Yates, C., Berninger, V. and Abbott, R.
Prufrock Press, Journal for the Education of the Gifted
Vol. 18, No. 2
This article discusses writing disabilities in detail. Specific studies are explained as they relate to comparisons between gifted students and average students in numerous areas of writing skills. Authored by Cheryl Yates, Virginia Berninger and Robert Abbott from the University of Washington.
Abstract: To determine whether some gifted children may have specific writing disabilities, we compared 10 gifted and 10 average children in grades 1 through 6 (N = 120). The primary question addressed was whether gifted writers exhibit more advanced skills than their average-IQ peers in both higher level cognitive writing processes (as manifested in quality of text generation) and lower level writing processes (as manifested in transcription), or solely at one of those levels. Results supported our hypothesis that gifted and average children differ in higher level cognitive processes but not in lower level processes in writing. Poor writing performance in gifted students may reflect low-level transcription deficits and is not necessarily the result of laziness, boredom, or lack of motivation as is often believed. Implications of this research for identifying writing-related learning disabilities in gifted students are discussed.
Although considerable research exists on writing disabilities (for review, see Newcomer & Barenbaum, 1991), hardly any research has focused on writing disabilities in gifted children (Ganschow, 1985). Four factors may contribute to this lack of attention to the writing problems of gifted children. One factor is difficulty in accepting the apparent paradox that gifted children may have learning disabilities. Another factor is the widespread belief that gifted children should excel in all academic domains. Yet another factor is the assumption that underperforming gifted students suffer primarily from low motivation and simply need more stimulation, writing practice, and opportunity to exercise their creativity (Master, 1983). A final factor is lack of awareness that specific disabilities may occur in some, but not in all, component writing processes.
None of these factors is a valid reason for ignoring the possibility that gifted children may have specific writing disabilities. The cooccurrence of superior intelligence and learning disabilities in general has been well documented (e.g.. Fox, Brody, &. Tobin, 1983), as has the fact that gifted children are not necessarily equally gifted in all academic domains (Keating, 1991). Simply encouraging a child to keep practicing a skill will not necessarily eliminate a problem in that skill; in fact, it may diminish motivation. Diagnosis of the underlying deficit interfering with skill acquisition is sometimes necessary. Progress is being made in identifying processing deficits that may interfere with acquisition of component writing skills (e.g., Berninger, Yates, Cartwright, Rutberg, Remy, &. Abbott, 1992).
In elementary school children, these processing deficits are often low-level developmental skills, such as orthographic or phonological coding, fine-motor, or orthographic-motor integration skills, rather than high-level reasoning skills (e.g., Berninger, Yates, et al., 1992; Berninger, Cartwright, Yates, Swanson, &. Abbott, in press). Gifted children are unlikely to have deficits in high-level reasoning skills but may have deficits in these low-level developmental skills. Thus, we tested the hypothesis that gifted children would differ significantly from age-matched average-IQ children on the high-level processes involved in writing but would not differ significantly in the low-level processes in writing. We tested this hypothesis by administering a battery of writing and writing-related measures validated in prior research with 300 primary grade children (Abbott & Berninger, 1993; Berninger, Yates, et al., 1992) and 300 intermediate grade children (Abbott & Berninger, 1993; Berninger, Cartwright, et al., in press). The battery included tests of both low-level and high-level processes involved in writing.
Current theory about writing processes suggests why deficits in low-level skills could cause gifted children to develop specific writing disabilities. Translating (Hayes & Flower, 1980) consists of two separable processes in beginning and developing writers: text generation and transcription (Berninger, Yates, et al., 1992). Text generation is the translation of ideas into language representations in working memory; transcription is the translation of those representations into written symbols on the printed page. High-level verbal skills are more likely to be related to text generation, but low-level skills are more likely to be related to transcription. Because of their superior high-level verbal skills, gifted children are most likely to differ from average students in the text generation component of translating. For example, they should produce higher quality content and use more sophisticated vocabulary. However, they may not differ from average students in low-level skills related to transcription. For example, they might not produce more text within the same time interval. Those with deficits in low-level skills will be "at risk" for disability in transcription, experiencing difficulty in producing written language despite the high quality of what they have to communicate.
We also examined whether individual children had disabilities in specific writing or writing-related skills. Based on previous work (Berninger &. Hart, 1993; Berninger, Hart, Abbott, &. Karovsky, 1992), we predicted that definitions based on absolute criteria (low-functioning independent of IQ) would most likely identify writing disabilities in average children, whereas definitions based on relative criteria (discrepancy between achievement and IQ) would most likely identify writing disabilities in gifted children.
The practical significance of using relative versus absolute criteria is best understood in the context of three recent developments in the field of learning disabilities. First, the use of IQ in defining learning disability has been questioned in two special issues of the Journal of Learning Disabilities (e.g., Siegel, 1989; Siegel, 1992). The rationale for not defining learning disabilities in terms of underachievement relative to IQ is the bidirectional influence between achievement and IQ. Not only does IQ set a limit on achievement but also achievement affects performance on IQ tests. Moreover, many of the low-level skills that differentiate reading-disabled children from children with normal reading development (e.g., nonword reading or word attack skills) do not correlate with IQ. Siegel and others therefore recommend using absolute criteria for low performance on relevant skills rather than relative criteria based on discrepancy from IQ to identify learning disabilities. Berninger and Hart (1993) are reluctant to eliminate relative criteria completely because the processing problems of gifted children, who may struggle to achieve near grade level, may go undetected, when in fact their potential is considerably above grade level.
Second, researchers are increasingly using their own criteria for identifying learning disabilities rather than assuming that learning disabilities are found only in children identified by schools as learning disabled (e.g., Shaywitz, Shaywitz, Fletcher, &. Escobar, 1990). The rationale is that criteria for qualifying for special education services vary greatly from state to state and even from school to school within states (Berninger, Hart, et al., 1992). Shaywitz et al. (1990) found that when researcher-criteria are used for learning disabilities, equal numbers of girls and boys are identified; whereas when school-criteria are used, more boys are qualified for special education.
Third, considerable intraindividual variation in reading, writing, and reading- and writing-related skills has been found in normally developing children in unreferred samples. For example, about 10% of a primary grade sample showed extreme developmental dissociations or uneven rate of development of reading, writing, and reading-and writing-related skills. That is, they scored in the bottom 5% of the normal distribution on at least one skill and in the top 5% of the normal distribution on at least one skill. Such intraindividual variation is also likely in gifted children. The implication of these intraindividual variations is that a distinction needs to be made between learning disabilities, which many children may have in a discrete skill, and classification of children as learning disabled for special education services. Children who are classified as learning disabled probably have extreme deficits in many rather than a single or a few isolated skills, whereas children who are not classified as learning disabled might have a deficit in a single or a few isolated skills (i.e., a specific disability). Also, we cannot readily classify children as learning disabled versus not learning disabled as all children have profiles of relative strengths and weaknesses. Rather, we need to determine whether a child has a disability in a specific component reading or writing skill or process related to those component skills.
The sample consisted of 120 students--20 from each of the first through sixth grades--drawn from a larger study of 600 students (100 at each grade level, one through six) from 5 urban and suburban schools. None of these children was receiving special education services. Of these 120 subjects, five percent were left-handed. Boys and girls were equally represented. The sample was ethnically diverse, consisting of 8% Asian-American, 13% Black-American, 74% Caucasian, 1% Hispanic, and 4% Other. More than half (55%) of the mothers had at least one college degree, 24% had some education beyond high school, 18% stopped with high school, 1% had not finished high school, and 2% did not report mother's level of education.
For each grade level, 10 subjects were selected as intellectually gifted, as defined by their obtaining Verbal-IQ scores on the WISC-R of 122 or greater. The mean scores for the gifted group were as follows: first grade, 128.1 (SD = 5.04); second grade, 139.0 (SD = 6.22); third grade, 139.0 (SD = 6.27); fourth grade, 128.9 (SD = 7.28); fifth grade, 132.4 (SD = 7.76); and sixth grade, 129.1 (SD = 5.70). The remaining 10 subjects in each grade are referred to as the "average" group. These children were matched to the gifted group in terms of gender, grade, and chronological age and were selected for having WISC-R Verbal IQs as close to 100 as possible. The means for the average group were as follows: first grade, 100.2 (SD = 1.48); second grade, 100.1 (SD = 1.20); third grade, 99.7 (SD = 2.21); fourth grade, 99.9 (SD = 2.81); fifth grade, 100.4 (SD = 3.27); and sixth grade, 100.1 (SD = 3.22). For all grades combined, the gifted students obtained a mean Verbal IQ of 132.8 (SD = 7.73), and the average students obtained a mean of 100.1 (SD = 2.40).
This battery assesses six component writing skills, including handwriting, spelling, compositional fluency (number of words or clauses produced within time constraints), sentence complexity (syntactic packaging rather than content), vocabulary sophistication, and compositional quality (mean of two raters' judgment of the content and text organization on a 5-point scale). It also assesses developmental skills related to writing, such as neuromotor finger function, phonemic skills (segmenting sound components of spoken words), orthographic coding (segmenting letter components of written words), orthographic-motor integration (retrieval of alphabet letters from memory), reading (word recognition), and Verbal IQ. Vocabulary sophistication, compositional quality, and Verbal IQ tap high-level processes, but the remaining measures in this battery tap low-level processes.
Below is a description of the administration and scoring for each of these measures. All the measures of writing skills were administered individually to the primary grade students and in a group session to the intermediate students.
Dependent Measures of Writing Skills
Handwriting. The Copy subtest of the Group Diagnostic Reading Aptitude and Achievement Tests (Monroe 61 Sherman, 1966) was given. Each child was asked to copy a short story as quickly as possible without mistakes. The score was the number of words accurately copied within the 90-second limit.
Spelling. The Wide Range Achievement Test--Revised (Jastak & Wilkinson, 1984) Spelling subtest, a dictation test, was administered according to standard procedures.
Compositions. Scoring for this analysis was based on each student's production of two compositions in response to topic frames held constant for all children. The first composition used the narrative frame, "One day _____ had the (best/worst) day at school." The second composition used the expository frame, "I like (person, place, or thing)______ because _______." For each composition, the child was instructed to choose the subject and then write about it for five minutes (for procedural details, see Berninger, Yates, et al., 1992). Deno, Mirkin, Lowry, and Kuehnie (1980) showed that number of words written during a brief writing sample (e.g., 3 minutes) correlates at greater than .70 with longer, standardized measures of writing achievement such as the version of Test of Written Language (Hammill &. Larsen, 1978) prior to its publication by Pro-Ed. Five measures were scored for each of the two compositions: number of words produced, number of clauses produced, sentence complexity, coded vocabulary sophistication, and mean quality rating for the whole text.
Number of words and number of clauses for both compositions were used as measures of compositional fluency and are thought to reflect lower-level transcribing skills. Sentence complexity and coded vocabulary sophistication were calculated by using modifications of the methodology described in Whitaker, Berninger, Johnston, and Swanson (1994).
Sentence complexity was determined by assigning a developmental maturity score of 1 to 5 to each sentence, with compound sentences receiving higher scores than simple sentences (see Appendix 1 for details). Each composition's total complexity score (summed over sentences) was divided by the number of sentences in the composition to obtain a score that represented the average sentence complexity for the composition.
Vocabulary sophistication was determined by assigning each word a score based on frequency of use (within the entire subject group across both compositions), excluding proper nouns and numerals. Plurality and tense variations of a word were treated as the same word for scoring purposes (see Appendix 1 for scoring procedures). A low score in this measure reflects use of less common words, and therefore indicates a high degree of vocabulary sophistication.
Mean quality ratings for each composition were determined by asking two experienced elementary teachers on our research team to read each of the compositions and rate them from 1 (considerably below grade level expectations) to 5 (considerably above grade level expectations). These raters were trained and instructed to base the rating on content and text organization. The two teachers' scores for each composition were averaged to arrive at a final score. The interrater Pearson product moment correlation on this scale was .75.
Dependent Measures of Writing-Related Skills
All measures were administered individually to primary grade students. For the intermediate grade students, all measures were administered individually except the Alphabet Task, Colorado Perceptual Speed Test, and the Homophone/Pseudohomophone Choice Test, which were group-administered.
Alphabet Task. Children were given pencils without erasers and told to print the lowercase alphabet in order as quickly as possible without making any mistakes (Berninger & Rutberg, 1992; Berninger, Yates, et al., 1992). Capital letters, omissions, reversals, transpositions, and substitutions were counted as errors, for which points were neither added or subtracted. One point was given for each correctly sequenced letter in the first 15 seconds. This task correlates with handwriting, spelling, and all compositional fluency measures used in this study (Berninger, Yates, et al., 1992).
Neuromotor Function (Finger Succession Task). The examiner modeled the finger succession task (Wolff, Gunnoe, & Cohen, 1983) and provided practice trials. The child was asked to touch the thumb to each finger in succession from the little finger to the index finger without looking. The time (in seconds) required to complete five cycles was recorded. Only the time for the child's dominant hand for writing was used in these analyses. This task is thought to draw upon motor planning and programming for the sequential finger movements needed for paper and pencil writing (Berninger &. Rutberg, 1992). Finger succession was the only neuromotor task included in the battery because it correlated more consistently and more highly with writing measures than any of the other finger function tasks originally used (Berninger, Yates, et al., 1992).
Phoneme Segmentation. The modified Rosner Test of Auditory Skills Analysis (see Berninger, Thalberg, DeBruyn, & Smith, 1987, for stimuli and procedures) required the child to repeat a word (e.g., game) and then say it again without a designated phoneme (e.g., /m/). The score was the percentage correct. Phoneme segmentation correlates with handwriting, spelling, and composition measures (Berninger, Yates, et al., 1992). This task was not used with intermediate grade students because third graders begin to approach ceiling.
Phoneme Articulation. The Phoneme Articulation Task (Vellutino & Scanlon, 1987; Vellutino, Scanlon, Small, & Tanzman, 1991) has a sufficient ceiling for intermediate students. The examiner read ten different pairs of words (half real words and half nonsense words) that differed in only one phoneme. For example, a real word pair was cup/cop and a nonsense pair was thope/fope. The child attempted to produce the sound that was different in each word or nonword pair. The score was the percentage correct. Phoneme articulation correlates with spelling (Berninger, Cartwright, et al., in press).
Orthographic Coding. The Receptive Orthographic Coding Test (Berninger, Yates, 61 Lester, 1991, Appendix II) was given to primary grade children. Orthographic coding is defined as the ability to hold a word in memory and then to access the whole word pattern or its components (single letter or letter cluster). For each trial, the student was shown a word for one second followed immediately by either a word, a letter, or a letter cluster and asked to determine whether the second stimulus matched the first or appeared in the first. (See Berninger, Yates, & Lester, 1991, for further procedural details.) The score was the percentage correct on whole word, letter, and letter cluster trials. This task was not used with intermediate grade students because third graders began to approach ceiling. Receptive Orthographic Coding correlates with handwriting, spelling, and compositional fluency measures.
The Colorado Perceptual Speed Test (Decker & DeFries, 1981; DeFries, 1985) has sufficient ceiling for intermediate students. Children were asked to match a target stimulus of alphanumeric symbols (e.g., acsr) with one of four alternatives on the same line to the right (e.g., rcas, acsr, sacr, rsca) by circling it. During three one-minute trials of 30 items each, the children were instructed to make as many correct matches as possible. The score was the percentage correct for the total number of items (90). This task correlates with handwriting, compositional fluency, and compositional quality measures.
The Homophone/Pseudohomophone Choice Task (adapted from a computer-based test developed by Olson, Kliegl, Davidson, &. Foltz, 1985; Olson, Wise, Conners, Rack, &. Fuller, 1989) required students to circle the real word in a pair of words pronounced the same, but spelled differently (e.g., sammon, salmon). The score was the percentage correct. This task correlates with spelling measures.
Expressive Orthographic Coding required students to look carefully at a nonword for one second and then to reproduce the word, a single letter in a designated position (e.g., the third letter), or a letter sequence in a designated position (e.g., the last two letters). This task corresponds to the Receptive Orthographic Coding Task given to the primary grade students in that students had to process orthographic units of varying size. The score was percent correct summed over the whole word, letter, and letter cluster trials. This measure correlates with handwriting and spelling measures.
Reading Real Words. The Word Identification subtest of the Woodcock Reading Mastery Test--Revised (Woodcock, 1987) was administered according to standard procedures. Raw scores were converted to standard scores using age norms in the test manual.
Reading Nonwords. The Word Attack subtest of the Woodcock Reading Mastery Test--Revised (Woodcock, 1987) was administered according to standard procedures. Raw scores were converted to standard scores using age norms in the test manual.
Verbal IQ. Four subtests of the Wechsler Intelligence Scale for Children--Revised (Wechsler, 1974)--Information, Similarities, Vocabulary, and Comprehension--were administered and scored according to standard procedures. A prorated Verbal IQ was derived from a table in the test manual. (See rationale below for not including Full-Scale IQ).
Measures were administered to the primary grade students individually during two one-hour sessions. They were administered to intermediate grade students in two (45-60 minute) group sessions plus one 45-minute individual session. Because of the current controversy over whether IQ should be used in defining learning disability (e.g., Bryan, 1989; Meyen, 1989; Siegel, 1989; Torgesen, 1989), we used both a relative criterion (discrepancy between IQ and achievement) and an absolute criterion (lowest 5% of the distribution) (see Berninger, Hart, et al., 1992) to identify children with specific writing disability. To identify "underachieving" students based on the relative criterion (those with lower performance on a task than expected based on IQ score), Verbal IQ was used because it is a better predictor of academic achievement than Full-Scale IQ, which includes Performance-IQ measures (Sattler, 1989). Also, rather than using traditional regression for evaluation of discrepancies between IQ and achievement, we used the Mahalanobis statistic (Stevens, 1986). The advantages of the Mahalanobis statistic are that (a) unlike the regression discrepancy model (Reynolds et al., 1983), which assumes a unidirectional relationship between IQ and achievement (IQ sets limits on achievement), it assumes a bidirectional relationship between IQ and achievement (IQ influences achievement and achievement influences IQ); and (b) because it is based on a random-effects rather than fixed-effects model, it allows for generalization beyond the specific IQ values in this study. (See Berninger, Hart, et al., 1992 for further discussion of these issues.)
The following results are based on independent 2 (ability groups) x 6 (grade levels) analyses of variance. This design was repeated for all dependent measures. An overview of the results based on analysis of variance, with means and standard deviations for significant effects, is provided in Tables 1 and 2.
For number of words on narrative compositions, neither the ability group nor the ability group x grade interaction was significant. However, the main effect for grade was significant, F (5, 108) = 20.073, p < .001. Sixth graders produced more words than fifth graders, who produced more words than fourth graders, and so on. For number of words in expository compositions, again, only the main effect for grade was significant, F (5, 108) = 21.532, p < .001.
Thus, for both narrative and expository compositions there were developmental changes, in that the older students were more fluent (i.e., wrote more within a constant time interval), but there were no ability group differences in compositional fluency. Gifted students were not significantly better than average students in low-level compositional fluency.
For number of clauses in narrative compositions, the ability group effect was not significant. However, the main effect for grade, F (5, 108) = 8.526, p < .001, and the ability group x grade interaction, F (5, 108) = 2.352, p < .05, were significant. The ability group x grade analysis indicated that gifted students in first, second, and fourth grades produced more clauses than their average-ability peers, whereas the opposite was true in third grade and fifth grades. Among sixth graders, gifted and average students produced the same number of clauses. The significant interaction of grade x ability group was most pronounced in first and second grades, where gifted students produced more clauses than their average-IQ grade peers. Thus, compositional fluency (as measured by the number of clauses produced in narrative compositions) was more advanced in gifted first and second graders than nongifted grade peers, but that relationship was not consistently evident in higher grades. For number of clauses in expository compositions, again, only the main effect for grade was significant, F (5, 108) = 12.012, p < .001.
Thus, the grade effect is reliable across narrative and expository compositions, but the interaction is not. There were developmental changes in low-level compositional fluency, as indexed by the number of words as well as the number of clauses, but there were not significant ability group differences in low-level compositional fluency, as indexed by the number of words (in either narrative or expository writing) or number of clauses (in expository writing). Significant interaction of grade x ability group was found only in the clause count of narrative compositions, but this finding did not replicate for the expository compositions. Therefore, in general, gifted students were not significantly better than average students in low-level compositional fluency, whether it was measured by number of words or number of clauses.
For sentence complexity on narrative compositions, only the main effect for grade was significant, F (5, 108) = 3.99, p < .002, but it did not reflect a clear developmental trend. Fourth graders had the most syntactically complex sentences, followed (in order of complexity) by sixth graders, fifth graders, second graders, first graders, and third graders. Apparently individual differences are greater than developmental trends for sentence complexity. Furthermore, for sentence complexity on expository compositions, there were no significant effects. Thus, the developmental changes were not reliable across the narrative and expository compositions. Gifted students did not show more low-level syntactic complexity than average students. This finding is consistent with prior research suggesting that measures of clause content and complexity are inherently unreliable as predictors of general writing performance (Deno, Marston, 61 Mirkin, 1982; Newcomer &. Barenbaum, 1991).
For coded vocabulary sophistication on narrative compositions, only the main effect for ability group was significant, F (1, 108) = 4.385, p < .04. High-IQ students showed more sophistication in vocabulary choice (i.e., used words that were less common, thereby producing lower scores) than average-IQ students. For coded vocabulary sophistication on expository compositions, the main effect for ability group, F (1, 108) = 8.434, p < .004, and the interaction between ability group x grade, F (5, 108) = 2.998, p < .02, were significant. At all grade levels except fifth grade, the gifted students showed more vocabulary sophistication in expository compositions than their average-IQ grade peers. Thus, only the ability group difference was reliable across narrative and expository compositions for vocabulary sophistication. Gifted students showed more sophistication in high-level vocabulary choice than did average students.
For quality ratings on narrative compositions, the main effects for ability group, F (1, 108) = 13.942, p < .001, and for grade, F (5, 108) = 3.661, p < .01, were significant. High-IQ students wrote compositions rated higher in quality than average-IQ students. No developmental pattern emerged to explain why the main effect for grade was significant, except the fact that first graders scored substantially below all other grades, but the ability group x grade interaction was not significant.
For quality ratings on expository compositions, the main effect for ability group, F (1, 108) = 22.696, p < .001, and for grade, F (5,108) = 3.464, p < .01, were significant. High-IQ students wrote compositions rated higher in quality than average-IQ students. Again, no developmental pattern emerged to explain why the main effect for grade was significant, except the fact that first graders scored substantially below all other grades, but the ability group x grade interaction was not significant.
Thus, for both narrative and expository compositions, there were ability differences in compositional quality, in that high-IQ students produced higher quality compositions than their average-IQ grade peers. Gifted students were significantly better than average students in high-level compositional quality.
The results for each of the 120 individual subjects were analyzed to determine which students showed either absolute or relative disabilities in one or more writing or writing-related skills. An absolute disability was defined as the lowest 5% of the normal distribution (≤1.65z-score) for a particular variable, based on the mean and standard deviation for each grade. A relative disability was defined as a score significantly below that expected by Verbal IQ, based on the Mahalanobis statistic at a .05 level, one-tail test.
The tasks used were placed into two general categories. The first category--Writing Skills--included handwriting, spelling, and compositional fluency (based on word count and clause count for both narrative and expository compositions). Because compositional fluency was assessed on four different indices (narrative words, narrative clauses, expository words, and expository clauses), only students meeting the disability criteria for two of the four indices were considered to have a disability in that area. The second category--Writing-Related Skills--included the Alphabet Task, Finger Succession, Phoneme Segmentation/Articulation, Orthographic Coding, Word Identification (Reading), and Word Attack (Reading). For orthographic coding, only students meeting the disability criteria for two of the three component skills (for primary grades: whole word, letter coding, and letter cluster coding; for intermediate grades: Colorado Perceptual Speed Test, Homophone/Pseudohomophone Choice Task, and Expressive Orthographic Coding) were considered to have a specific disability in that area.
We computed the percentage of children who had at least one writing or writing-related disability based on relative criteria. In first grade, 40% of the gifted students and 10% of the average students met the relative criterion for a disability in at least one of the writing or writing-related skills. In second grade, 50% of the gifted students and 10% of the average students met the relative criterion. In third grade, 60% of the gifted students and 0% of the average students met the relative criterion. In fourth grade, 20% of the gifted students and 10% of the average students met the relative criterion. In fifth grade, 40% of the gifted students and 20% of the average students met the relative criterion. In sixth grade, 50% of the gifted students and 0% of the average students met the relative criterion. For all grades combined, 43.3% of the gifted students and 8.3% of the average students met the relative criterion in at least one of the writing or writing-related skills. Thus, our prediction was confirmed that relative criteria are more likely to identify writing disabilities in gifted children.
However, we emphasize that these percentages do not reflect the number of gifted children who would qualify for special education services as learning disabled. For most of these children the specific writing or writing-related disability occurred on only one of the nine skills examined. Thus, they had a relative weakness in one aspect of writing, which might or might not compromise their writing performance in the classroom, depending on the pattern of the relative strengths and weaknesses in their overall profile of processing and writing abilities. This study was not an epidemiological study of the incidence of writing disabilities in the gifted population. Rather, it was an exploratory study of whether writing and writing-related problems are more likely to be detected on the basis of relative or absolute criteria in gifted children.
Conversely, using absolute criteria, gifted students are underrepresented. In first grade, 0% of the gifted students and 30% of the average students met the absolute criterion for a disability in at least one of the writing or writing-related skills. In second grade, 20% of the gifted students and 20% of the average students met the absolute criterion. In third grade, 30% of the gifted students and 30% of the average students met the absolute criterion. In fourth grade, 0% of the gifted students and 30% of the average students met the absolute criterion. In fifth grade, 10% of the gifted students and 20% of the average students met the absolute criterion. In sixth grade, 10% of the gifted students and 20% of the average students met the absolute criterion. For all grades combined, 11.7% of the gifted students and 25% of the average students met the absolute criterion in at least one measure. Thus, absolute criteria are more likely to identify disabilities in average children than in gifted children.
Again, however, we caution that these percentages do not reflect the number of gifted children who would qualify for special education services as writing disabled. Rather, they reflect the number of children who might have at least one relative writing or writing-related weakness that could compromise writing performance, depending on the relative strengths and weaknesses in their overall profile of writing and writing-related abilities. Many gifted children's profiles contain many relative strengths that may allow them to compensate for an isolated relative weakness or disability. That is not always the case, however, without intervention, as the two case studies discussed in the next section show.
The results of this study support our hypothesis that the advanced high-level writing skills of gifted children do not necessarily indicate similarly advanced low-level writing skills. Although translation of ideas into written words involves many subtle processes, our focus was on (a) high-level skills involving text generation (transforming ideas into language representations) and (b) low-level skills related to transcription (transforming language representations into written symbols on the printed page). High-level skills were indeed found to be consistently more advanced in gifted students than their average peers. Two measures of high-level text generation reflected a strong influence by IQ grouping--vocabulary sophistication and composition quality. Because these measures reflect high-level linguistic and cognitive ability , the results confirmed our hypothesis that the gifted students would show relatively more advanced abilities in the high-level processes of writing.
On the other hand, the gifted and average groups did not differ in the two measures of lower-level compositional fluency skills--number of words and number of clauses (without regard to content). Additionally, the gifted and average groups did not tend to differ in low-level sentence complexity (packaging of language without regard to content). These findings were consistent with our hypothesis that gifted children do not differ from their average-IQ peers in low-level transcription skills.
It follows that for gifted children, specific writing disabilities likely result from deficits in low-level skills rather than in high-level skills. Our individual subject analyses shed light on what these lower-level skills might be: fine-motor, orthographic coding, phonological coding, orthographic-motor integration, and letter-sound correspondences in word recognition.
The individual subject analyses also support the use of relative criteria in identifying specific writing disabilities (Berninger, Hart, et al., 1992), in gifted children. Because of their advanced intellectual skills, gifted children may compensate for low-level deficiencies up to a point. Their low-level writing deficiencies may not fall at the bottom of the normal distribution but may be significantly discrepant from their high-level writing capabilities. Such discrepancies will only be noted if low-level skills are evaluated relative to Verbal IQ.
In general, our study has three major implication for teachers of gifted students. First, it is important to distinguish between high-and low-level writing processes. Gifted students may have superior text generation abilities that are masked or compromised by average or below average transcription abilities. A distinction should be drawn between text generation and transcription.
Second, teachers should be cognizant of the potential frustration of gifted children struggling with such low-level disabilities. Creative children are likely to be very frustrated by their inability to adequately represent their ideas on paper in a way that meets parent, teacher, and peer expectations. Their emotional responses may include loss of motivation or self-esteem, general reluctance to engage in writing, and/or increased verbal precocity or other behavior changes in an attempt to compensate. Instead of attributing these children's lack of interest in the mechanical aspects of writing to boredom, we should identify the low-level skills in which an individual child is deficient and either remediate them or teach them ways to compensate for these deficiencies.
It follows that gifted children with specific writing disabilities will require explicit teaching to deal with these disabilities. In general, programs for the gifted focus on high-level thinking and problem solving skills and not on low-level mechanical skills. However, some of these children also will need instruction directed toward the mechanical aspects of writing. Two gifted third-grade boys who participated in our treatment program for writing disabilities (Berninger, Abbott, Whitaker, &. Sylvester, in preparation) illustrate these points. The first, AZ, was removed from a gifted program because of his writing problems, which were attributed to motivational problems. According to his mother, he could read Moby Dick and discuss it intelligently, but he could not write a book report about it. The second, BY, was on the verge of being removed from the gifted program because of failure to keep up with its writing requirements. He had begun to refuse to write. Both boys were given our writing assessment battery prior to the sixteen 1-hour individualized tutorial sessions. Tutoring consisted of handwriting warm-up (activities designed to automatize low-level letter production skills), spelling instruction (3 strategies: orthographic imaging, letter-sound relationships, syllable patterns), and composition (planning, translating, reviewing, and revising). The emphasis was on getting thoughts down on paper without the pressure to make the first draft perfect.
At pretest, AZ's Verbal IQ fell in the very superior range, but only his verbal working memory was developed to the level expected on the basis of his Verbal IQ. His performance on the alphabet task was average for grade, but his handwriting fluency on the timed-copy task was more than a standard deviation below the mean for his grade level. His spelling was in the average range (.5 standard deviation below the mean). His compositional fluency was in the average range for grade. His phonological coding skills (phoneme and syllable segmentation on the Modified Rosner) were more than a standard deviation below the mean for his grade level. His phonological processing problems explained his relative difficulty in both spelling and word attack (low average range) compared to his superior verbal ability. Orthographic coding was in the average to above average range. In addition to the standard protocol, AZ also was given training in phonological coding. At posttest, he had made significant gains in both phonological coding and spelling (+2/3 standard deviation) and word attack (+3/5 standard deviation). Although his rate of output remained slow, he made qualitative gains in his attitude toward writing and the strategies he used for spelling. At posttest, he stated that what he had learned was, "It isn't important to be the best on everything; what is important is to do your best and get as many words as you can written down--speed does not count. If you try different ways of writing, it is easier when you find the way that works for you. Two ways worked for me--syllable patterns and orthographic imaging strategy. I learned different ways to learn words and that if I can't figure out a word try writing a different word that means the same." This boy's writing problems had previously been attributed to lack of motivation because his writing skills appeared to be average. Yet when relative criteria based on IQ were used, it was apparent that he had disabilities in phonological coding and rate of written output. The former responded to short-term intervention, whereas the latter did not. Nevertheless, his former reluctance to write (because he could not spell words he wanted to use) gave way to a willingness to write once he was taught strategies for spelling and the pressure to produce writing under time constraints was removed.
At pretest, BY's Verbal IQ fell in the superior range. His handwriting and spelling skills were developed to the level expected on the basis of IQ, but his compositional fluency (ability to get ideas on paper under time constraints) was only in the average range--considerably below what was expected relative to IQ. He also had a disability in working memory,1 based on relative criteria (a difference of almost 2 standard deviations between IQ and working memory standard score). BY reported that he has a writer's block, which was probably related to his frustration in juggling all the requirements of writing--setting goals, making plans, generating ideas, transcribing ideas into the written symbol system, etc.--when his limited capacity working memory overloaded. A child with superior verbal reasoning skills may become very frustrated by losing ideas before they get transcribed on paper. By posttest, BY made gains in both working memory (+.75 standard deviation) and compositional fluency (+1 standard deviation). At posttest, when asked what he had learned, he stated, "I put 100 sticks of dynamite in the writer's block and it blew up!" When asked what he wants to do better as a writer, he wrote, "Write more." His initial disability was apparent only when relative criteria based on IQ were used. The tutorial program helped him overcome the diagnosed specific disability by developing strategies for avoiding an overload in working memory. Instruction directed toward dealing with this specific disability resulted in both quantitative and qualitative gains, whereas failure to identify and remediate the specific disability may have resulted in continuing refusal to write and avoidance of writing.
Because gifted children have more advanced high-level cognitive abilities (as measured by IQ) than their average-IQ peers, we are often inclined to assume that they will also master lower-level processes sooner and/or more completely than their average-IQ peers. Teachers and parents often assume that gifted students who exhibit poor performance in reading and writing "must be lazy" because they obviously are "smart" enough to do better. As this study shows, that is an erroneous assumption; it is possible for a child with an IQ of 130 to require much the same remediation of orthographic or phonological coding, fine-motor, or orthographic-motor integration skills as a child of the same age with an IQ of 100 (or lower). In order to maximize each child's unique potential, we must discover and remediate as necessary whatever obstacles stand in that child's way. When specific writing disabilities exist in gifted children, their teachers, specialists, and parents should be careful not to ignore basic low-level skills, which may hold the key to setting free the cognitive and creative potential waiting to manifest itself in written expression.
1The working memory measure was not included in this study because different measures of working memory were used in the primary and intermediate grade samples.
The authors thank Nancy Robinson for helpful comments on a prior version of this manuscript.
Scoring Criteria for Student Compositions
- Coded Vocabulary Sophistication
All words were tabulated except proper nouns. The following were considered identical for scoring purposes:
Singular and plural forms of the same noun
Different tenses of the same verb
Different representations of the same cardinal number
Different representations of the same ordinal number
The score for each word was its frequency count among all words used by all subjects in this analysis.
- Sentence Complexity
1 Point: A sentence fragment
2 Points: An independent clause (simple sentence)
3 Points: Use of a linking word when there is only one independent clause
4 Points: Use of a linking word to join two independent clauses (compound sentence)
5 Points: Use of a subordinate/dependent clause as well as independent clause (complex sentence)
- Compositional Quality Rating
Each composition was evaluated by two experienced elementary teachers and given an overall compositional quality score by each according to the following scale:
1 Point: Significantly below grade level expectation
2 Points: Below grade level expectation
3 Points: At grade level
4 Points: Above grade level expectation
5 Points: Significantly above grade level expectation The mean of the two raters' scores was used.
Cheryl M. Yates is an advanced doctoral student in school psychology at the University of Washington. She is interested in early acquisition of writing and mathematics skills. Virginia W. Berninger is Professor and Coordinator of School Psychology at the University of Washington. She is interested in reading and writing acquisition in gifted, normally developing, and handicapped children and has recently written Reading and Writing Acquisition: A Developmental Neuropsychological Perspective, published by WCB Brown &. Benchmark. Robert D. Abbott is Professor of Educational Psychology at the University of Washington. He is a fellow of the American Psychological Association and has published extensively on the application of structural equation models and longitudinal data analysis strategies to study children with language and learning problems.
This research was supported, in part, by Grant No's. 25858-02 and 25858-04 from the National Institute of Child Health and Human Development to the second author.
Abbott, R., &. Berninger, V. (1993). Structural equation modeling of relationships among developmental skills and writing skills in beginning and developing writers. Journal of Educational Psychology, 85,478-508.
Berninger, V., &. Hart, T. (1993). From research to clinical assessment of reading and writing disorders: The unit of analysis problem. In R.M. Joshi 61 C.K. Leong (Eds.), Reading disabilities: Diagnosis and component processes (pp. 33-61). The Netherlands: Kluwer Academic Publishers.
Berninger, V., & Rutberg, J. (1992). Relationship of finger function to beginning writing: Application to diagnosis of writing disabilities. Developmental Medicine & Child Neurology, 34, 155-172.
Berninger, V., & Yates, C. (1993 ). Formal operational thought in the gifted: A post-Piagetian perspective. Roeper Review, 15, 220-224.,/p>
Berninger, V., Abbott, R., Whitaker, D., & Sylvester, L. (in preparation). Integrating low-level skills training and high-level process instruction in treatment protocols for writing disability. To appear in Learning Disability Quarterly.
Berninger, V., Cartwright, A. C., Yates, C. M., Swanson, H. L., & Abbott, R. D. (in press). Developmental skills related to writing and reading acquisition in the intermediate grades. Reading and Writing: An Interdisciplinary Journal.
Berninger, V., Hart, T, Abbott, R., Karovsky, P. (1992). Defining reading and writing disabilities with and without IQ: A flexible, developmental perspective. Learning Disabilities Quarterly, 15, 103-118.
Berninger, V., Thalberg, S., DeBruyn, I., &. Smith, R. (1987). Preventing reading disabilities by assessing and remediating phonemic skills. School Psychology Review, 16, 554-565.
Berninger, V., Yates, C., & Lester, K. (1991). Multiple orthographic codes in reading and writing acquisition. Reading and Writing: An Interdisciplinary Journal, 3, 115-149.
Berninger, V., Yates, C., Cartwright, A., Rutberg, J., Remy, E., & Abbott, R. (1992). Lower-level developmental skills in beginning writing. Reading and Writing: An Interdisciplinary Journal, 3, 1-25.
Bryan, T. (1989). IQ and learning disabilities: A perspective from research on social factors. Journal of Learning Disabilities, 22, 480-481.
Decker, S. N., & DeFries, J. C. (1981). Cognitive ability profiles in families of reading-disabled children. Developmental Medicine and Child Neurology, 23, 217-227.
DeFries, J. C. (1985). Colorado reading project. In D.B. Gray &. J.F. Kavanaugh (Eds. ), Biobehavioral measures of dyslexia, pp. 107-122. Parkton, MD: York Press.
Deno, S., Marston, D., & Mirkin, P. (1982) Valid measurement procedures for continuous evaluation of written expression. Exceptional Children, 48, 358-371.
Deno, S. L., Mirkin, P. K., Lowry, L., & Kuehnie, K. (1980). Relationship among simple measures of spelling and performance on standardized achievement tests. (Research Report No. 21). Minneapolis: University of Minnesota, Institute for Research on Learning Disabilities. (ERIC Document Reproduction Service No. ED 197 508)
Fox, L., Brody, L., &. Tobin, D. (1983). Learning disabled/gifted children. Baltimore: University Park Press.
Ganschow, L., 1985. Diagnosing and remediating writing problems of gifted students with language learning disabilities. Journal for the Education of the Gifted, 9, 25-43.
Hammill, D., & Larsen, S. (1978). The effectiveness of psycholinguistic training: A reaffirmation of position. Exceptional Children, 44, 402-417.
Hayes, J., & Flower, L. (1980). Identifying the organization of the writing process. In L. W. Gregg & E. R. Steinberg (Eds. ), Cognitive processes in writing, pp. 3-30. Hillsdale, NJ: Erlbaum.
Jastak, S., & Wilkinson, G. (1984). Wide Range Achievement Test-- Revised. Wilmington, DE: Jastak Associates.
Keating, D. (1991). Curriculum options for the developmentally advanced: A developmental alternative to gifted education. Exceptionality Education Canada, 1, 53-83.
Master, D. L., (1983). Writing and the gifted child. Gifted Child Quarterly, 27, 162-168.
Meyen, E. (1989). Let's not confuse test scores with the substance of the discrepancy model. Journal of Learning Disabilities, 22, 482-483.
Monroe, M., & Sherman, E. (1966). Group Diagnostic Reading Aptitude and Achievement Tests. Bradenton, FL: C. H. Nevins Printing Co.
Newcomer, P. L., &.Barenbaum, E. M. (1991). The written composing ability of children with learning disabilities: A Review of the Literature from 1980 to 1990. Journal of Learning Disabilities, 24, 578-593.
Olson, R. K., Kliegl, R., Davidson, B. J., & Foltz, G. (1985). Individual and developmental differences in reading disability. In G. E. Mac-Kinnon & T. G. Waller (Ed.), Reading research: Advances in theory and practice: Vol. 4, (pp. 1-64). New York: Academic Press.
Olson, R. K., Wise, B., Conners, F., Rack, J., & Fuller, D. (1989). Specific deficits in component reading and language skills: Genetic and environmental influences. Journal of Learning Disabilities, 22, 339-348.
Reynolds, C. R., Berk, R. A., Boodoo, G. M., Cox, J., Gutkin, T. B., Mann, L., Page, E. B., & Wilson, V. L. (1983). Critical issues in learning disabilities. Washington, DC: U. S. Department of Education, Special Education Work Group on Measurement Issues in the Assessment of Learning Disabilities.
Sattler, J. M. (1989). Assessment of children. San Diego: Author.
Shaywitz, S., Shaywitz, B., Fletcher, J., & Escobar, M. (1990). Prevalence of reading disability in boys and girls. Results of the Connecticut Longitudinal Study. Journal of the American Medical Association, 264, 998-1002.
Siegel, L. S. (1989). IQ is irrelevant to the definition of learning disabilities. Journal of Learning Disabilities, 22, 469-486.
Siegel, L. (1992). An evaluation of the discrepancy definition of dyslexia. Journal of Learning Disabilities, 25, 618-629.
Stevens, J. (1986). Applied multivariate statistics for the sciences. Hillsdale, NJ: Laurence Erlbaum Associates.
Torgesen, J. K. (1989). Why IQ is relevant to the definition of learning disabilities. Journal of Learning Disabilities, 22, 484-486.
Vellutino, F. R., Scanlon, D. M. (1987). Phonological coding, phonological awareness, and reading ability: Evidence from a longitudinal and experimental study. Merrill-Palmer Quarterly, 33, 321-363.
Vellutino, F. R., Scanlon, D. M., Small, S. G., & Tanzman, M. S. (1991). The linguistic basis of reading ability: Converting written to oral language. Text, 11, 99-133.
Wechsler, D. (1974). Manual for the Wechsler Intelligence Scale for Children--Revised. San Antonio: The Psychological Corporation.
Whitaker, D., Berninger, V., Johnston, J., &. Swanson, H. L. (1994). Intraindividual differences in levels of language of developing writers: Implications for the translating process. Learning and Individual Differences, 6, 107-130.
Wolff, P. H., Gunnoe, C., & Cohen, C. (1983). Associated movements as a measure of developmental age. Developmental Medicine &
Child Neurology, 25, 417-429.
Woodcock, R. (1987). Woodcock Reading Mastery Test--Revised.
Circle Pines, MN: American Guidance Service.