation should stop after two consecutive failures. However, we were particularly interested in variability in performance across the levels, and as the complete test took only a few minutes to administer, it was presented in its entirety to each child.

RESULTS

Sound categorization

Mean results and standard deviations are shown in Table 1. All three control groups performed at or near 100%, with the majority of the C8 group obtaining maximum scores. By contrast, the children with dyslexia made errors in all three conditions. The alliteration condition was the most difficult task for all the children. The data were analysed separately for the three conditions, but the pattern of results was the same.

In view of the posit ive skew of the data, the scores were converted to probabilities and the modified arc sine transformation 2 (Winer 1971: 400) was applied. Significant differences between the children with dyslexia and the

Table 1. Mean performance (percent correct) for the four phonological tasks*

Dyslexia Control

Group D8 D13 DI7 C8 C13 C17

Rhyme 97,3 100 95.0 100 100 100 (4.67) (0.00) 7,98) (0,00) (0,00) (0.00)

Middle sound 92.7 92.2 95.8 98.0 I00 100 (7.86) (8.33) (9.00) (4.22) (0.00) (0.00)

Alliteration 82.7 93,3 91.7 96.0 94.2 100 (11.9) (7.07) (14.7) (9.66) (7,93) (0.00)

Overall sound 90.9 95.2 94.2 98,0 98,1 100 categorization (6,34) (3.77) (6,98) (3,58) (2.64) (0.00)

TAAS test 65.4 73,5 76,9 83,0 96,8 100 (27,2) (13.4) (15,7) (9.21) (5,14) (0.00)

* Standard deviations in parentheses.


controls were found for each condition, with the pattern of results being equivalent across conditions. Consequently, analyses here are presented taking the overall score collapsed across the three conditions.

Comparison with chronological age controls. A two factor analysis of variance was undertaken for the six groups, with the factors being age (three levels) and dyslexia (presence/absence). The main effect of dyslexia was highly significant, whereas that of age was not [F(1,59) = 24.7, p < 0.0001; F(2,59) = 2.1, NS respectively]. The interaction between age and dyslexia did not approach significance.

Comparison with reading age controls. A further two factor analysis of variance was conducted, omitting the C17 and D8 groups and comparing the two older dyslexic groups with their reading age controls. The factors were therefore reading age (two levels) and dyslexia (presence/absence). The main effect of dyslexia was significant, but that of reading age was not [F(1,37) -- 4.4, p < 0.05; F(1,36) = 0.1, NS respectively]. The interaction between reading age and dyslexia did not approach significance.

Overall, therefore, there was no significant effect of age, but the dyslexic groups performed significantly worse than both their chronological and reading age controls.

TAAS test

The mean data and standard deviations are also shown in Table 1. It may be seen that the two older control groups were close to 100%, but all four of the other groups made substantial numbers of errors. The performance of the youngest controls is nevertheless better than that of the oldest children with dyslexia. The data were again converted to proportions and the modified arc sine transformation applied.

Comparison with chronological age controls. A two factor analysis of variance was undertaken for the six groups, with the factors being age (three levels) and dyslexia (presence/absence). The main effects of both age and dyslexia were significant [F(2,59) = 7.3, p < 0.01; F(1,59) = 38.6, p < 0.0001 respectively]. The interaction between age and dyslexia did not approach significance.

Comparison with reading age controls. A further two factor analysis of variance was conducted, with the factors being reading age (two levels) and dyslexia (omitting the C 17 and D8 groups, as above). The main effects of both reading age and dyslexia were significant [F(1,37) = 10.6, p < 0.01; F(1,37) = 16.0, p < 0.001 respectively]. The interaction between reading age and dyslexia did not approach significance.

Overall, therefore, for the TAAS phoneme deletion test, the dyslexic groups again performed significantly worse than both their chronological and reading age controls, but in this case there was also a significant overall improvement in performance with age.


DISCUSSION

The children with dyslexia performed significantly worse, even than their reading age controls on all the tasks investigated. It may be seen from Table 1 that, in every test, performance of the oldest children with dyslexia was inferior even to that of the youngest controls. This pattern of results suggests that children with dyslexia have persistent, and unexpectedly severe, problems in phonological awareness skill, even with short highly familiar lexical items.

The ceiling effects on the sound categorization task for the normally achieving children reduce the sensitivity of the analyses, and so the significant differences obtained are especially interesting. It is hard to conceive of a much simpler task than the rhyming condition, and the alliteration condition also has minimal memory load. The major thrust of this research therefore is to show that even under the most advantageous method of presentation possible, using early age of acquisition one syllable words, where the stimuli were clearly discriminable and easily remembered, the children with dyslexia still had problems in sound categorization. However, although the sound categorization task was designed explicitly to minimize memory processing load, it might still be argued that the sound categorization deficits reflect some inappropriate response bias strategy. Consequently, we undertook an analysis of the errors made.

The overall incidence of errors (collapsed across the three control groups) was 0, 2 and 7 (out of 33 participants) for conditions 1, 2 and 3 respectively. For the three dyslexic groups (34 participants) the corresponding incidence levels were 7, 14 and 18 respectively. Although almost all the errors made by the normally achieving children were in the alliteration condition, the errors of children with dyslexia were spread across all three conditions. The failure of four of the D17 group to identify correctly whether two three-letter words rhymed is particularly remarkable, with over half errors of commission, that is rejecting rhymes. In the three dyslexic groups overall, 78% of the errors were incorrect rejection of rhymes, with 57% of those errors made by the D17 group. In the alliteration condition, 69% of the errors in the dyslexic groups were based on the rejection of a correct match, with 25% of these errors attributable to the D17 group. It appears therefore that any response bias interpretation is untenable.

It is also interesting to consider the type of error made on the TAAS test, whether in simple syllable deletion, omitting the first or last phoneme, or the more complex task of deleting part of a consonant cluster. Surprisingly enough, 5 (39%) of the D17 group and 5 (50%) of the D13 group failed to simply delete the cu from cucumber (item 3 in difficulty in the list of t3), whereas none of the C17 group and only two of the Ct3 group had this problem. The two youngest groups, the D8 and C8 group made an equal number of errors, 4 each, on cucumber. In fact, the most common error for dyslexic children at all ages was the failure to break down a consonant cluster, with consistent errors in clap, play, and stale, whereby both phonemes of the


consonant cluster were deleted instead of one. This meant that in response to 'Say clap, say it again, but don't say/k/)' both phonemes in the cluster were deleted to produce 'ap'. These errors occurred in 10 D17, 6 D13 and 5 D8 children, who generated a total of 14, 11 and 12 errors of this type respectively. Only the youngest controls (C8) made this type of error, and even they were considerably more accurate than the D17 group, with 6 C8 children producing 8 errors in total. Interestingly enough, some of the dyslexic children also deleted the last phoneme when they meant to delete the first and vice- versa (for example deleting the s rather than the m from same, producing ame for say), reversed the order of the remaining phonemes, or mispronounced the string (for example ple for plea). Only three children with dyslexia (one in each age group), scored full marks on this test. It is useful to bear in mind here that the Rosner (1979) guidelines suggest that children in grade 3 (around age 9) should achieve perfect performance on the TAAS test. In normative terms (Rosner 1979), the performance of the 8 year old and 13 year old children with dyslexia was at the level of the average 6 year old, whereas the 17 year old children with dyslexia were at around the 7-8 year old 1eve1. Our results suggest, that even by the age of 17, some children with dyslexia have still not fully mastered breaking words into syllables, and the majority continue to have significant problems in breaking apart clusters.

The primary motivation for this study was investigation of the issue of whether, for children with dyslexia, phonological awareness deficits persist into adolescence, or whether the existing evidence of persisting deficits in complex processing tasks such as Pig Latin might be partly accounted for in terms of impaired working memory or processing speed. It is clear from our results that deficits do persist in simple phonological awareness skills, in that all three groups of children with dyslexia performed significantly worse than their reading age controls, indicative of a disorder in phonological awareness skills. Moreover, assuming that the reading age match equates orthographic skill, the significant deficit compared with the reading age match suggests that orthographic knowledge alone is not sufficient for successful completion of the phoneme awareness tasks by the dyslexic groups. This deficit was not eliminated by the acquisition of reading skills, in that, although all of the children with dyslexia showed evidence of some skill in phonological awareness, the mean performance overall of the 17 year old children with dyslexia was in no case better than that of the 8 year old controls. Furthermore, these deficits occurred not only on the relatively complex phoneme deletion task, but also on the extremely simple sound categorization tasks which were constructed to cause negligible memory load. None of the 34 children with dyslexia performed perfectly on all tasks, whereas over half the controls did. These findings provide strong evidence that children with dyslexia suffer from a disorder in phonological awareness skills which persists at least into late adolescence.


ACKNOWLEDGEMENTS

The research reported here was supported by a grant from the Leverhu lme

Trust to the Univers i ty of Sheffield. We thank Alan Baddeley and Tim Miles

for va luable suggest ions for the design of the studies. We gratefully acknowl- edge the dedicated support of the participants and their parents, and the support

of the teachers and pupils at Ecclesal l Junior School and Silverdale Secondary

School, Sheffield.

APPENDIX 1: PSYCHOMETRIC DATA FOR THE SIX GROUPS OF PARTICIPANTS

Group N IQ (WISC-R) Chronological age Reading age

C17 11 107.3 [92 to 130] 17.4 [17:0 to 18:0] 15.&

D17 13 105.0 [88 to 126] 17.4 [16:0 to 18:2] 12.5 [8.2 to t4.6] (11) b {105.0 [91 to 126]} {13.2 [10.9 to 14.6]}

C13 12 111.6 [96 to 129] 13.4 [13:0 to 14:4] 13.7 [12.8 to 14.6]

D13 10 11t.2 [101 to 128] t3.3 [11:7 to 14:9] 9.9 [7.9 to 12.3]

C8 10 115.1 [101 to 133] 8.2 [7:0 to 8:8] 9.2 [7.7 to 11.1]

D8 11 113.4 [96 to 133] 8.7 [7:7 to 9:9] 6.6 [5.5 to 7.9]

Mean data are presented, with ranges in brackets. 15 represents ceiling on the Schonell test of reading age used. All this group were reading

at this level, and the majority had reached this level at around the age of 15. b In order to improve the match for reading age with the 13 year old controls, two D17 subjects were omitted from the reading age analysis. Figures in parentheses show the psychometric data excluding these subjects.

APPENDIX 2: SOUND CATEGORIZATION STIMULI FOR RHYMING

(From Bradley & Bryant 1978, 1983)

Condition 1: Condition 2: Condition 3: End of word rhyme Middle of word rhyme Beginning of word rhyme

pin, win, sit pot, cot, hut pin, pig, hill hop, top, doll bun, gun, pin bus, bun, rug bun, gun, hut pig, wig, hug hat, ham, tap map, cap, pal bed, head, lid cup, cut, fun head, bed, men bag, rag, leg net, neck, bed wig, fig, pin bell, well, doll dog, doll, pot rod, nod, sock tin, pin, man jam, jack, lad mud, bud, sun dog, log, mug kick, kiss, fill nut, cut, fun cup, pup, lip wet, web, bell well, bell, pet cat, bat, fit log, lock, mop

P H O N O L O G I C A L S K I L L AND D Y S L E X I A 373

APPENDIX 3: THE ROSNER (1971) TEST OF AUDITORY ANALYSIS SKILLS (TAAS)

Say sunshine Say it again, but don' t say shine Say picnic Say it again, but don' t say pic Say cucumber Say it again, but don' t say cu Say coat Say it again, but don' t say/k/ Say meat Say it again, but don' t say/m/ Say take Say it again, but don' t say/ t / Say same Say it again, but don' t say/m/* Say wrote Say it again, but don' t say/ t / Say please Say it again, but don' t say/z/ Say clap Say it again, but don' t say/k/ Say play Say it again, but don't say/p/ Say stale Say it again, but don't say/ t / Say smack Say it again, but don' t say/m/

* We substituted the word 'same' for the original "game', because in pilot studies we established that even the 8 year old children were reluctant to produce the answer "gay'.

NOTES

1. It should be noted, however, that in all cases where the middle phoneme in these stimuli is the same, the words also rhyme. Performance on the task was equivalent to that on the rhyming condition. The stimuli were used for compatibility with Bradley & Bryant (1983).

2. For a proportion p lying between 0 and 1, the transformation is p --~ arcsin(p) unless p = 1 when p ~ 2 arcsin (~/[p - 1/2n]), where n is the number of observations on which the proportion p is based. A log linear transformation may also be used, but Winer suggests use of the arc sine for values of X either close to 0 or close to unity.

REFERENCES

Ackerman, P. T., Dykman, R. A. & Gardner, M. Y. (1990). ADD students with and without dyslexia differ in sensitivity to rhyme and alliteration, Journal of Learning Disabilities 23: 279-283.

Adams, M. J. (t990). Beginning to read: Thinking and learning about print. Cambridge, MA: MIT Press.

Ball, E. W. & Blachman, B. A. (1991). Does phoneme segmentation training in kindergarten make a difference in early word recognition and developmental spelling, Reading Research Quarterly 26: 49-66.

Ball, E. W. (1993). Phonological awareness: What 's important and to whom?, Reading and Writing: An Interdisciplinary Journal 5: 141-159.

Bertelson, P., de Gelder, B., Tfouni, L. V. & Morais, J. (1989). Metaphonological abilities of adult illiterates: New evidence of heterogeneity, European Journal of Cognitive Psychology 1 : 239-250.

Blachman, B. A. (1984), Relationship of rapid naming ability and language analysis skills to kindergarten and first-grade reading achievement, Journal of Educational Psychology 76: 610-622.

Bradley, L. & Bryant, P. E. (1978). Difficulties in auditory organisation as a possible cause of reading backwardness, Nature 271: 746-747.

374 A. J. F A W C E T T AND R. I. N I C O L S O N

Bradley, L. & Bryant, P. E. (1983). Categorising sounds and learning to read: A causal connection, Nature 301: 419-421.

Bruck, M. & Treiman, R. (1990). Phonological awareness and spelling in normal children and dyslexics: The case of initial consonant clusters, Journal of Experimental Child Psychology 50: 156-178.

Bryant, P. E. & Goswami, U. (1986). Strengths and weaknesses of the reading level design, Psychological Bulletin 100: 101-103.

Bryant, P.E., MacLean, M., Bradley, L. & Crossland, J. (1990). Rhyme and alliteration, phoneme detection and learning to read, Developmental Psychology 26: 429-438.

Byrne, B., Freebody, P. & Gates, A. (1992). Longitudinal data on the relations of word-reading strategies tO comprehension, reading time and phonemic awareness, Reading Research Quarterly 27: t41-151.

de Gelder, B. & Vroomen, J. (1991). Phonological deficits: Beneath the surface of reading- acquisition problems, Psychological Research 53: 88-97.

Denckla, M. B. & Rudel, R. G. (1976). Rapid 'Automatized' naming (RAN): Dyslexia differ- entiated from other learning disabilities, Neuropsyehologia 14:471-479.

Felton, R. H. (1994). Effects of instruction on decoding skills in children with phonological processing problems, Journal of Learning Disabilities 26: 583-590,

Fowler, A. E. (1991). How early phonological development might set the scene for phoneme awareness. In S. A. Brady & D. P. Shankweiler (eds.), Phonological processes in literacy (pp. 97-117). Hillsdale, N J: Erlbaum.

Fox, B. & Routh, D. K. (1984). Phonemic analysis and synthesis as word attack skills: Revisited, Journal of Educational Psychology 76:1059-1064.

Gathercole, S. E. & Baddeley, A. D. (1990). Phonological memory deficits in language disordered children: ts there a causal connection?, Journal of Memory and Language 29: 336-360.

Goswami, U. & Bryant, P. E. (1990). Phonological skills and learning to read. Hitlsdale, NJ: Erlbaum.

Hatcher, P., Hulme, C. & Ellis, A. W. (t994). Ameliorating early reading failure by integrating the teaching of reading and phonological skills, Child Development (in press).

Hurford, D. P. (1990). Assessment and remediation of a phonemic discriminatien deficit in reading disabled second and fourth graders, Journal of Experimental Child Psychology 50: 396-415.

Jorm, A. F. (1983), Specific reading retardation and working memory: A review, British Journal of Psychology 74:311-342.

Liberman, I. Y., Rubin, H., Duques, S. & Carlisle, J. (1985). Linguistic abilities and spelling proficiency in kindergarteners and adult poor spellers. In D. B. Gray & J. F. Kavanagh (eds.), Biobehavioral Measures of Dyslexia (pp. 163-176). Baltimore, MD: York Press.

Lundberg, I., Olofsson, A. & Wall, S. (1980). Reading and spelling skills in the first school years predicted from phonemic awareness skills in kindergarten, Scandinavian Journal of Psychology 21: 159-173.

MacLean, M., Bryant, P. E. & Bradley, L. (1987). Rhymes, nursery rhymes and reading in early childhood, Merrill-Palmer Quarterly 33: 255-282.

Manis, F. R., Custodio, R. & Szeszulski, P. A. (1993). Development of phonological and orthographic skill: A two years longitudinal study of dyslexic children, Journal of Experimental Child Psychology 56: 64-86.

Mann, V. A. (1984). Reading skill and language skill, Development Review 4: t-15. Mann, V. A. (1986). Phonological awareness: The role of reading experience, Cognition 24:

65-92. Morais, J., Cary, L., Alegria, J. & Bertelson, P. (1979). Does awareness of speech as a sequence

of phones arise spontaneously?, Cognition 7:323-331. Morais, J., Bertelson, P., Cary, L. & Alegria, J. (1986). Literacy training and speech segmen-

tation, Cognition 24: 45-50.

P H O N O L O G I C A L SKILL AND D Y S L E X I A 375

Nicolson, R. I. & Fawcett, A, J. (1990). Automaticity: A new framework for dyslexia research, Cognition 35: 159-182.

Nicotson, R. I. & Fawcett, A. J. (1994). Reaction times and dyslexia, Quarterly Journal of Experimental Psychology 47: 29-48.

Olson, R. K., Davidson, B. J., Kliegel, R. & Fottz, G. (t985). Individual and developmental differences in reading disability. In G.E. MacKinnon & T.G. Waller (eds.), Reading research: Advances in theory and practice (Vol. 4, pp. 1-64). New York: Academic Press.

Olson, R. K., Wise, B. W. & Rack, J. P. (1989). Dyslexia: Deficits, genetic etiology and computer based remediation, lrish Journal of Psychology 10: 494-508.

Olson, R. K., Giltis, J. J., Rack, J. P., DeFries, J. C. & Fulker, D. W. (1991). Confirmatory factor analysis of word recognition and process measures in the Colorado reading project, Reading and Writing: An Interdisciplinary Journal 3: 235-248.

Pennington, B.F., Van Orden, G.C., Smith, S,D. & Green P.A. (1990). Phonological processing skills and deficits in adult dyslexics, Child Development 61: 1753-1778.

Perfetti, C. A., Beck, I., Bell, L. C. & Hughes, C. (1987). Phonemic knowledge and learning to read are reciprocal, Merrill-Palmer Quarterly 33: 283-289.

Pratt, A. C. & Brady, S. (1988), Relation of phonological awareness to reading disability in children and adults, Journal of Educational Psychology 24: 28-37.

Rack, J. (1985). Orthographic and phonetic coding in normal and dyslexic readers, British Journal of Psychology 76: 325-340.

Rosner, J. & Simon, D. P. (1971). The auditory analysis test: An initial report, Journal of Learning Disabilities 4: 384-392.

Rosner, J. (1974). Auditory analysis training with pre-readers, The Reading Teacher 27: 379- 384.

Rosner, J. (1979). Helping children overcome learning disabilities, 2nd ed. New York: Walker. Sanchez, E. & Rueda, M. I. (1991). Segmental awareness and dyslexia: Is it possible to learn

to segment well and yet continue to read and write poorly?, Reading and Writing: An Interdisciplinary Journal 3:11-18.

Shiffrin, R. M. & Schenider, W. (1977). Controlled and automatic human information processing II: Perceptual learning, automatic attending and general theory, Psychological Review 84: 127-190.

Snowling, M., Goulandris, N., Bowlby, M. & Howell, P. (1986). Segmentation and speech perception in relation to reading skill: A developmental analysis, Journal of Experimental Child Psychology 4i: 489-507.

Stanovich, K. E. (1988). The right and wrong places to look for the cognitive locus of reading disability, Annals of Dyslexia 38:154-177.

Stanovich, K. E., Cunningham, A. E. & Cramer, B. B. (1984). Assessing phonological skills in kindergarten children: Issues of task comparability, Journal of Experimental Child Psychology 38: t75-190.

Tangel, D. M. & Blachman, B. A. (I992). Effect of phoneme awareness instruction on kindergarten children's invented spelling, Journal of Reading Behaviour 24: 233-261.

Temple, C. M. (1988). Developmental dyslexia and dysgraphia persistence in middle age, Journal of Communication Disorders 21:189-203.

Treiman, R. & Baron, J. (1983). Phonemic analysis training helps children benefit from spelling- sound rules, Memory and Cognition 11: 382-389.

Uhry, J. K. & Shephed, M. J. (1993). Segmentation/spelling instruction as part of a first grade reading program: Effects on several measures of reading, Reading Research Quarterly 28: 219-233.

Vellutino, F. R. (1979). Dyslexia: Theory and Research. Cambridge, MA: MIT Press. Wagner, R. K., Torgeson, J. K., Laughon, P., Simmons, K. & Rashotte, C. A. (1993).

Development of young readers' phonological processing abilities, Journal of Education Psychology 85: 83-103.

Wimmer, H., Landerl, K., Linortner, R. & Hummer, P. (1991). The relationship of phonemic

376 A. J. F A W C E T T AND R. I. N I C O L S O N

awareness to reading acquisition: More consequence than precondition, but still important, Cognition 40: 219-249.

Winer, B. J. (1971). Statistical principles in experimental design, 2nd ed. New York: McGraw- Hill.

Wechsler, D. (1976), Wechsler Intelligence Scale for Children Revised (WISCR). Slough, UK: NFER.

Yopp, H. K. (1988). The reliability and validity of phonemic awareness tests, Reading Research Quarterly 23: 159-177.

Address for correspondence: Dr Angela J. Fawcett, Department of Psychology, University of Sheffield, Western Bank, Sheffield S 10 2UR, UK Phone: +44 742 826-546; Fax: +44 742 766-515

Persistence of phonological awareness deficits in older children with dyslexia

Documents