Longitudinal course of rapid naming in disabled and nondisabled readers
Meyer, Marianne S
The Rapid Automatized Naming Test (Denckla and Rudel 1974) was studied cross-sectionally in an sample of kindergartners (n = 342) atrisk for reading disability (Study 1), and longitudinally in an n = 160 epidemiological normal sample of children tested in first, third, fifth, and eighth grades (Study 2). Study 1 showed faster absolute naming speeds for those with near perfect untimed alphabet recitation, but the stronger and more orderly relation (at r = .31, p
to reading level and vocabulary across grades. Norms for the Rapid Automatized Naming Test based on the epidemiological normal sample tested in Grades 1, 3, 5, and 8 are presented in the appendix.
In 1974 Denckla and Rudel, noting the relationship between acquired alexia and childhood dyslexia, developed the Rapid Automatized Naming Test (RAN), which required rapid sequential naming of printed colors, numbers, objects, and letters. Two years later, they reported that deficits in rapid sequential naming differentiated children with dyslexia from learning disabled students without dyslexia and from normal controls (Denckla and Rudel 1976). At about the same time, LaBerge and Samuels (1974) proposed an information processing model that posited that good reading requires not only accurate, but also automatic, retrieval so that the reader’s attention can be focused on meaning and content. Both sets of investigators considered automatic, rapid, and accurate retrieval an important component of skilled reading. These researchers laid the groundwork for much of the current research on rapid naming deficits which are increasingly considered to be among the underlying cognitive correlates of dyslexia.
Although there have been numerous subsequent studies on rapid naming and its role in reading acquisition, most of these studies have been either cross sectional or have followed children for only a few years. Moreover, studies have varied in their choice or preference within the four stimulus categories of the original RAN (Wolf, Bally, and Morris 1986; Lovett 1987; Walsh, Price, and Cunningham 1988; Bowers, Steffy, and Tate 1988; Badian, McAnulty, Duffy, and Als 1990; Cornwall 1992; Badian 1993; Ackerman and Dykman 1993; Torgesen, Wagner, and Rashotte 1994; Fawcett and Nicholson 1994; and Korhonen 1995). To date, no study has specifically addressed the longitudinal course of rapid serial naming over the elementary school years, nor have norms been collected for large groups of dyslexic versus nondisabled readers over an extended period of time.
Beginning with Denckla and Rudel (1976), many studies have shown that a slowed rate of rapid naming is a strong correlate of dyslexia. Findings differ on the relative sensitivity of the four rapid naming stimuli. Lovett (1987) found in a group of primarily nine to eleven year olds that naming speed deficits for all RAN stimuli differentiated fluent normal readers from readers weak in both rate and/or accuracy. When Lovett looked separately at readers who were slow and those who were inaccurate, both groups exhibited a comparable degree of disadvantage on colors, objects, and numbers, but not on letters. Based on these findings, Lovett suggested that accuracy disabled readers have a selective problem with accessing letter names in addition to a general deficit in naming speed for visual stimuli. Our recent findings (Meyer, Wood, Hart, and Felton 1998), based on color/object and number/letter composites, show that both graphological (letters and numbers) and nongraphological (colors and objects) naming speeds are equally sensitive to, and predictive of, sight word identification in third through eighth graders. Others also show this generality (Cornwall 1992; Fawcett and Nicholson 1994; Korhonen 1995). Other investigators find graphological stimuli to be more sensitive measures. For example, Wolf, Bally, and Morris (1986) found that kindergarten rate of response, for both graphological and nongraphological stimuli, predicts impaired readers from average readers in Grade 2. However, by second grade, only the speed of naming of letters and numbers-the graphological symbols-predicted Grade 2 reading level. Consistent with the Wolf et al. findings are those of Bowers, Steffy, and Tate (1988) who looked only at colors and numbers. Bowers et al. (1988) found that number naming, but not color naming, contributed significantly to the variance in sight word identification (p
The present study documents the longitudinal course of rapid naming in kindergarten, elementary, and middle-school students. Data were obtained from two different groups within the same urban school system, identified at different stages of their educational careers. The first group included kindergartners rated by teachers as average, below average, and very poor in terms of expected future reading acquisition. (Those rated above average or superior were excluded.) The second group included randomly selected, normally distributed students (an appropriate random portion of whom were poor readers) ascertained in first grade and followed longitudinally through eighth grade.
The principal dependent measures for all of the above sample were the RAN tests developed by Denckla and Rudel (1974). On the RAN, children are required to name, as rapidly as possible, items presented visually on the four different charts. Each individual chart contains 50 items consisting of five different items presented in horizontal rows of ten items per row and randomly repeated. Stimulus sets are as follows:
(a) colors-red, yellow, green, blue, and black;
(b) single digit numbers-2, 4, 6, 7, and 9;
(c) line drawings of objects-key, scissors, umbrella, watch, and comb; and
(d) high-frequency lower case letters-o, a, s, d, and p.
Each test is scored for the seconds it takes to name each of the items on an individual chart in left-to-right and top-to-bottom sequence.
To introduce the task, children (or adolescents or adults) are told that we want to see how fast they can name some things with which they are familiar. The color chart is presented first, and the examiner points to each color on the chart and asks, “What color is this?” to determine if the child can identify the colors correctly. If the child can identify them correctly, the examiner then says, “Now I want you to start here (point) at the top of this chart and name the colors in this row (point), then the next row (point) all the way to the bottom as fast as you can without making mistakes and without skipping any.” This same procedure is followed for the three subsequent charts (numbers, objects, letters). In each case, the examiner makes certain, before starting the timer, that the child can identify the stimuli accurately. Note that in the case of the object chart, the subject can choose to identify the watch as either a watch or a clock, but must consistently use that word throughout or it is counted as an error. The time it takes to complete the chart is recorded and the child’s responses, including any uncorrected errors, are noted. Although errors are not corrected, the child is allowed to correct his or her own errors spontaneously. The time it takes for spontaneous selfcorrection is counted in the total time. At most, the total task requires 10 minutes, but by mid-elementary school, children often take less than 5 minutes.
To facilitate analysis, RAN raw scores were combined into a color and object composite (COLOBJ) and a letter and number composite (NUMLET). The composite letter/number and color/object scores were calculated as the mean of their respective component scores. We and others (Denkla and Rudel 1974; Wolf, Bally, and Morris 1986; Badian et al. 1991; and Korhonen 1995) have shown that numbers and letters require almost equal time to complete at all age levels. Colors and objects require a longer time to complete, with objects typically requiring approximately a third again as much time as colors. These two composites were chosen because numbers and letters are both printed symbols, whereas objects and colors are not. Therefore, number/letter scores are, on their face, more likely to reflect the impact of early learning to read, including alphabet mastery, whereas color/object scores are less obviously related to prior alphabet or reading mastery itself. Color/object composite scores might then be seen as “purer” measures of naming speed whereas number/letter composite scores might reflect more of a combination of “pure” naming speed and fluency acquired by means of exposure to alphabet or print.
Consistent with the above analysis, an informative way to represent the distinctiveness of the two composites was to report the color/object composite (COLOBJ) as one major dependent variable, representing naming speed that is uncontaminated by print stimuli; and a ratio score (RATIO) that expresses the extent to which the number/letter composite exceeds the color/object composite. This ratio score thus represents the relative number/letter advantage. The second variable isolates and represents the particularly lexical, print-related component distinct from the general naming speed component. RATIO is calculated as the quotient of number/letter (NUMLET) speed minus color/object (COLOBJ) speed over color/object (COLOBJ) speed. Speed is defined as the number of items named per second. Because there are 50 items, speed is 50 divided by the total number of items named. Speed is different from time which is simply the number of seconds required to name all 50 items.l The statistical alternative to this RATIO construct is to take NUMLET speed as a dependent measure with COLOBJ speed partialled out by covariance.
Since the raw rapid naming scores are measured as seconds to complete the page of stimuli, and since such measures are naturally positively skewed, we used the reciprocal of the time measure in seconds, thereby converting it to a speed measure.
STUDY 1-KINDERGARTEN SAMPLE PARTICIPANTS
Participants for Study 1 were drawn from all the kindergarten classes in the eight schools in our local public school system. During their kindergarten year, all children had been exposed to the Writing to Read program developed by IBM, but had not been taught using basal readers. Letter identification and letter recognition, alphabet recitation, and sound-symbol correspondence had also been taught using the Houghton-Mifflin kindergarten curriculum and Alphatime. In the spring of the year, all students were ranked by their teachers for their ability to master basic reading skills, using categories of superior, above-average, average, below average, or very poor. No criterion, other than expected reading acquisition and mastery was given to the teachers. Because these kindergartners were initially tested as part of an intervention study on teaching methods for at-risk readers, those who were rated as superior or above average in potential for reading success, and those who scored below a standard score of 80 on the recently administered Otis-Lennon Mental Ability Test (Otis and Lennon 1968), were not included in the study Of the 469 children remaining in the sample after these exclusions, permission was obtained to evaluate 365 children; 355 were subsequently tested on an individually administered battery of reading readiness tests which included the Rapid Naming Test. Thirteen of these children could not recognize lower case letters or single digit numbers on the RAN, and were, therefore, not included in the analysis. The remaining 342 children, whose average age was 6.13 (SD 0.39) and who were rated as average, below average, or poor in reading potential by their teachers, had a mean Otis Lennon IQ of 107.11 (SD 12.34). Gender and race were reflective of the public school population from which this sample was drawn (53 percent female; 24 percent minority). We do not have data about the subsequent achievement scores of all of these 342 students.
For this total group of 342 kindergartners, rated by their teachers as average or below average in reading potential, the time to name letters/numbers was already faster than the time to name color/objects, although the variances did overlap considerably. (Number/letters mean = 63.0, SD = 20.0; colors/objects mean = 76.2, SD = 18.0.) Therefore, a number/letter advantage already exists in the second half of the kindergarten year. To test the hypothesis that a number/letter advantage will emerge as alphabet mastery is achieved, we compared the number/letter advantage to concurrent level of alphabet mastery. Alphabet recitation, measured here by having children orally recite the alphabet from memory without reference to print, is one of the most elemental and commonly rehearsed of all kindergarten skills. Note that it is an untimed measure and relies on knowledge, not speed. However, like the RAN, alphabet recitation relies on automaticity of response when fully achieved, despite the fact that it is an untimed measure. It is, therefore, the underlying automatic nature of both tasks which unites them.
Those children whose alphabet knowledge was so minimal that they did not recognize the RAN stimuli had already been excluded, but within the remaining sample of 342, a considerable range of alphabet recitation ability was found. Children were divided into three categories (limited, moderate, perfect) based on the number of letters in the alphabet that they could recite. Although the vast majority recited most of the alphabet in sequence, correct order was not a requirement of the task.
Orderly changes in rapid naming performance were found across the three alphabet recitation groups (see figure 1). Specifically, the poorest “limited” recitation group (n = 24) took almost exactly the same time to name colors/objects and numbers/letters. The middle or “moderate” group (n = 56) took 10 seconds less to name numbers/letters compared to colors/objects, although color/object naming was no faster than in the poorest group. The “perfect” group, who could recite the alphabet almost flawlessly, showed a relative time advantage of slightly more than 14 seconds in favor of letter/number naming. Their times to name both classes of stimuli was also significantly faster than the times of the first two groups. In terms of the major dependent measures described in the Methods, RATIO (the relative number/letter advantage) showed an orderly increase across the three alphabet recitation groups, with the extreme groups differing significantly at p
In brief, poor alphabet reciters (limited group) showed no number/letter with RATIO scores near zero. Children with better alphabet recitation performance (moderate and perfect groups) showed a letter/number advantage, and this advantage was itself proportional to their alphabet recitation ability Even in kindergarten, then, the number/letter composite may be reflective of prior achievement and exposure to alphabet, if not exposure to print (Cunningham and Stanovich 1990). Alternatively, there could be chronic individual deficiencies in the ability to automatize, despite equal amounts of exposure (Bowers 1993).
At the same time, another effect is also apparent: particularly good alphabet recitation (again, operationalized as accuracy, not speed) is associated with color/object naming speed in the absolute. On its face, this relationship is less obviously related to alphabet mastery; it could reflect a more long-standing, perhaps congenital, ability. At minimum, the data seem to suggest two different mechanisms: absolute naming speed on the one hand, and relative number/letter advantage on the other.
Let us now examine the longitudinal “fate” of these two components over Grades 1 through 8 in an epidemiologically derived normal sample.
STUDY 2-LONGITUDINAL SAMPLE PARTICIPANTS
A longitudinal sample of 160 students was derived from a larger, normally distributed, random sample of 485 students who were accessed in first grade. Since they were recruited in the same year as the sample in Study 1, they were then one year older and one year farther along than the previous Study 1 sample. The n = 160 subset, comprising the present sample, was tested in first, third, fifth, and eighth grades, and is virtually identical to the longitudinal sample in Meyer et al. (1998). Students were excluded who could not be tested in all four grades, but sampling was still stratified so as to ensure a normal distribution of reading ability in first grade (and, as it turned out, in the other grades as well). Students were also excluded if their Peabody Picture Vocabulary (PPVT-R) score was below 70 or above 130. The mean Woodcock Johnson reading standard score in first grade was 107.20 (SD 14.73); by eighth grade it was 102.66 (SD 12.48). The first grade PPVT-R score was 103.84 (SD 15.22). The demographics of this sample were appropriately representative of their public school system: males and females are equally represented, and race was 73 percent Caucasian and 27 percent African-American. All students were English-speaking.
Reading level was determined by measuring single-word reading which is usually considered the core defining weakness in reading disabled students (Stanovich 1986; Adams 1990; Olson et al. 1994; Beck and Juel 1995; Lyon 1995). Two measures of single-word reading in third grade were used to establish the three groups of readers: (1) Woodcock-Johnson Psycho-Educational Battery-Reading Cluster, (WJWID) (Woodcock and Johnson 1977); and (2) Decoding Skills Test-Part II: Phonetic Patterns, Real Words (DSTREAL) (Richardson and DiBenedetto 1985). The WJWID is norm referenced and requires identifying letters and both phonetically regular and irregular sight words within a fivesecond time limit. The DSTREAL is a criterion-referenced test, where participants read up to 60 phonetically regular words30 monosyllabic and 30 polysyllabic-chosen to represent common orthographic patterns. Normal-reading fifth graders get all of them right. Reading ability was defined on the basis of third grade rather than first grade tests. This allowed us to avoid first grade floor effects, particularly on WJWID.
The poorest reading group consisted of those scoring in the bottom 16 percent of the sample on both tests; this designation required raw scores of less than 29 on the WJWID and less than 38 on the DSTREAL. This converging requirement yielded 17 students at the lowest reading level which comprised 10.6 percent of our total sample at the time. A similar requirement that good readers score in the top 16 percent of the sample on both tests identified 13 top readers, comprising 8 percent of the total group. The remaining cases (who scored above the sixteenth percentile on at least one test and below the eighty-fourth percentile on at least one test) comprised the middle-level group with an n of 130.
In order to obtain a fuller composite or overall measure of reading skill, we administered the other two subtests from the Woodcock Johnson: Word Attack and Passage Comprehension. These, together with Word Identification, yield an age-referenced standard score (Woodcock and Johnson 1977, p. 34). The Word Attack subtest requires untimed reading of a list of words defined as “letter combinations that are not actual words or are extremely low frequency words in the English Language.” The Passage Comprehension subtest uses a cloze procedure that requires reading sentences that are missing a word and supplying a word that makes sense in the sentence or passage.
Finally, certain brief estimates of intelligence were taken. In first grade, we administered both the Peabody Picture Vocabulary Test (Dunn and Dunn 1981), a measure of receptive vocabulary knowledge, and Raven’s Colored Progressive Matrices (Raven, Court, and Raven 1984), a nonverbal measure of reasoning by analogy. We administered the Kaufman Brief Intelligence Test (Kaufman and Kaufman 1990) to the eighth graders. According to its authors, the K-BIT’s two subtests measure crystallized thinking (assessed by a two-part vocabulary task requiring word knowledge and verbal concept formation) and fluid thinking (assessed by a matrices which requires nonverbal reasoning to solve new problems). RESULTS
Table I shows rapid naming, reading, and vocabulary scores for the three reading level subgroups, and figure 2 shows the growth curves of rapid naming for the three reading level subgroups.
The growth curves show an obvious floor effect as naming speeds approach an asymptote by about eighth grade. Many of the subsequent observations noted below can be explained, at least in part, by this floor effect. First, there is a steady improvement of rate of responding over time, with the greatest improvement in rate occurring between first grade and third grade for all groups, marginally more so for the poor reading group than for the other groups. Second, with increasing age, there is a narrowing of the gap between the groups as they approach the eighth grade asymptote. Third, the number/letter advantage (RATIO) remains similar across Grades 1 through 8. Finally, naming speed is itself related to reading level, with the greatest difference between the normal and poor reading groups. Again, perhaps for reasons of a floor effect, the relationship is stronger in the early grades.
These fairly straightforward observations on the growth curves aside, the question then turns to the dissociation observed in Study 1 between absolute color/object naming speed (COLOBJ) and the relative number/letter advantage (RATIO) expressed as the quotient of the difference of NUMLET minus COLOBJ over COLOBJ. As the longitudinal curves in figure 2 show, naming speeds improve considerably over the eight grades, but RATIO does not. In the context of the findings of Study 1, showing the orderly relationship of RATIO to alphabet recitation knowledge, the implication of these growth curves is that much of the number/letter advantage expressed in the RATIO measure is related to early processes in kindergarten and first grade, thereafter to show little change over time despite the considerable improvement in absolute color/object and number/letter naming speeds. The differential longitudinal course of these absolute and ratio measures reinforces the impression that they are separate mechanisms.
If these are two separate mechanisms, then it would be important to test whether or not they have separate relationships to reading achievement longitudinally. The present longitudinal data permitted that test to be formally conducted, to assess the extent to which variance in reading achievement was separately related to these two mechanisms. In consideration of the large sample size and the multiple analyses involved, alpha was set at p = .01. Correlations detailing the prediction of reading and brief IQ measures, from rapid naming scores at each grade level, are shown in table II. This table presents the absolute color/object naming scores and the absolute number/letter naming scores, along with the RATIO that expresses the degree of number/letter advantage, as defined above. Table II also shows the partial correlations involving NUMLET, where COLOBJ is partialled out by covariance (the outcome of a Type III sums of squares multiple regression analysis).
Table II clearly shows that the number/letter composite (NUMLET) is usually the strongest predictor of reading achievement. The color/object composite (COLOBJ) is almost as strong a predictor, however, so that the difference between them, i.e., the relative number/letter advantage (RATIO), is always weak and sometimes negative. RATIOs at any grade never account for as much as 4 percent of the variance and are never significant predictors at p
Of particular interest is the fact that the first and eighth grade verbal and nonverbal IQ estimates are dissociated in their relations to COLOBJ, NUMLET, and RATIO. The nonverbal tests-Raven’s in first grade and K-BIT Matrices in eighth grade-are never significantly related to any naming variable. Moreover, eighth grade K-BIT Verbal IQ is usually marginally better predicted by COLOBJ at any grade than by NUMLET, and not at all by the unique NUMLET variance. First grade picture vocabulary, PPVT-R, is related to COLOBJ at each grade, but to NUMLET only concurrently at Grade 1. Within the limits of the content distinctions among the IQ estimates, therefore, COLOBJ as well as NUMLET are best interpreted as verbal, rather than as nonverbal or fluid skills.
Of interest also are the long-term predictive relations. COLOBJ in first grade is a somewhat better longitudinal predictor of reading than is COLOBJ at other grade levels, and COLOBJ in first grade predicts reading at all grade levels equally strongly. First grade NUMLET, on the other hand, is a particularly strong concurrent predictor of first grade reading, but its strength of prediction declines somewhat for subsequent years. The remarkable stability of the prediction by COLOBJ, selective to verbal skills, and its ability to predict subsequent reading as well as concurrent reading in first grade, may suggest a lesser role for experience and a greater role for a continuing, perhaps congenital, trait.
In brief, Study 2 shows the maintenance of the two separate mechanisms first identified in the kindergarten sample in Study 1: COLOBJ, or absolute color/object naming speed, improves to asymptote over the eight grades of school whereas RATIO, the relative advantage for number/letter speed over color/object speed, remains relatively constant across the eight grades of school. COLOBJ, especially in first grade, is by itself a strong predictor of subsequent reading performance and is a correlate and predictor of vocabulary or estimated verbal IQ. RATIO is not a strong correlate or predictor of reading.
Taken together, Studies 1 and 2 suggest conclusions in three broad areas: kindergarten experience, development in Grades 1 through 8, and intervention implications.
First, it is a safe assumption that there is some period, however brief and early in kindergarten or even prekindergarten, when color/object and number/letter naming speeds (COLOBJ and NUMLET) are equally fast (or slow). However, the relative speed advantage of numbers/letters over colors/objects (RATIO) develops rapidly in kindergarten, is strong in first grade, and does not increase in magnitude thereafter. There are individual differences in the degree of this advantage, although almost all first graders have the advantage to some degree. It is difficult to avoid the conclusion, therefore, that the major determinants of the degree of the number/letter advantage tend to exert their effect in kindergarten or early first grade. These determinants could include exposure to print or alphabet, and to that extent, they could be environmental. There could, of course, be a more chronic individual trait difference in the ability to automatize new information, whether in alphabet recitation or in number/letter naming. However, since it a learned stimulus-response automaticity that is emerging at different levels of fluency, it is almost by definition an interactive and reciprocal process relating a prior predisposition to the current instructional situation. Both pre-existing traits and current environmental stimulation are almost certainly interacting to produce the kindergarten and first grade level of number/letter advantage. Indeed, the very speed of the emergence of the number/letter advantage, and its subsequent long term stability over eight grades of school, suggests a process that is amenable to environmental modification in that learning phase.
Second, in regard to the longitudinal sample, the general form of the growth curve for rapid naming rate holds for all students, regardless of reading ability, with the greatest growth in naming speed occurring between first and third grade, after the number/letter advantage has already been established. More important, the number/letter advantage itself has little further predictive or concurrent validity. This contrasts with the strong predictive effect of the absolute COLOBJ or NUMLET scores. The conclusion is inescapable that speed in general, not strongly dependent on the type of stimuli, is the important variable in the grade school years. That COLOBJ in first grade has predictive validity as strong as its concurrent validity suggests that something stable is measureable in first grade and is not much changed thereafter. In a separate paper (Meyer, Wood, Hart, and Felton 1998), we report that most of the predictive variance in rapid naming is within the group of poor readers, distinct from average or superior readers. This suggests that only some poor readers, especially those with the poorest prognosis, are constrained by their automaticity or fluency deficit. A chronic, possibly even congenital, trait affecting some initially poor readers is implied, and this trait becomes the background reality against which intervention efforts should be viewed.
Taken together, the above two considerations suggest that the implications for intervention are appropriately considered in the context of the reciprocal causation hypothesis (Stanovich 1986; Perfetti et al. 1987) which states that some skills both facilitate learning to read and result from reading itself. Cunningham and Stanovich (1990) invoke exposure to print as that mediating variable, and our Study 1 may represent a generalization of that concept to what could be described as “exposure to (hence mastery of) alphabet” in kindergarten. The Study 1 data do not rule out the possibility that interventions in kindergarten designed to increase exposure to print or alphabet could have an impact on reading related processes, in particular on the number/letter advantage. At the same time, it must be admitted that individual differences in the number/letter advantage, already well established by first grade, had little further impact on subsequent reading performance. If naturally occurring differences in kindergarten environment had a beneficial effect on the number/letter advantage, perhaps due to variations in exposure to print or alphabet, and if these persisted until first grade as the main component of the number/letter advantage, then the data from Study 2 suggest that this effect dissipates in subsequent grades with little long-term relation to reading achievement. Kindergarten interventions that would be successful in the long term might, therefore, have to exceed the best of the naturally occurring variations in kindergarten environments. The data suggest that kindergarten or first grade, not later grades, are the most productive times at which to intervene to enhance fluency.
One obvious next test of these implications would be to provide direct fluency training as early as kindergarten in an early effort to improve both absolute fluency and the number/letter advantage. Attempts to improve it would plausibly be made through consistent fluency training beginning in kindergarten with alphabet mastery and single-letter calling, and continuing in the school grades with a focus on fluent and automatic reading itself through sentence and passage repetition (such as the Repeated Readings method in Samuels 1987; Mathes, Simmons, and Davis 1992; Young, Bowers, and MacKinnon 1996; or the RAVE-0 method in Wolf and Obregon 1997). In the context of the reciprocal causation hypothesis, a continuous feedback mechanism might be imagined whereby even small gains in the number/letter advantage could result in improvements in reading and reading readiness, which, in turn, might cause further gains in fluency and so on for several grades of school. Bowers (1993) reports that repeated reading of a text causes more improvements in the passage reading fluency of children who are rapid namers than in that of slower namers even when initial text speed is controlled. This suggests not only differential rates of improvement among children but also the need to titrate the amount of fluency training provided (more for the less able). While the number/letter advantage has some relevance for the earliest years, as a possible marker of improvement, it is likely that more generalized improvements in fluency, across all categories of stimuli, would be the broader goal since speed itself across all classes of stimuli was the major correlate and predictor of reading in the grade school years.
This study, part of a larger research project on dyslexia, was funded by the National Institute of Child Health and Human Development, Research Grant HD 21887-10. The participation of the Winston-Salem/Forsyth County Schools and their students is acknowledged with gratitude.
1 It is appropriate to express the arithmetic difference between NUMLET and COLOBJ as a quotient over COLOBJ, since the size of the arithmetic difference is not independent of the absolute speed of either variable. See Harshman and Krashen 1972 for an early treatment of this question in the context of laterality indices. In the present case, because of the theoretical assumption that COLOBJ represents the purer measure of absolute speed, we use it alone in the denominator of the quotient instead of using the sum of COLOBJ and NUMLET.
Ackerman, P. T., and Dykman, R A. 1993. Phonological processes, confrontational naming, and immediate memory in dyslexia. Journal of Learning Disabilities 26(9):597-609.
Adams, M. J. 1990. Beginning to Read. Cambridge, MA: MIT Press. Badian, N. A. 1993. Phonemic awareness, naming, visual symbol processing, and read
ing. Reading and Writing: An Interdisciplinary Journal 5:87-100. Badian, N. A. 1995. Predicting reading ability over the long term: The changing roles of letter naming, phonological awareness and orthographic processing. Annals of Dyslexia 45:79-96.
Badian, N. A., Duffy, F. H., Als, H., and McAnulty, G. B. 1991. Linguistic profiles of dyslexics and good readers. Annals of Dyslexia 41:221-45. Badian, N. A., McAnulty, G. B., Duffy, F .H., and Als, H. 1990. Prediction of dyslexia in
kindergarten boys. Annals of Dyslexia 40:152-69.
Beck, I. L., and Juel, C. 1995. The role of decoding in learning to read. American Educator 19: 8-12.
Biddle, K. R. 1996. Timing deficits in impaired readers: an investigation of visual naming speed and verbal fluency. Ph.D. Diss., Tufts University, Boston. Blachman, B. 1984. Relationship of rapid naming ability and language analysis skills in kindergarten and first-grade reading achievement. Journal of Educational Psychology 76(4):610-22.
Bowers, P. G., Steffy, R., and Tate, E. 1988. Comparison of the effects of IQ control methods and naming speed predictors of reading disability. Reading Research Quarterly 23:304-9.
Bowers, P. G. 1993. Text reading rereading: Predictors of fluency beyond word recogition. Journal of Reading Behavior 25:133-53.
Bowers, P. G., and Wolf, M. 1993. Theoretical links among naming speed, precise timing mechanisms and orthographic skill in dyslexia. Reading and Writing: An Interdisciplinary Journal 5:69-85.
Bowers, P. G., and Swanson, L. B. 1991. Naming speed deficits in reading disability: Multiple measures of a singular process. Journal of Experimental Child Psychology 51:19-219.
Cornwall, A. 1992. The relationship of phonological awareness, rapid naming, and verbal memory to severe reading and spelling disability. Journal of Learning Disabilities 25:532-38.
Cunningham, A. E., and Stanovich, K. E. 1990. Assessing print exposure and orthographic processing skill in children: A quick measure of reading experience. Journal of Educational Psychology 82(4):733-40.
Denckla, M. B., and Rudel, R. G. 1974. Rapid automatized naming of pictured objects, colors, letters and numbers by normal children. Cortex 10:86-202. Denckla, M., and Rudel, R. G. 1976. Rapid automatized naming (RAN): Dyslexia differ
entiated from other learning disabilities. Neuropsychologia 14:471-79. Dunn, L. M., and Dunn, L. M. 1981. Peabody Picture Vocabulary Test. Circle Pines, MN: American Guidance Service.
Fawcett, A., and Nicholson, R. 1994. Naming speed in children with dyslexia. Journal of
Learning Disabilities 27(10):641-46.
Felton, R. H., and Brown, 1. 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.
Felton, R. H., Naylor, C. E., and Wood, F. B. 1990. Neuropsychological profile of adult dyslexics. Brain and Language 39:485-497.
Felton, R.H. 1992. Early identification of children at risk for reading disabilities. Topics in
Exceptional Children and Special Education 12(2):212-29. Flowers, D. L. 1993. Biological basis for reading disability: A review of work in progress. Journal of Learning Disabilities 26:575-82. Lovett, M. W. 1987. A developmental approach to reading disability: Accuracy and
speed criteria of normal and deficient reading skill. Child Development 58: 234-60. Lyon, G. R., 1995. Toward a definition of dyslexia. Annals of Dyslexia 45:3-30. Kaufman, A., and Kaurman, N. 1990. Kaufman Brief Intelligence Test. Circle Pines, MN: American Guidance Service.
Korhonen, T. 1995. The persistence of rapid naming problems in children with reading disability: A nine year follow-up. Journal of Learning Disabilities 28(4):232-39. LaBerge, D., and Samuels, J. 1974. Toward a theory of automatic information processing in reading. Cognitive Psychology 6:293-323.
Mathes, P. G., Simmons, D. C., and Davis, B. I. 1992. Assisted reading techniques for developing reading fluency. Reading, Research, and Instruction 31(4):70-77. Meyer, M. S., Wood, F. B., Hart, L., and Felton, R. H. 1998. Selective predictive value of rapid automatized naming within poor readers. Journal of Learning Disabilities 31(2):10S17.
Olson, R., Forsberg, H., Wise, B., and Rack, J. 1994. Measurement of word recognition orthographic and phonological skills. In G.R. Lyon, ed., Frames of Reference for the Assessment of Learning Disabilities: New Views on Assessment Issues. Baltimore: Paul Brooks Publishing: 243-77.
Otis, A.S., and Lennon, R. T. 1968. Otis-Lennon Mental Ability Test. New York: Harcourt
Perfetti, C., Beck, I., Bell, L., and Hughes, C. 1987. Merrill-Palmer Quarterly 33(3):283-319. Porter, R., and Hughes, L. 1983. Dichotic Listening to CV’s: Method, Interpretation, and Application. In J. Hellige, ed., Cerebral Hemispheric Asymmetry: Method, Theory, and Application. New York: Praeger.
Raven, J., Court, J., and Raven, J. 1984 Coloured Progressive Matrices. London: H.K Lewis & Co. Ltd.
Richardson, E., and DiBenedetto, B. 1985. Decoding Skills Test. Los Angeles, CA: Western Psychological Services.
Samuels, S. J. 1987. Information processing abilities and reading. Journal of Learning Disabilities 20(1):18-22.
Shankweiler, D., and Liberman, I. Y. 1989. Phonology and Reading Disability. Ann Arbor, MI: The University of Michigan Press.
Snyder, L. S., and Downey, D. M. 1995. The serial rapid naming skills of reading disabled children. Annals of Dyslexia 45:31-50.
Stanovich, K. E. 1986. Matthew effects in reading: Some consequences of individual dif
ferences in the acquisition of literacy. Reading Research Quarterly 21:360-407. Stanovich, K. E. 1988. Explaining the differences between the dyslexic and the gardenvariety poor reader: The phonological-core variable-difference model. Journal of Learning Disabilities 21:590-612.
Torgesen, J. K., Wagner, R. K., and Rashotte, C. A. 1994. Longitudinal studies of phonological processing and reading. Journal of Learning Disabilities 27:276-86. Torgesen, J. K., Wagner, R. K., Simmons, K., and Laughon, P. 1990 Identifying phonological coding problems in disabled readers: Naming, counting and span measures? Learning Disabilities Quarterly 13: 236-45.
Wagner, R. K., and Torgesen, J. K. 1987. The nature of phonological processing and its casual role in the acquisition of reading skills. Psychological Bulletin 101:192-212. Walsh, D. J., Price, G., and Gillingham, M. G. 1988. The critical but transitory importance of letter naming. Reading Research Quarterly 23:108-22. Wolf, M., Bally, H., and Morris, R. 1986. Automaticity, retrieval processes, and reading: A longitudinal study in average and impaired readers. Child Development 57:988-1000.
Wolf, M. 1990. The word retrieval deficit hypothesis and developmental dyslexia. Presented at American Educational Association Meetings, Boston, Mass. Wolf, M. 1991. Naming speed and reading: The contribution of the cognitive neurosciences. Reading Research Quarterly 26:123-40.
Wolf, M., and Bowers, P. 1997. The “double-deficit hypothesis” for the developmental dyslexias. Unpublished paper, submitted for publication. Wolf, M., and Obregon, M. 1997. The “double-deficit” hypothesis: Implications for diagnosis and practice in reading disabilities. In L. Putman, ed., Readings on Language and Literacy. Cambridge, MA: Brookline Books, 177-210. Woodcock, R. W., and Johnson, M.B. 1977. Woodcock-Johnson Psycho-educational Battery. Allen, TX: DLM.
Young, A., Bowers, P., and MacKinnon, G. 1996. Effects of prosodic modeling and repeated reading on poor readers’ fluency and comprehension. Applied Psycholinguistics 17: 59-84.
Wake Forest University School of Medicine
Correspondence may be sent to: M. S. Meyer, Section of Neuropsychology, Wake Forest University School of Medicine, Medical Center Blvd., WinstonSalem, N.C. 27157-1043. e-mail: mmeyerEwfubmc.edu
Copyright International Dyslexia Association 1998
Provided by ProQuest Information and Learning Company. All rights Reserved