The foregoing discussion has highlighted the importance of "prediction" studies that tell us a great deal about the knowledge and skills young children bring to bear on the task of learning to read. Research-generated phonological tasks might, if given to sufficiently large samples of children, provide norms for the purposes of screening young children, or against which individual "slow starter" youngsters might be compared. However, whether such studies can suggest a strategy for reliably identifying those specific children who go on to have persisting phonologi-cally based reading problems that necessitate special needs intervention is a rather more complex issue. When considering individual children, it is not always possible to conclude confidently that a child who obtains a low score on a measure of phonological awareness will necessarily go on to have significant reading difficulties. In Bradley and Bryant's (1983) longitudinal study of early readers, only 30% of those children who initially produced good sound categorization scores went on to become exceptionally good readers. Of greater relevance to the early identification issue is the finding that just 28% of those who initially produced poor sound categorization scores became exceptionally poor readers. These authors suggest that a phonological awareness test on its own might not be a particularly effective way of predicting persisting reading problems. More recent studies that have used a number of independent predictors have reported higher sensitivity ratings of 48% (Schneider & Naslund, 1993), 42% (Badian, 1988), and 56% (Butler, Marsh,
Sheppard, & Sheppard, 1985). An even higher sensitivity was achieved in a study by Badian (1994) in which she was able to predict the problems of 14 out of 15 poor readers. However, a high sensitivity such as this may be at the cost of a relatively high number of false alarms, depending on where the cutoff point for group inclusion is placed. So, in the Badian' (1994) study, 10 out of 24 children did not develop reading difficulties as predicted. At the extreme, adopting a large number of predictors in order to achieve higher prediction rates also has its disadvantages. Using a large number of independent measures will provide an almost perfect, or even a perfect, prediction (Elbro et al., 1998). Beyond the practical limitations of time and cost, this is not in itself desirable theoretically or methodologically. Prediction is best when each measure is strongly correlated with the outcome measure of reading, but uncorrelated with the other measures (Tabachnick & Fidell, 1989). Thus, the goal of any screening or early assessment measure is to select the fewest independent measures necessary to provide a good prediction of reading outcome where each measure predicts a substantial and independent proportion of the variability in reading outcome.
There are a number of tests that have employed a phonological awareness measure as a screening or early diagnostic instrument. In the United States, Diane Sawyer's (1987) Test of Awareness of Language Segments (TALS) can be used as a screen for children at the kindergarten or Grade 1 level, but is also recommended for use diagnostically and prescriptively for older children who are already exhibiting delay in learning to read. In this test, children are required to segment language, first from sentences to words, then words to syllables, and finally from words to sounds. Predictive validation studies have shown that the TALS administered at age 4 years 6 months to 5 years 6 months significantly predicted reading right through to Grade 3. Also in the United States, Torgesen and Bryant's (1994) Test of Phonological Awareness (TOP A) is a group administered test in which children use pictorial material to demonstrate their ability to identify initial sounds (kindergarten version) or end sounds (elementary version) within words. The TOPA-Kindergarten was normed on 857 children aged 5 and 6, while the TOPA-Early Elementary was normed on 3,654 children aged 6 through 8 years. Predictive validity from kindergarten TOPA scores showed that the correlation with reading analysis skill 1 year later was 0.55.
Bearing in mind Bradley and Bryant's (1983) caution against relying on a single measure to predict reading outcome reliably, the present author, together with Charles Hulme and Margaret Snowling developed a multi-measure screening and early diagnostic instrument aimed at being simple to use, economic, and able to provide good prediction, while at the same time being congruent with recent research findings and with current theoretical perspectives on early reading development. The Phonological Abilities Test (PAT) was published by the Psychological Corporation UK in 1997 (Muter, Hulme, & Snowling, 1997). Tests were selected that tap the skills young children bring to bear on their earliest reading experiences and which reflect the authors' theoretical position on the important relationship between phonological abilities and reading. Since phonological awareness skills are the most stable and robust of the available predictor skills, and in view of their ability to predict reading outcome over long periods of time, they were given selection priority over more transitory predictors like vocabulary or naming skill. The PAT contains two measures of children's ability to segment words into syllables or phonemes—a test of syllable and phoneme completion, and a test of phoneme deletion. While current evidence seems to favor segmentation over rhyming as being the better predictor of beginning literacy, measures of onset-rime awareness were not excluded. There is the well-documented view that rhyming skill may have a bearing on later stages of learning to read, and, in particular, on children's ability to adopt analogical strategies. We have seen that there is increasingly strong evidence emerging as to the paramount importance of letter knowledge acquisition in early reading development. Thus, a test of letter knowledge as an indicator of the thoroughness and possibly ease with which the letter identities have been learned was viewed as an obligatory component of an early screening battery. A measure of children's phonological memory and access, speech rate completes the battery.
The PAT comprises four phonological awareness subtests (Rhyme Detection; Rhyme Production; Word Completion—Syllables and Phonemes; Beginning and End Phoneme Deletion), a Speech Rate test (timed repeating of the word "buttercup"), and a test of Letter Knowledge. The test was standardized on 826 children aged 4-7 years, and norms provided in 6-month age bands between 5 and 7 years (and in a 12-month age band for the 4-year-olds). When given in full, the PAT takes approximately 25-30 min to administer, with each subtest varying in administration time from around 3 min (Letter Knowledge and Speech Rate) to up to about 8 min (Beginning and End Phoneme Deletion). The individual subtests demonstrated good internal and test-retest reliability. Criterion-related validity studies have shown that the individual subtests correlate significantly with a concurrently administered standardized reading test (British Abilities Scales Reading Test, Elliott et al., 1983). Multiple regression analyses have also demonstrated that the subtests are significant predictors of concurrent reading skill in the age range 5-7 years, with the best and most consistent predictor subtests being Phoneme Deletion and Letter Knowledge. The predictive validity of the PAT has been recently evaluated in a longitudinal study of almost 100 children. The present author administered the PAT to rising 5-year-olds who had had only very minimal exposure to formal reading instruction. The measures of Letter Knowledge, Phoneme Completion, and Beginning and End Phoneme Deletion together accounted for 55% of the variance in children's reading skills 1 year later. In effect, they predicted, at an accuracy rate of 90%, whether the children would be categorized as good or poor readers 1 year later. The PAT, like the TALS, may be used as a screening measure in the age range 5-7 years (either when given in its entirety or using a subset of tests), or it may be administered to selected older children who are already experiencing reading problems and for whom a diagnostic and prescriptive phonological profile is required.
For older children for whom the PAT may be used diagnostically, reference is made not only to the norms (given in centiles), but also to the graphically represented PAT Profile. The following single case study illustrates how the PAT might be used diagnostically and as a prescription for teaching. Andrew aged 7 years 11 months had a WISC III Verbal IQ of 101 (Wechsler, 1992), and could thus be regarded as being of average ability. He scored at barely the 6-year level on a standardized test of single word reading (Wechsler Objective Reading Dimensions, WORD, Wechsler, 1993), and he was unable to read any nonwords from the Graded Nonword Reading Test (Snowling et al., 1996). Andrew's scores on the Rhyme Detection, Rhyme Production, Beginning and End Phoneme Deletion, and Letter Knowledge subtests were all at or under the tenth centile. He had no difficulty with Sentence or Phoneme Completion (fiftieth centile) or with Speech Rate (seventy-fifth centile). His PAT profile is shown in Figure 1. While Andrew clearly has no problems in respect of phonological short-term memory, and his simple segmentation abilities are established, he is nonetheless experiencing great difficulty in many other aspects of phonological skill and in acquiring letter knowledge. His pattern of difficulty, taken together with his marked reading underachievement and his total lack of decoding ability, is clearly indicative of his having a specific literacy difficulty that has the hallmarks of a dyslexic problem. He needs to embark on a systematic literacy training program which emphasizes phonological awareness and related skills. Of the phonological abilities in which Andrew is deficient, rhyming is the ability which appears earliest in the developmental progression of phonological skills. It is, therefore, recommended that this be trained first. Later on, Andrew needs to work on his phonological manipulation skills through exercises which teach him to add, delete, substitute, or transpose phonemes within words. Andrew also needs to be trained in his letter knowledge, with a teaching approach that emphasizes multi-sensory learning (feeling, writing, naming) of both letter names and sounds and where there is the opportunity for a lot of practice and reinforcement. When Andrew is more phonologically aware, and when he knows all his letters, he should be exposed to "linkage" exercises which help him make important connections between his improving speech sound sensitivity and his experience of print. After Andrew has worked through a program such as this, he should be ready to embark on a structured phonic-based program that teaches him about grapheme-to-phoneme consistencies and about sequential decoding skill.
The PAT is an instrument to be used for assessing phonological skills in children in the early stages of learning to read. However, as we have already seen, phonological skills continue to have a significant long-term bearing on children's reading progress and their ultimate literacy outcome. A UK-based phonological test battery geared toward somewhat older children is the Phonological Assessment Battery (PhAB) (Frederickson, Frith, & Reason, 1997). The PhAB provides norms for children aged 6 years to 14 years 11 months and consists of six subtests: alliteration, naming speed (for pictures and digits), rhyme, spoonerisms, fluency, and nonword reading. The PhAB is intended not so much as a screening
instrument for early identification (the main purpose of the PAT), but as a diagnostic and prescriptive instrument "in particular... for children whose literacy progress is causing concern" (p. 3). Criterion-related validation studies have confirmed that the PhAB tests show significant correlations with a standardized reading measure, the Neale Analysis of Reading Ability (Neale, 1989). The clinical applicability of the PhAB was demonstrated by comparing the PhAB scores of children with significant reading underachievement with those in the standardization sample; the specific reading disabled group obtained significantly lower scores on all the PhAB subtests apart from the fluency measures.
A recently available, and highly recommended, phonological assessment instrument is the Comprehensive Test of Phonological Processing (CTOPP), developed by Wagner, Torgesen, and Rashotte (1999). It is based on their model of phonological processing ability which comprises three separate, though related, skills: phonological awareness tapped by tests of phonological blending and segmentation and the like; phonological working memory assessed by tests of memory span or nonword repetition; and phonological access in long-term memory evaluated by naming speed measures. There are two versions of the CTOPP, one for use in the 5-6 year age range and the other developed for use with older individuals, aged from 7 to 24 years. The CTOPP was normed on 1,656 individuals across the United States, and its manual reports extensive data on the impressive reliability and validity of the instrument. The core phonological awareness tests are measures of phoneme blending and phoneme deletion (referred to as "elision"); these measures in the 5-7 year age range correlate with reading 1 year later with correlations as high as 0.7 and 0.8. The CTOPP has measures of both memory for digits and nonword repetition which tap phonological working memory skills, together with color, picture, letter, and digit naming tasks that are sensitive to phonological access in long-term memory. Rapid naming and working memory tasks correlate significantly with later reading skills but the correlations are not as impressive as for phonological awareness with later reading; rapid naming tasks in 5-6-year-olds have correlations with later reading in the 0.6-0.7 region, while working memory tasks given at the same age correlate with later reading at a rather lower level (0.4-0.5). Children with severe dyslexia are likely to experience problems on all three types of phonological measure. However, there are some children who may have difficulty with one or two of the phonological tasks, but not all three. Phonological awareness tests are particularly strongly diagnostic of dyslexic difficulties but it is not uncommon to assess children who are able to blend and segment phonemes within words, but who exhibit short-term verbal memory limitations or naming speed problems or both. Whether these apparent individual differences in phonological skill reflect varying degrees of severity or complexity in dyslexia or whether there are qualitatively different subtypes of dyslexia is still an area of some debate.
Was this article helpful?
When parents help their children learn to read, they help open the door to a new world. As a parent, you can begin an endless learning chain: You read to your children, they develop a love of stories and poems, they want to read on their own, they practice reading, and finally they read for their own information or pleasure. They become readers, and their world is forever expanded and enriched.