Contributions of Phonological Awareness and Rapid Automatic Naming Ability To The Grown of Word-Reading Skills in Second To Fifth Grade Childen

You might also like

Download as pdf
Download as pdf
You are on page 1of 26
SCIENTIFIC STUDIES OF READING, 1(2), 161-185 Copyright © 1997, Lawrence Erlbaum Associates, Inc. Contributions of Phonological Awareness and Rapid Automatic Naming Ability to the Growth of Word-Reading Skills in Second- to Fifth-Grade Children Joseph K. Torgesen, Richard K. Wagner, Carol A. Rashotte, Stephen Burgess, and Stephen Hecht Department of Psychology Florida State University A longitudinal-correlational design was used to test the hypothesis that individual differences in rapid automatic naming make a unique contribution to explaining the growth of orthographic reading skills in 2 overlapping periods of development: second to fourth grade, and third to fifth grade, Separate analyses were done on the entire sample of approximately 200 children as well as on subsamples selected for impairment in word-reading development (bottom 20% and bottom 10% of readers). When second- and third-grade reading skills were not included in the multiple regressions, both rapid automatic naming and phonological awareness skills were strongly predictive of individual differences in reading 2 years later. With prior levels, of reading skill included in the predictive equation, rapid automatic naming ability did not uniguely explain variance in any of the reading outcome measures. In contrast, individual differences in phonological awareness in both second and third grades did uniquely explain growth in a variety of reading skilis over this developmental period. Results are discussed in the context of methodological issues in the use of longitudi- nal-correlational designs to study reading growth. Over the last 10 to 15 years, one of the most intensely examined questions in research on reading has involved the role of individual differences in phonological Requests for reprints should be sent to Joseph K. Torgesen, Department of Psychology, Florida State University, Tallahassee, FL 32306-1051. E-mail: torgesen @ psy feu.edu 162 | TORGESENET AL. processing ability in explaining variability in the rate of reading acquisition in young children. As the term is used in reading research, phonological processing refers to the mental operations that are involved when the phonological, or sound structure, of oral language is utilized in decoding written language. The vast majority of research in this area has used three broadly different kinds of tasks to assess phonological skills. These tasks are designed to assess phonological aware- ness, phonological short-term memory, and rate of access to phonological informa- tion in long-term memory. Performance on all three kinds of tasks has been shown to predict growth in early reading skill, and children with reading disabilities also differ consistently from normal readers on each kind of task (see Adams, 1990; Brady & Shankweiler, 1991; Share & Stanovich, 1995, for recent reviews of this work). Phonological awareness can be formally defined as one’s awareness of, and access to, the sound structure of oral language. Less formally, it involves the ability to notice, think about, or manipulate the individual sounds, or phonemes, in words. Phonological memory is usually assessed by tasks that require the brief, verbatim retention of sequences of familiar verbal items like digits or words, but it can also be assessed by tasks requiring the repetition of novel verbal strings like nonsense words (Stone & Brady, 1995). These tasks assess phonological short-term memory because they require the representation, or coding, of information in terms of its phonological features (Baddeley, 1986). Rate of access to phonological information in long-term memory has typically been assessed by rapid automatic naming tasks that require children to name, as rapidly as possible, a series of 30 to 50 familiar items (digits, colors, letters, or objects) printed on a page. When they are given in the serial format, rapid naming tasks obviously assess processes in addition to phonological ones (Wolf, Pfeil, Lotz, & Biddle, 1994). However, these tasks have sufficient face and predictive validity as phonological measures to include them in the “phonologi- cal family,” at least for the time being (Wagner, Torgesen, Laughon, Simmons, & Rashotte, 1993). For the past 7 years, we have been studying the nature and development of these Phonological processes, as well as their causal relations with reading growth, in a longitudinal study involving over 200 children. We have modeled the structure of individual differences on tasks measuring all three areas, and our analyses have shown that it is most accurate to think of these tasks as assessing separate, though correlated, abilities, rather than as assessing one general phonological ability (Wagner et al., 1993). We have also determined that these abilities are highly stable across the developmental span from kindergarten through fourth grade (Torgesen & Burgess, 1996; Wagner et al., in press). In fact, when these abilities were measured as latent variables, the median stability coefficient (which can range between values of 0.0 and 1.0) for the phonological measures was .77 between kindergarten and fourth grade, and .88 between first and fourth grades. These values GROWTH OF WORD READING SKILLS 163. were substantially higher than those for word-level reading skills over the same periods of time, which were .27 and .62, respectively. In our first report of the causal relations between phonological processing variables and reading growth (Wagner, Torgesen, & Rashotte, 1994), we covered the developmental span from kindergarten through second grade in 1-year incre- ments, That is, we examined the influence of phonological variables measured in kindergarten on reading acquisition in first grade, and performed a similar analysis of growth from first to second grade. Although we found that each of the phonologi- cal variables played a causal role in the development of word-reading ability when considered by itself, when they were considered as simultaneous causes in a single equation, only phonological awareness emerged as a unique predictor of reading growth. The causal influence of each of the other two variables, phonological memory and rate of access, were redundant with, and smaller than, that of phonological awareness. With the acquisition of additional waves of data in the longitudinal project, we were able to perform more sensitive analyses of causal relations between phonologi- cal processes and reading (Wagner et al. in press). Because word-level reading skills are very stable over 1-year time intervals, there is not very much individual difference variability to be accounted for in analyses that extend only over 1 year. Our new analyses were more sensitive because they extended over 2-year periods (kindergarten to second grade, first to third grade, and second to fourth grade). They showed that both phonological awareness and rapid naming ability make unique causal contributions to the growth of word-reading ability from kindergarten to second, and from first to third grades. However, only individual differences in phonological awareness continued to have a unique influence on growth in word- reading ability from second to fourth grades. In other words, individual differences in both phonological awareness and rapid naming ability are important in explain- ing variability in the growth of early word-reading skills (through the beginning of third grade), whereas differences in phonological awareness continue to be impor- tant in explaining growth at least through fourth grade. Our results for the early elementary grades are consistent with a number of other reports that document independent roles for both phonological awareness and rapid naming skills in explaining early growth of word-reading ability (Badian, 1993; Blachman, 1984; Felton & Brown, 1990). However, our results for the period from second to fourth grade are inconsistent with both theory (Bowers, Golden, Kennedy, & Young, 1994; Bowers & Wolf, 1993b) and other data (Bowers, 1995; Cornwall, 1992; Meyer, Wood, Hunt, & Felton, 1995; Wimmer, 1993), indicating that the abilities indexed by rapid naming tasks may be critically important as children move toward greater fluency in word-teading ability. Most accounts of reading development (Barron, 1986; Ehri, 1992; Share & Stanovich, 1995) suggest that the acquisition of fluent reading skills depends on the development of orthographic representations for words in long-term memory. 164 | TORGESENET AL. Although there is some divergence of opinion in this area (Wagn-~r & Barker, 1994), orthographic representations are usually thought to involve memory for the visual and spelling patterns that identify individual words, or word parts, on the printed page. Unitized orthographic representations provide a direct link between a word’s spelling and its pronunciation and meaning. These representations allow rapid identification of individual words as whole units, or as units composed of patterns larger than individual letters. Share and his colleagues (Share, 1995; Share & Jorm, 1987; Share & Stanovich, 1995; see also Ehri, 1992, 1996) have recently presented a compelling case for the role of phonological reading skills as a critical base for the development of fully specified orthographic representations of words. In this model, emergent skills in phonological decoding, which consist of letter-sound knowledge and a basic level of phonological awareness, provide the basis for acquiring accurate orthographic representations of words from the very beginning of the learning process. A basic proposition of their argument is that if children use partial or complete phonological cues to derive an approximate pronunciation for a word in text, and combine this approximate pronunciation with contextual constraints to identify the fully correct pronunciation, the prior attention to individual letters that is involved in alphabetic decoding provides a solid basis for acquisition or refinement of an orthographic representation for the word. Orthographic representations are acquired by repeated associations of a word’s correct pronunciation with its visual representation. As children’s increasingly developed alphabetic reading skills lead to more detailed analysis of the internal structure of words in print, they begin to acquire increasingly explicit and more fully specified orthographic representations. It is clear from Share and Stanovich’s (1995) analysis that phonological reading skills provide necessary, but not entirely sufficient, support for the development of good orthographic reading ability. That is, a child may be able to identify words accurately by using phonological and analytic strategies in combination with context, but if these skills are not applied in extensive exposure to print, the development of a rich orthographic reading vocabulary will not take place. Addi- tionally, a number of studies have now shown that there is unexplained variability in orthographic reading skill in children and adults even after partialling out the effects of general ability, phonological abilities, and print exposure (Barker, Tor- gesen, & Wagner, 1992; Cunningham & Stanovich, 1991; Stanovich & West, 1989). Although it is possible that this unexplained variability in orthographic skill is due to unreliable measurement of phonological reading skills or incomplete assessment of reading experience, it might also be due to an as yet unidentified information processing ability that directly affects the rate at which orthographic Tepresentations are formed. Recently, Bowers and her colleagues (Bowers et al., 1994; Bowers & Wolf, 1993b) suggested that this ability, or abilities, might be indexed by rapid automatic naming tasks. They proposed that “‘a largely inde- pendent individual difference factor, indexed by speed of naming simple visual GROWTH OF WORD READING SKILLS 165 symbols such as single digits or letters, has a significant effect upon learning and retrieving orthographic patterns” (Bowers et al., 1994, p. 173). Bowers and her colleagues argued against viewing rapid automatic naming tasks as primarily phonological in nature, and instead they emphasized the visual and speed components of these tasks. They suggested that rapid naming tasks may assess the functioning of a “precise timing mechanism” that is important in the complex interactive processes involved in amalgamation of phonological and visual representations to form orthographic codes for words. They hypothesized, “that slow letter- (or digit-) naming speed may signal disruption of the automatic processes which support induction of orthographic patterns, which, in turn, result in quick word recognition” (Bowers & Wolf, 1993b, p. 70). Although this theoreti- cal position is just beginning to be developed, part of the idea is that if individual letter identities cannot be accessed with sufficient ease and rapidity, they will not be processed in close enough temporal proximity to allow the learner to acquire sensitivity to the letter patterns in words. The evidence cited in support of this hypothesis is suggestive, but incomplete. For example, Bowers (1995) recently reported the results of a longitudinal study in which second-grade performance on measures of phonological awareness and rapid naming speed were used to predict performance on a variety of reading measures in fourth grade. One of the central findings of this study was that rapid naming ability made a significant independent contribution to the development of reading comprehension and word-reading latency, but not to the development of phonological reading skills (word analysis or word attack). Rapid naming’s inde- pendent contribution to the prediction of word-reading accuracy just failed to reach statistical significance. This evidence for the role of rapid naming ability in the development of orthographic representations for words is incomplete for at least two reasons. First, Bowers did not include second-grade reading level as one of the predictors of fourth-grade reading in her multiple regression analyses. Unless the autoregressive effects of prior reading level are included in these kinds of predictive analyses, it is impossible to know whether rapid naming ability independently influenced growth in reading over the developmental period in question, or whether its relations with fourth-grade reading were obtained because of its correlation with second-grade reading skills. Bowers and her colleagues proposed that individual differences in rapid naming skill have a causal relation with growth in orthographic representations after initial levels of accuracy in word reading are acquired. Causal inferences can only be drawn from longitudinal correlational studies to the extent that all relevant variables are included in the predictive equation. Although one can never be sure about the extent to which this assumption is met, early levels of reading are clearly one of the most powerful potential causes of later individual differences in reading and should be included in any analysis that attempts to explain growth or test theoretically based causal explanations. We will acknow- ledge (this issue will be discussed again later in the article) that the inclusion of the 166 TORGESEN ET AL. autoregressive effects of early reading on later reading in longitudinal regressions can itself introduce certain types of interpretive distortions. However, unless the autoregressor is included, longitudinal data cannot be used to examine causal hypotheses. “The second sense in which results such as those reported by Bower are incom- plete is that they show a relation between rapid naming speed and reading speed, but not between rapid naming speed and reading accuracy. If rapid automatic naming tasks are linked to reading fluency because they serve as a simple index of general processing speed (Kail & Hall, 1994), this does not necessarily imply that they play a unique role in explaining the ability to form orthographic repre- sentations. If, indeed, processes indexed by rapid naming tasks play a uniquely important role in support of the acquisition of orthographic representations for words, their effects should be seen in measures of orthographic accuracy as well as in speed of reading. With regard to this last point, two types of evidence have been reported. Golden and Bowers (1993a; as cited in Bowers et al., 1994) reported that children in first, second, and third grade who performed well on measures of rapid automatic naming for digits were more sensitive to letter patterns within words than were children who named digits more slowly. They also indicated that phonological awareness was related to sensitivity to letter clusters, but that there was a significant degree of independence between the contributions of these two factors to letter-cluster ability. Bowers and Wolf (1993a; as cited in Bowers et al., 1994) also reported a reanalysis of the data from Bower’s longitudinal study (Bowers, 1995) in which they divided 38 fourth-grade children into four groups: (a) fast readers who were also good at phonological decoding, (b) fast readers poor at phonological decoding, (c) slow readers who were good at phonological decoding, and (d) slow readers who were poor at phonological decoding. In comparing the reading skills of these groups, they found that phonological decoding skills were strongly associated with scores on untimed word-reading accuracy, untimed reading comprehension, and phonological awareness tests. Speed of reading was associated with untimed reading comprehension, number of correct identifications of both exception words and less frequent words, story-reading speed and errors, and digit-naming speed. The important point for our purposes is that speed of reading, but not phonological decoding skill, was associated significantly with accuracy of reading exception words that require specific orthographic knowledge if they are to be read accurately out of context. The major limitation of both of the studies cited here, of course, is that they are based on concurrent assessment of naming speed and outcome measures, and so are not really appropriate to explain individual differences in the acquisition of orthographic reading skills. The purpose of this study was to evaluate the contribution of rapid automatic naming ability to the growth of orthographic reading skills within a large randomly selected sample for whom both measures of orthographic knowledge and reading GROWTH OF WORD READING SKILLS 167 speed are available. Because Bowers and her colleagues (Bowers et al., 1994) made the point that variation in rapid naming ability may be more important in explaining differences in growth of orthographic skills when it is severely impaired within reading-disabled samples than when it falls within normal limits in large random samples, our analyses will be conducted with both the entire sample and with a subset of children with reading disabilities. We will report analyses of growth in reading skills from the beginning of the second grade through the end of the fifth grade in overlapping 2-year increments, as this is a developmental period in which important and rapid development of orthographic knowledge takes place. If performance on rapid naming tasks does index visual and timing processes uniquely important to the induction of orthographic patterns, it should contribute independently to growth on untimed word-reading ability, accuracy of orthographic reading, reading comprehension, and reading speed, and should be less important in explaining growth in phonological reading ability. Additionally, if these processes contribute uniquely to individual differences in growth of orthographic reading skill that takes place during the second- to fifth-grade period, they should also explain differences in reading outcome that are independent of prior reading status. METHOD Participants The number of participants involved in our analysis of growth from second to fourth grade, and from third to fifth grade, was slightly different because of attrition in the longitudinal sample. There were 215 children in the first analysis, and 201 in the second, We originally began with 284 children randomly sampled from kindergarten classrooms in six elementary schools in Tallahassee, Florida, All participant attrition occurred as a result of children leaving the Tallahassee area. Earlier analyses of dropouts through fourth grade indicated that there were no differences, in terms of major demographic variables or performance on reading and phonological tasks, between the remaining and original sample (Wagner et al., 1994; Wagner et al., in press). The 15 children who left the sample between the fourth- and fifth-grade testing also did not significantly affect the overall characteristics of the sample. The 216 children in the second- to fourth-grade analyses consisted of 53% girls and 26% minority groups, primarily African American, and the composition of the sample (201 children) in the third- to fifth-grade analyses was exactly the same. Materials Each year, the measures given to our sample included (a) tests of phonological skill including measures of phonological awareness, phonological memory, and rapid 168 TORGESEN ET AL. automatic naming; (b) measures of reading ability; and (c) a measure of general verbal aptitude. In the fourth and fifth grades, measures of orthographic processing and reading fluency were added to the battery. Only those measures used in the present analyses will be described here. Descriptions of other measures given in the early grades can be found in earlier reports on this sample (Wagner et al., 1994). Phonological processing tasks included measures of phonological awareness and rapid automatic naming. The awareness tasks were: 1. Phoneme elision: Children were asked to say a word, then to say the word after deleting a specified phoneme. For example, children were instructed to “Say the word ‘book.’ Now tell me what word would be left if I said ‘book’ without saying /b/.” The phoneme to be deleted was always a consonant, the position of which varied at random. The remaining phonemes formed a word. For all grades, there were four practice items. For the second grade, there were 15 test items consisting of words containing three to five phonemes and one or two syllables. At third grade, there were 15 different test items that consisted of words containing three to six phonemes and one or two syllables. For fourth and fifth grades, 5 more difficult items were added to the list of 15 items that were given in the third grade. 2. Phoneme segmentation: Children listened to words and were instructed to “tell me each sound you hear in the word in the order that you hear it.” There were 4 practice items. For second grade, there were 15 test items consisting of words containing two to five phonemes and one or two syllables. At third grade, 4 test items consisting of words of five to eight phonemes and three syllables were added to the original 15 items. Finally, at fourth and fifth grades, 6 more test items were added, giving a total of 25 test items. The 6 additional items consisted of words containing seven to eleven phonemes and four to six syllables. 3. Blending phonemes into words: Children listened to words presented pho- neme by phoneme at a rate of two per second and were asked to pronounce the words that resulted when the phonemes were blended together. There were six practice items. The number of test items for second grade students was 15, consisting of two- to six-phoneme, one- to two-syllable words. At third grade, 5 more items were added that included seven- to eight-phoneme, three- to four-syllable words. The total of test items at fourth and fifth grades was 25, with 5 items added to the 20 third-grade test items. These final 5 items consisted of nine- to eleven-phoneme, three- to four-syllable words. The rapid automatic naming tasks were: 4. Naming digits serially: Six rows of five single digits per row were arrayed on a card and children were instructed to name the digits as fast as they could, beginning with the top row and continuing to the bottom. Two trials were given using two cards with differently arranged numbers. Times were recorded on a GROWTH OF WORD READING SKILLS 169 stopwatch, and scores were based on average time for the two series. Performance was converted to number of items named per second, so as to make higher scores indicative of better performance. 5, Naming letters serially: This task was identical to the naming digits serially task, except that the stimuli were common letters instead of digits. Again there were two trials with scores based on average time for the two series. ‘The reading measures included: 6. Word identification: This subtest, which required children to name individu- ally presented words, was taken from the Woodcock Reading Mastery Test (Wood- cock, 1987). 7. Word analysis: This subtest, which required children to name individually presented nonwords, was also taken from the Woodcock Reading Mastery Test. 8. Passage comprehension: This subtest uses the cloze procedure to assess passage-level comprehension. Children were required to supply a missing word within a sentence or brief passage that they read themselves. It was also taken from the Woodcock Reading Mastery Test. 9. Word-reading speed-isolated trials: Words appeared one at a time on a computer screen in approximately 1-in. high letters, and the child’s task was to name the word as rapidly as possible. Instructions were, “I want you to read some words on the computer screen as fast as you can. Try to read them correctly, but fast.” The task was controlled by an Apple Ie computer with the words appearing in white on a black background. Rate of presentation was controlled by the child, who pressed the space bar to indicate readiness for the next word. Speed of response was measured by a voice key, and was the time from onset of the visual signal until the child began to make a verbal response. After four practice trials, the child was shown how the computer kept track of errors and time of response. The test trials began when the child pushed the spacebar indicating readiness for the first item. The measure derived from this task was the median response time for correct responses. For the third-grade assessment, there were 20 items consisting of high-frequency one- and two-syllable words of approximately second-grade level of difficulty. In the fourth and fifth grades, the 20 items were changed to short exception words (e.g., chorus, lose, deaf, dough, and anchor). 10. Word-reading efficiency: Four columns of 26 words were arrayed on an 8-by-11-in. card and children were instructed to read the words column by column as fast as they could without making errors. The words were printed in 16-point Courier typeface. The children were told to “read the words in order, but if you come to one you can’t read, just skip it and go to the next one.” They were also encouraged to use their finger to keep track of their place in the columns. They were also told, “the words start out pretty easy, but they get harder as you go along. Just read as many words as fast as you can until I tell you to stop.” The children first 170 TORGESEN ET AL. practiced reading a column of eight single-syllable words. They were then shown a covered stimulus card, and instructed to begin reading as soon as the cover was lifted. Timing began when the first response was made, and the child was told to stop after 45 sec. The word list began with very simple single-syllable words (me, up, no, he, to), and ended with multisyllable words of much greater difficulty (participate, historical, repugnant, permutation). The same word lists were pre- sented to both fourth- and fifth-grade children. There were two equivalent forms of the test, and the child’s score was the average number of words read in 45 sec across the two forms. 11. Orthographic choice: The children were told, “You will first see a square in the middle of the screen that tells you to get ready. Then you will see two words. One of these is a real word, and one is not a real word.” They were then told to press a key on the right or left side of the keyboard, corresponding to the side of the screen on which the real word appeared. This task is sensitive to orthographic knowledge because the word pairs were phonologically similar, but only one of them represented the word’s correct spelling (i.e., smoke vs. smoak, coat vs. cote, scare vs. scair). Phonological analysis of both words would lead to the same pronunciation, so a correct response had to be based on a fully specified ortho- graphic representation for the word in question. Stimuli were similar to those for the word-reading speed-isolated trials task, and response timing was accomplished ina similar manner, except that timing was terminated by a button press rather than a vocal response measured by a voice key. The child received 4 practice trials followed by 30 response trials. Rate of presentation of trials was controlled by the child who pushed a spacebar when ready for the next trial. The same words were used on this task for both the fourth- and-fifth grade administration. Speed of response on this task was the median response time for correct trials, and number correct was the total number of items answered correctly. 12. Lexical verification: On this task, the children were told, “I will say a word to you, and then I will press the space bar. If the word I said is spelled correctly, press this key. ... Work as fast as you can, but also try to get as many correct as you can.” The rationale for this task is similar to that for orthographic choice, as the misspellings that were presented were all phonologically correct, and the child’s response had to be based on an orthographic representation of the word’s correct spelling (ie., peeple vs. people, frunt vs. front, enuff vs. enough). On half the trials, the word was spelled correctly, whereas on the other half a phonologi- cally correct misspelling was presented. The same words were used for both the fourth- and fifth-grade versions of this task. Stimuli and timing were presented and timed in a manner similar to the orthographic choice task. After four practice trials, the child received 20 test trials. Speed of response on this task was the median response time for correct trials, and number correct was the total number of items answered correctly. GROWTH OF WORD READING SKILLS 171 We estimated general verbal aptitude by administering a standardized measure of oral vocabulary. 13. Stanford-Binet Vocabulary: Children were asked to orally define words using the Vocabulary subtest of the Stanford-Binet intelligence test (4th ed.; Thorndike, Hagen, & Sattler, 1986). Vocabulary consistently is the subtest most highly correlated with general verbal ability (Sattler, 1992). Procedure The tasks were individually administered by trained research assistants in the fall of the second and third grades, and in the spring of the fourth and fifth grades. Thus, in each of our developmental analyses, approximately 2% years elapsed between testing points. All testing was done in either mobile research trailers on school grounds, or in space provided within the schools. Task order was randomized, and the tasks were administered over three or four sessions within a 2-week period for each child. RESULTS Basic statistics for the variables used in the second- to fourth-grade analyses are presented in Table 1 and those for the third- to fifth-grade analyses are presented in Table 2. These statistics include means, standard deviations, reliabilities, and bivariate correlations. The results reported in Tables 1 and 2 suggest that the tasks had adequate psychometric characteristics for the age ranges included in this study. All reliabilities were in the acceptable range, and performance was free from floor and ceiling effects on almost all the tasks, Modest ceiling effects occurred for the orthographic choice and lexical verification accuracy measures in the fourth and fifth grade. All correlations were in the expected direction, and there were no major differences in the pattern of correlations at the different age levels involved in the two sets of analyses. In previous analyses of our longitudinal data, we used multiple measures of each of our constructs to form latent variables representing the common variance among several tasks measuring the same construct (Wagner et al., 1994; Wagner et al., in press). Because they represent only the variance common across tasks, latent variables provide relatively error-free measurement by excluding task-specific sources of error. In this study, we did not construct latent variables because some of our analyses involved sample sizes that were insufficient to support reliable estimation of latent variables. However, we did continue to employ multiple measures for some of our constructs, but in this case we constructed unit-weighted composite scores rather than latent variables. Although unit-weighted composites do not eliminate task specific variance to the same extent as latent variables, they ood ot 8% st © we ror eet of ss Cor Tu vO so ee ee Bt as sor ese GSS OLE TLE ORGIES LEE OL TLL n Ose Oe crese pO event OO seve oy aeeiOOL cser Ovusesice veseviep sensu: OL ene cseereseieriaeita gE st argssod uname ST Te ver eset Chee Oreo tec erred Gees z 1 6s) 6s i 66S 8 to 9 w is ey er av Ty p-Aoemooe uoneoyt9a [roo] 7 a a woos ow wv we 6k ve p-fovanooe earoyp omydetoyiQ, (6) se wo or tr ue uw o Ok us ‘p-AoUaIOYJ® PLO Gs) 6s Sot ey 9 +s wu fy ee 6z or p-peads Surpear prog, (6) sor tL $9 Rei eeeeeeeeaee ow os p-Uorsuayaidwod aBesseq CCG) om seses Genero enn raeetiACatae Oc seat Dy aneiOO caHeiTeacaeae(0)" sKTeUE PAOAK, 66) w sek Oe 6S fr ‘p-uoHeaAUAPT PION @) oor sr OPEN ae eeeea 8c se q-Areingesoy, (96) 9L nn a sr com z-uorsusyasduos adusseg 06) 98 oH zestsdyome p1ogy Ws) 9 uw Mm ea (6) 98 66 (6) vee we ‘c-Stoquinu SuwreN, (sx) 9g" OL ‘g-Burpuaiq euiauoyg (es) ar TYoneuewdes swoUoyE (06) Zeorsyo su9HOKU sesheuy epeip-YpiNO 0} -PUOdEg UI PEAOAUI SeqeUeA 10} SoRSAEIS ennduoseg pue ‘(jeuobeig Uo) SeAIIGeNEY HEH-HIGS 40 “eydiy YoRquoLD ‘suOyeIaL0D - aay -rouuo oeudosdde we ut soup asuodsou renprsrpur aavs 10u pxp raandutoo amp asne99q 1801 sINN Yo} pare|NoqE aq 10U pInoo AamIENeN BWoKpU SoNSECT 2ON v ¢ wm ef ¢ Ta ol 6 of ce FY 68 Tm ct 0 SO oY OF eF as ol 8 TU ce Of «GW Fe oe CUM Om FL MOE EI «Ts Of Le ce c# FO " -— = @ %& — wi 8 SS OO F — 9 S&F om — — w@ o st sqqssod wwe Cer eehiee Op eacrasaa pn saterens oraHeired orien aan grants aio tgueivae nese raaey We) eo 0 oO | Oe OU Oa s-paeds woneogtion e821 G6) 1 so ws ve we se wa ws ew or ar wr wr a we ‘S-poads aoyoup atudeBo4210 (9) 00 ts or we ov v oe w ve sy Te 0€ or or st g-Aopanaoe wopeagtiag ST as) « c we w@ ew we w © © MW Or ¢-fornaoe oopoyp oTydedouIO oe ow oe «© ow ot 6 Mm ww ie -paads Buypeas pio, ue yw w $ fous. iON @ wow ow oe wb ow uw om sr s-uopsuayouduios fesse o) of swe or (96°) ss wae ie or a Sy a @ «cis w ¢ tw @ eat x iu is us ae" eo) ow ey aw Go) oe or : Ww) wos (os) oe ) st se s-sraquinu dune (6s) es ‘¢-TuIpuayg owuoyd (8) g-uoqrenmourdos auonoy SeshTeuy @PEAD-UYI4 O} -PUILLL Ul PAAIOAU ‘sejqeueA 10} sonSHEIS eAnduoseg puke ‘{jeuobeIq uo) saMIIqeIey HEH-HIdS 40 ‘eydiy YoRquosD ‘suORE|eL0D E-worsHe auiauoYe 174 TORGESEN ET AL. do reduce the possibility that estimates of relations in regression analyses will be specific to one particular set of stimulus materials or method of measurement. All composite variables were formed as an average of the standard scores for each task used to construct the composite. The tasks used to construct each variable, at each age level, are given in Table 3. For most of the variables, the composite was formed from all of the tasks in the battery that were included to measure the construct in question. However, in the case of phonological awareness, the three variables used were those that had the largest weights for the awareness factor in a factor analysis involving all awareness and naming tasks in the battery. As can be seen from Tables 1 and 2, the variables used to form each composite were substantially correlated with one another. The primary analytic goal of this study was to determine the extent to which individual differences in phonological awareness and rapid naming ability contrib- ute to changes in the pattern of individual differences for reading skills over two developmental periods. Results for the period from second to fourth grade, for the whole sample, are reported in Table 4. The results in this table were obtained from two different sets of regression analyses. The top panel reports the proportion of variance in reading outcome uniquely accounted for by phonological awareness and naming without controlling for the autoregressive effects of second-grade TABLE 3 ‘Tasks Used to Form Each Composite Variable at Each Age Level Grade Level Variable 2 3 4 5 Phonological awareness Blision Elision Segmenting Segmenting Blending Blending Rapid naming Naming digits Naming digits Naming letters Naming letters Word reading speed Word reading Word reading speed Word reading speed speed Word efficiency Word efficiency Orthographic accuracy Orthographic choice Orthographic accuracy choice accuracy Lexical verification- Lexical accuracy verification- accuracy Orthographic speed Orthographic choice-speed Lexical verification- speed GROWTH OF WORD READING SKILLS 175 TABLE 4 Proportion of Variance in Fourth-Grade Reading Outcomes Uniquely Attributable to Individual Differences in Second-Grade Phonological ‘Awareness and Rapid Naming Ability Fourth-Grade Reading Outcomes Word Word Reading Reading — Orthographic Second-Grade Variables Identification Analysis. Comprehension Speed Accuracy Vocabulary ait age ase 20" 06 Phonological awareness a4 pees “2s o2** ore Rapid naming et o4s* oan ae ore Total variance explained 55 46 31 31 30 Autoregressor est sist Sane age Vocabulary oi 00 oo ot Phonological awareness 03* 068 03" 00 ‘Rapid naming 00 or 00 00 Total variance explained 69 59 60 Al Now. N= 215. *p <.05.*p < OL. ***p < O01. reading skill. The bottom panel reports the same data when controlling for these autoregressive effects. For Teasons explained earlier, the bottom panel results are those most appropriate for examining the influence of phonological awareness and naming on growth in reading skills over the developmental period in question, but. the top panel results are also reported for comparison with other research that has not used the autoregressor in analyses such as these. All analyses were performed using hierarchical regression procedures. In analyses involving the autoregressor and vocabulary scores, these variables were entered first, in the order listed, so the proportion of variance assigned to vocabulary is that obtained after the effects of the autoregressor are accounted for. The values reported for phonological awareness and naming represent the proportion of variance they uniquely account for after all other variables are entered in the equation, regardless of the order in which they appear in the table. Thus, the bottom panel shows that individual differences in second-grade phonological awareness accounted for 3% of the variance in fourth-grade reading comprehension after second-grade reading comprehension, vocabulary, and rapid naming ability were entered in the equation. In contrast, the same type of analysis showed that rapid automatic naming skill accounted for no unique variance in reading comprehension after the other variables were entered, In all cases except for the measure of orthographic accuracy, the autoregressor used was the corresponding reading skill measured in second grade. Because we did not give the orthographic tasks in second grade, the autoregressor used in this analysis was word- reading accuracy. There was no analyses in the bottom panel for word-reading speed, because reading-speed measures were not given in second grade. 176 TORGESEN ET AL. ‘The results in Table 4 indicate that both phonological awareness and rapid naming ability made significant contributions to the prediction of fourth-grade reading skills independently of each other and of general verbal ability. However, when level of reading skill in second grade was taken into account, only individual differences in phonological awareness continued to be important in explaining variability in fourth-grade outcomes. Although rapid naming was independently related to growth in phonological decoding ability, the relation was very weak. Both phonological awareness and rapid naming predicted orthographic reading ability in fourth grade, although neither of them explained variance beyond that accounted for by word-reading ability in second grade. To study the relations among these variables in a sample of impaired readers, we repeated these analyses using children who fell in the bottom 20th percentile in performance on the Word Identification subtest in second grade. This sample of 43 children had an average reading-grade level of 1.4 on the Word Identification subtest, and their corresponding sample-based standard score (M = 100, SD = 15) was 78.0 (range = 65 to 87). These children were approximately a year behind in reading development in the first semester of second grade. The results of our analyses for this subsample are reported in Table 5. In comparing these results to those for the whole sample, the major difference is that individual differences in phonological awareness in second grade play a much more important role in explaining growth in word-reading ability and reading comprehension. Whereas differences in phonological awareness explain only 3% of the variability in growth of reading comprehension skills in the whole sample, they uniquely account for 19% of this variability in the sample of poor readers. Once again, individual differences in rapid automatic naming ability do not explain variability in the growth of orthographic reading skills. We also repeated these analyses with the 21 children who fell below the 10th percentile in reading ability in second grade, and obtained essentially the same results. This was a severely impaired sample, with average grade-level performance on the Word Identification subtest falling at the 1.2 grade level, with a corresponding standard score of 73.2. Analyses for the whole sample covering the period from third to fifth grade are presented in Table 6. A composite variable representing speed of response on orthographic reading tasks was added to this analysis, and we were able to perform an analysis of growth in word-reading speed because we had a word-reading speed measure available from third grade. This reading-speed measure was used as the autoregressor for both the word-reading and orthographic-speed measures. The overall pattern of results in this analysis was similar to those for the second- to fourth-grade results. Once again, individual differences in both phonological awareness and rapid naming ability uniquely explained significant proportions of the variance in fifth-grade reading skills when the autoregressive effects of third- grade reading skills were not taken into account. However, only individual differ- ences in phonological awareness continued to play a small role in explaining TABLE 5 Proportion of Variance in Fourth-Grade Reading Outcomes Uniquely Attributable to Individual Differences in Second-Grade Phonological Awareness and Rapid Naming Ability Fourth-Grade Reading Outcomes Word Word Reading Reading Orthographic Second-Grade Variables Identification Analysis Comprehension Speed Accuracy Vocabulary 06 05 at 09 1 Phonological awareness 250 2a 2g 00 06 Rapid naming .06* 00 05** aoe 05 ‘Total variance explained 56 32 2 ST 20 Autoregressor ae Agee Bott 30+ Vocabulary 01 00 01 00 Phonological awareness 4 03 sl9*** 00 Rapid naming 00 00 01 00 Total variance explained —_.69 a) ey Al Note. For the bottom 20% of students, +p < 05. "p< 01. ***p < 001. TABLE 6 Proportion of Variance in Fifth-Grade Reading Outcomes Uniquely Attributable to Individual Differences in Third-Grade Phonological Awareness and Rapid Naming Ability Fifth-Grade Reading Outcomes Word Word Reading Reading Orthographic Orthographic Third-Grade Variables Identification Analysis. Comprehension Speed Accuracy Speed Vocabulary aoe gee ages a6 owe 06 Phonological awareness 1s 20%" alee ogee 00 1 Rapid naming ogres oe oe Ast ores ate Total variance 60 32 60 st 14 aL explained Autoregressor sore anes samen eee se ate ‘Vocabulary 00 OL lores 00 00 00 Phonological awareness .02* 02 02" a1 oO o1 Rapid naming 00 00 00 00 oO 00 ‘Total variance 2 68 1 ST 4 2s explained Note N= 201. *p < 05. *p <1. *p < 001. 177 178 TORGESEN ET AL. variability in growth, or change in the pattern of individual differences in reading skill, from third to fifth grade. Once again, these analyses were repeated with a sample of 41 children who obtained scores below the 20th percentile on the Word Identification subtest in third grade. This group of children had an average grade equivalent performance of 1.9 on the Word Identification subtest, with a corresponding mean sample standard score of 78.8 (range = 45 to 90). The results of this analysis are reported in Table 7. For this sample of poor readers, rapid naming ability in third grade explained a significant proportion of variance in orthographic accuracy in fifth grade when the effects of third-grade reading accuracy were not included. How- ever, with the autoregressor included, it was individual differences in phonological awareness that accounted for unique variance in the outcome measure. This suggests that the processes important for orthographic development that are indexed by naming speed were fully captured by the word-reading accuracy measure at third grade. When these analyses were repeated with the more severely impaired (grade level = 1.7; standard score = 71.1 on the Word Identification test) children in the bottom. 10th percentile of reading ability, the pattern was the same except that the analyses including the autoregressor explained a much larger proportion of variance in orthographic accuracy and speed (.70 and .57, respectively). In addition, with the autoregressor in the equation, rapid naming ability explained a significant propor- TABLE 7 Proportion of Variance in Fifth-Grade Reading Outcomes Uniquely Attributable to Individual Differences in Third-Grade Phonological Awareness and Rapid Naming Ability Fifth-Grade Reading Outcomes Word Word Reading Reading Orthographic Orthographic Third-Grade Variables Identification Analysis. Comprehension Speed Accuracy Speed Vocabulary 20° 30° aie 00 age 00 Phonological awareness. 08 2arew ove 06 0 00 Rapid naming aa 0 oot 13 aa 12" “otal variance 49 36 55 30 25 a3 explained Autoregressor Tae gguee sere ane Aor ats Vocabulary 02 0 18* 00 OL 00 Phonological awareness OL om om or ose 0 Rapid naming 1 o o OL 00 04 Total variance 16 48 61 42 49 as explained Note For bottom 20% of students, n = 41. *p <.05.™%p < 01. ™p < 001. GROWTH OF WORD READING SKILLS = 179 TABLE 8 Unique Contributions of Orthographic Reading Skill to Word Identification Accuracy and Passage Comprehension in Fourth- and Fifth-Grade Children Fourth Grade Fifth Grade Reading Variable Predictor R’ R’ Change RR Change Word identification Vocabulary AL 38 Phonological awareness 63 .22** = 59 24 Orthographic accuracy 73.10" 65 06* Passage comprehension Vocabulary 54 50 Phonological awareness 65 .11** 62 2s Orthographic accuracy 73 .08** = 66 ose *p <.0L. p< .001. tion of variance (15%) in speed of response on the orthographic reading measure, while phonological awareness uniquely explained 12%, which just failed to reach significance. Neither of these variables explained a significant proportion of variance in orthographic accuracy beyond that explained by the autoregressor of third-grade word-reading accuracy. Because there are unresolved questions about the best methods of measuring orthographic reading skills (Bowers et al., 1994; Wagner & Barker, 1994), it is important to demonstrate that the composite measure of orthographic accuracy used in this study measured skills similar to those assessed in previous research. This was done by examining the concurrent relations between our orthographic measures and other reading variables. Specifically, we conducted a series of hierarchical regressions to determine the proportion of variance in word-reading accuracy and passage comprehension explained by our orthographic accuracy measures after general verbal ability and phonological ability were entered in the analyses. In contrast to our earlier longitudinal analyses, these are concurrent analyses, with all variables at each grade level given at approximately the same point in time. In all the analyses, the phonological measure is the same composite of phonological awareness tasks used earlier, and the orthographic measure is the same composite for accuracy used in our developmental analyses. The analyses are reported in Table 8. In all cases, our measures of orthographic reading skill uniquely account for a significant proportion of variance in general word-reading ability and passage comprehension at both grades. Because the level of magnitude of these effects is very similar to that reported in earlier studies (Barker et al., 1992; Cunningham & Stanovich, 1991; Stanovich & West, 1989), we conclude that our measures of orthographic skills provided an acceptable assessment of this area of reading skill. 180 | TORGESENET AL. DISCUSSION The focus of this article is on the independent contributions of individual differences in phonological awareness and rapid naming ability in explaining variability in growth of orthographic reading skills between second and fifth grades. The clearest test of the hypothesis that rapid naming tasks measure processes critical to the formation of orthographic representations for words was provided in the analysis of growth on the orthographic accuracy measure that included the autoregressive effects of earlier reading skill. These analyses indicated that performance on rapid naming tasks did not explain variability in fourth- and fifth-grade orthographic accuracy beyond that explained by level of word-reading ability in second and third grade, This was true not only for analyses involving the whole sample, but also for analyses involving subsets of moderate and severely impaired poor readers. Is this finding inconsistent with the hypothesis proposed by Bowers and Wolf (1993b)? It is clearly inconsistent with a strong form of the hypothesis suggesting that individual differences in rapid naming ability can explain which students, among groups of poor readers with similarly impaired word-reading skills, will show growth in orthographic reading skill, and which students will not. That is, we found no evidence that performance on rapid naming tasks can predict growth in accuracy of orthographic processing beyond the prediction provided by level of prior word-reading accuracy. We did have one result that was consistent with this hypothesis, involving growth in speed of response on the orthographic tasks from third to fifth grade in severely impaired (bottom 10%) poor readers. In this analysis, performance on rapid naming tasks uniquely accounted for 15% of the variance beyond that explained by third-grade word-reading speed. However, this result is weakened by the fact that we did not obtain a similar result for the word-reading speed measure, which should also have been sensitive to development in speed of orthographic processing. Although inclusion of the autoregressive effects of prior reading level is critical if we are interested in causal explanations of individual differences in growth over specified developmental periods, it is important to recognize that our conclusions apply only to the specific developmental periods examined in the analyses. Further, because inclusion of the autoregressor in the predictive equation may mask the real effects of other variables (by capturing any causal effects it may have in common with them), our analyses may underestimate the influence of both rapid automatic naming and phonological awareness skill on growth in reading from second to fifth grade. For example, we have already established that rapid naming ability plays a unique causal role in explaining differences in growth of word-reading ability between kindergarten and second, and between first and third grades (Wagner et al., in press). If our second-grade autoregressor for orthographic accuracy (word- identification accuracy) was influenced by individual differences in ability to form orthographic representations for which rapid naming ability has been proposed as GROWTH OF WORD READING SKILLS 181 an index), then this ability is accounted for in our regressions prior to the entry of rapid naming scores in the predictive equations. Thus, a unique factor related to the ability to form orthographic representations could have a strong influence on the development of children’s orthographic vocabulary during the late elementary school period, but this factor would be represented in our equation by word-reading ability rather than the rapid naming tasks. If this is the case, then it would be an interpretive distortion to conclude from our results that abilities indexed by rapid automatic naming tasks have no direct influence on the growth of orthographic reading skills from second to fifth grades. In defense of our strategy to include the autoregressor, it is important to remember that Bowers (Bowers et al., 1994) proposed a causal hypothesis that applies expressly to the developmental period we studied. She proposed that rapid naming ability may particularly affect the development of orthographic representations, which undergo particularly rapid development from the third through fifth grades. During this period of time, children are expected to learn new words at a much faster rate than during the first two elementary grades, and they are also exposed to a much broader range of new words (Share & Stanovich, 1995). Thus, if processes indexed by rapid automatic naming skill provide special support for the acquisition of orthographic representations, their effects should be particularly apparent during this period. Although the autore- gressor may reflect to some extent the earlier influence of rapid naming ability, if Bowers’s hypothesis is correct, processes indexed by rapid automatic naming tasks should be even more influential during the later elementary period exam- ined in this study. This differential pattern of influence over early versus later development of word-reading skills should be manifest in our equations by unique variance in orthographic accuracy associated with second- and third- grade performance on the rapid naming tasks. Although we have suggested that there are important interpretive limitations of analyses that include an autoregressor, the analyses reported at the top of our tables (those without the autoregressor) have even more important limitations when it comes to making causal inferences. For this task, they are no better than concurrent correlations, in that they have implications only for prediction and none for causation. In addition to the ambiguities in the interpretation of our results arising from limitations in the longitudinal-correlational methods available to us, the results themselves may have been influenced by certain characteristics of our sample. First, our sample was more heterogeneous, with regard to demographic variables such as socioeconomic status and ethnicity, than many other samples used in this area of research (Bowers, 1995; Wolf, Bally, & Morris, 1986). Thus, the relatively subtle contributions of rapid naming tasks to the prediction of reading outcomes beyond the autoregressive effects of prior reading may have been lost within the context of these other, more powerful noncognitive variables. 182 TORGESEN ET AL. Another variable that is likely to influence the findings of studies such as the present one is the instructional opportunities available to students that have an impact on growth of reading skills across the elementary school period. For example, in Bowers’s (1995) longitudinal study, she reported that half of the children classified as poor readers in second grade were no longer classified as poor readers in fourth grade. Similarly Meyer et al. (1995) reported a longitudinal study in which they emphasized the considerable growth in reading of many children in their sample of poor readers, and which also reported one result that was directly inconsistent with our findings. These investigators used a sample of 64 children with reading disabilities, and they found that rapid naming ability, measured by tasks very similar to ours, accounted for 12% of the variance in fifth-grade word-reading accuracy after third-grade reading accuracy was controlled. In a very similar analysis with our sample, we found that third-grade word-reading ability accounted for 73% of a total 76% of variance in fifth-grade reading that was explained by all variables. Rapid naming tasks accounted for only a nonsignificant 1% of variance. The high correlation of the third- with fifth-grade reading in our sample suggests that the pattern of individual differences in reading within this group of poor readers was very stable over this period. Compared to the sample used by Meyer et al. (1995), our sample had very similar beginning scores on the Word Identification measure (our standard score = 78.8, theirs = 79.5) and general verbal ability (our standard score = 89.6, theirs = 90). However, very few of the children in our sample of poor readers showed significant growth in word-reading ability (relative to normal performance) over the period from third to fifth grade. Only 10% of the children made substantial gains (greater than 1 standard deviation) or achieved performance in the average range in fifth grade. Unless poor readers are exposed to effective instruction across the entire elementary school period, whatever cognitive strengths they do possess (i.e., good orthographic processing skills) will have little opportunity to manifest themselves in significant growth in reading ability. The considerations about sample characteristics outlined earlier suggest that our understanding of the relative contributions of the skills and knowledge indexed by phonological awareness and rapid naming tasks to the growth of reading ability will be enhanced to the extent we are able to include more of the relevant variables in our analyses of reading growth. In addition to the factors mentioned earlier, characteristics of a language’s orthography itself may pow- erfully impact the relative contributions of phonological awareness and rapid naming ability to the growth of word-reading ability. For example, Wimmer (1993) provided evidence suggesting that, as a language’ s orthography becomes more transparent (relations between letters and sounds more predictable), individual differences in phonological awareness may become less important in explaining reading growth, whereas differences in rapid naming ability become more important. GROWTH OF WORD READING SKILLS) 183. In addition to the inclusion of more relevant variables in our longitudinal analysis of reading growth, our understanding of rapid naming ability’s relation to reading development, in general, and orthographic development, in particular, will be enhanced to the extent that we make progress in dissecting the component skills involved in performance on rapid naming tasks. Serial naming tasks require the complex coordination of many different processes (Wolf, 1991). If we can isolate the particular components of rapid naming performance that are responsible for its predictive relations with reading, this may take us an important step forward in our understanding of the development of orthographic reading skills. Although work is proceeding rapidly in this area (Kail & Hall, 1994; Wolff, Michel, & Ovrut, 1990), the work needs to move eventually from component analyses of the tasks themselves to study of the components in their relations with reading growth. Apart from the important qualifications about the meaning and generality of our results outlined earlier, we want to clearly distinguish between a study such as this one that tests a specific causal hypothesis over a limited developmental period, and studies that address issues of prediction. Our results should not be taken as evidence that rapid automatic naming tasks are not important predictors of growth in word-level reading, though indeed they are (Torgesen & Wagner, 1995; Wolf, 1991). In our earlier work, rapid automatic naming tasks were shown to be particularly powerful predictors of word-level reading growth during the early elementary grades (Wagner et al., 1994, in press), and during this developmental period, they predict variance in reading growth beyond that predicted by earlier word-reading ability. These results should also not be taken as evidence that rapid automatic naming tasks might not be useful in the diagnosis of particular kinds of reading problems in individual children (Bowers et al., 1994; Fletcher et al., 1994). Given the qualifications to our findings related to sample characteristics and instructional issues, it is apparent that we still have a great deal of work to do in sorting out the complex interactions among variables that influence reading out- comes for individual children. ACKNOWLEDGMENT This research was supported by Grant HD23340 from the National Institute of Child Health and Human Development. ‘We thank the children, teachers, and principals of the Leon County public schools who participated in this project. REFERENCES ‘Adams, M. J. (1990). Beginning to read: Thinking and learning about print. Cambridge, MA: MIT Press Baddeley, A. D. (1986). Working memory. New York: Oxford University Press. 184 TORGESEN ET AL, Badian, N. A. (1993). Phonemic awareness, naming, visual symbol processing, and reading. Reading and Writing: An Interdisciplinary Journal, 5, 87-100. Barker, T. A., Torgesen, J. K., & Wagner, R.K. (1992) The role of orthographic processing skills on five different reading tasks. Reading Research Quarterly, 27, 334-345 Barron, R. (1986). Word recognition in early reading: A review of the direct and indirect access hypothesis. Cognition, 24, 93-119 Blachman, B. A, (1984). Relationship of rapid naming ability and language analysis skills to kindergar- ten and first-grade reading achievement. Journal of Educational Psychology, 76, 610-622. Bowers, P. G. (1995). Tracing symbol naming speed’s unique contributions to reading disabilities over time, Reading and Writing: An Interdisciplinary Journal, 7, 1-28. Bowers, P. G., Golden, J., Kennedy, A., & Young, A. (1994). Limits upon orthographic knowledge due to processes indexed by naming speed. In V. W. Beminger (Ed), The varieties of orthographic knowledge I: Theoretical and developmental issues (pp. 173-218). Dordrecht, The Netherlands: Kluwer. Bowers, P. G., & Wolf, M. (1993a, March). A double-deficit hypothesis for developmental reading disorders. Paper presented to meetings of the Society for Research in Child Development, New Orleans, LA. Bowers, P. G., & Wolf, M. (1993b). Theoretical links between naming speed, precise timing mecha- nisms and orthographic skill in dyslexia, Reading and Writing: An Interdisciplinary Journal, 5, 69-85, Brady, S., & Shankweiler, D. (1991). Phonological processes in literacy: A tribute to Isabelle Y. Liberman. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. ‘Comwall, A. (1992). The relationship of phonological awareness, rapid naming, and verbal memory to severe reading and spelling disability. Journal of Learning Disabilities, 25, 532-538. Cunningham, A.E., & Stanovich, K.. (1991). Tracking the unique effects of print exposure in children: Associations with vocabulary, general knowledge, and spelling, Journal of Educational Psychology, 83, 264-274. Ehri, L. C. (1992). Reconceptualizing the development of sight word reading and its relationship to recoding. In P, B. Gough, L. C. Ebri, & R. Trieman (Eds.), Reading acquisition (pp. 107-143). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Ehri, L. C. (1996). Grapheme-phoneme knowledge is essential for learning to read words in English. In J. Metsala & L, Ehsi (Eds.), Word recognition in beginning literacy. Manuscript submitted for publication. Felton, R., & Brown, I. S. (1990). Phonological processes as predictors of specific reading skills in children at risk for reading failure. Reading and Writing: An Interdisciplinary Journal, 2, 39-59. Fletcher, J. M., Shaywitz, S.E., Shankweiler, D. P., Katz, L., Liberman, I. Y., Stuebing, K. K., Francis, D..,Fowler, A. B., & Shaywitz, B, A. (1994). Cognitive profiles of reading disability: Comparisons of discrepancy and low achievement definitions. Journal of Educational Psychology, 86, 6-23. Kail, R. V., & Hall, L. K. (1994). Processing speed, naming speed, and reading. Developmental Psychology, 30, 949-954. Meyer, M. M., Wood, F. B., Hunt, L. A., & Felton, R. H. (1995). Predictive value of rapid automatized ‘naming for later reading. Unpublished manuscript, Bowman Gray Schoo! of Medicine, Winston- Salem, NC. Sattler, J. M. (1992). Assessment of children: Revised and updated third edition. San Diego: Sattler. Share, D. L. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition. Cognition, 55, 151-218. ‘Share, D. L., & Jorm, A. F. (1987). Segmental analysis: Co-requisite to reading, vital for self-teaching, requiring phonological memory. European Bulletin of Cognitive Psychology, 7, 509-513. GROWTH OF WORD READING SKILLS = 185, Share, D. L., & Stanovich, K. B. (1995). Cognitive processes in early reading development: A model of acquisition and individual differences. Issues in Education: Contributions from Educational Psychology, 1, 1-35. Stanovich, K. E., & West, R. F. (1989). Exposure to print and orthographic processing. Reading Research Quarterty, 24, 402-433, Stone, B., & Brady, S. (1995). Evidence for phonological processing deficits in lese-skilled readers. Annals of Dyslexia, 45, 51-78. ‘Thorndike, R. L., Hagen, B. P., & Sattler, J. M. (1986). Guide for administering and scoring the Stanford-Binet Intelligence Scale (4th ed.). Chicago: Riverside. Torgesen, J. K., & Burgess, S. R. (1996). Consistency of reading-related phonological processes throughout early childhood: Evidence from longitudinal-correlational and instructional studies. In J. Metsala & L. Ehri (Eds.), Word recognition in beginning literacy, Manuscript submitted for publication. Wagner, R. K., & Barker, T. A. (1994). The development of orthographic processing ability. In V. W. Beringer (Ed.), The varieties of orthographic knowledge I: Theoretical and developmental issues (pp. 173-218). Dordrecht, The Netherlands: Kluwer. Wagner, R. K., Torgesen, J. K., Laughon, P., Simmons, K., & Rashotte, C. (1993). The development ‘of young readers’ phonological processing abilities. Journal of Educational Psychology, 85, 83-103. Wagner, R. K., Torgesen, J. K., & Rashotte, C. A. (1994). The development of reading-related phonological processing abilities: New evidence of bi-directional causality from a latent variable longitudinal study. Developmental Psychology, 30, 73-87. Wagner, R. K., Torgesen, J. K., Rashotte, C. A., Hecht, S. A., Barker, T. A., Burgess, S. R., Donahue, 4,, & Garon, T. (in press). Changing causal relations between phonological processing abilities and word-level reading as children develop from beginning to fluent readers: A five-year longitudinal study. Developmental Psychology. ‘Wimmer, H. (1993). Characteristics of developmental dyslexia in a regular writing system. Applied Psycholinguistics, 14, 1-33. Wolf, M. (1991). Naming speed and reading: The contribution of the cognitive neurosciences. Reading Research Quarterly, 26, 123-141. Wolf, M., Bally, H., & Morris, R. (1986), Automaticity, retrieval processes, and reading: A longitudinal study in average and impaired readers. Child Development, 57, 988-1000. Wolf, M., Pfeil, C., Lotz, R., & Biddle, K. (1994). Towards a more universal understanding of the developmental dyslexias: The contribution of orthographic factors. In V. W. Berninger (Ed.), The varieties of orthographic knowledge I: Theoretical and developmental issues (pp. 173-218), Dordrecht, The Netherlands: Kluwer, Wolff, P., Michel, G., & Ovrut, M. (1990). Rate variables and automatized naming in developmental dyslexia. Brain and Language, 39, 556-575. Woodcock, R. W. (1987). Woodcock Reading Mastery Tests-Revised. Circle Pines, MN: American Guidance Service. Manuscript received February 29, 1996 Final Revision received September 10, 1996 Accepted October 8, 1996 Copyright © 2002 EBSCO Publishing

You might also like