Reflective spelling strategies for elementary school students with severe writing difficulties: a case study.

Abstract. A case study with two Finnish 10-year-old boys evaluated an intervention designed to promote the spelling skills of elementary school students with severe writing difficulties. The intervention comprised strategy instruction, procedural facilitation, and computer-assisted tutoring. Transfer was facilitated by involving teachers and parents. The results showed gains in spelling accuracy, spelling revision skills, decoding accuracy, and knowledge about the writing process from baseline to posttreatment phase. Transfer, parents' and teachers' participation, and the stability of intervention effects over a six-month follow-up period are discussed.

**********

Skilled writing includes the cognitive subprocesses of planning, translation, and reviewing (Berninger, Abbott, Whitaker, Sylvester, & Nolen, 1995; Hayes, 1996; Hayes & Flower, 1980). In the early school years, learning the skills related tO the translation process has conventionally had a central role. This is understandable since translation skills enable students to make language, more or less elaborated, visible on paper. As McCutchen, Covill, Hoyne, and Mildes (1994) stressed: "The written product may not be very good if the writer fails to plan or revise, but there will be no product at all if the writer fails to translate" (p. 256).

In this article, we describe an intervention designed to promote spelling, one component of the translation subprocess of writing. In order to define the target skills of the intervention, we begin by describing the processes involved in skilled spelling. Then we describe and argue our instructional approach.

Spelling as a Component of the Translation Process

The translation process of writing has two component processes, text generation and transcription (Berninger, 1999; Berninger et al., 1995). Text generation refers to the translation of ideas into language representations. Transcription, in turn, refers to the translation of these generated language representations into written language by means of handwriting and spelling. Spelling requires that student can name letters, connect phonemes with corresponding letters, and produce letters with a pencil or word processor without too much fine-motor difficulty (Berninger, 1999; Maki, Voeten, Vauras, & Poskiparta, 2001; Sandler et al., 1992; see also Graham, 1999, for a review). A beginning speller should also be able to segment spoken words into syllables and phonemes in order to decide which letters are needed and in which order (e.g., Bradley & Bryant, 1983; Maki et al., 2001; Schneider & Naslund, 1993; see also Graham, 1999, for a review). Furthermore, spelling requires that the student has a knowledge of spelling rules, such as capitalization, punctuation, and specific irregular phoneme-grapheme correspondences.

An essential part of spelling is the ability to evaluate and revise the text (Wong, 1986). Specifically, in order to revise, students must be able to detect and correct their spelling errors. This seems to involve two separate but interrelated skills (Hacker, Plumb, Butterfield, Quathamer, & Heineken, 1994; Hayes, Flower, Schriver, Stratman, & Carey, 1987). First, students have to find spelling errors in order to correct them. However, to find spelling errors, they have to apply some spelling knowledge. Thus, detection of spelling errors requires that the student can decode the produced text in a way that allows detailed evaluation of the phonetic and syllabic structure of the words, as well as the application of specific spelling roles and lexical knowledge.

In addition to content generation and transcription of content into written language by spelling, skilled text production requires that the writer masters self-regulation strategies, which enable independent monitoring and regulating of the use of the writing subskills (Graham & Freeman, 1986; Harris & Graham, 1996). Harris and Graham (1996) stressed four self-regulation strategies that are important in writing: goal-setting, self-instructing, self-assessment, and self-reinforcement. Applied to spelling this would mean, for example, (a) deciding what to spell and activating standards against which to compare the spelling performance and outcomes; (b) telling oneself what to do while spelling with internal self-verbalizations; (c) monitoring the fluency of various spelling subskills; and (d) reinforcing oneself whenever spelling outcomes and performance have met the preset criterion.

On the basis of the viewpoints presented above, the writer should be able to coordinate the use of several spelling subprocesses, as well as plan and review simultaneously. The success of this task depends largely on working memory processes. The writer should be able to store in working memory all the necessary writing processes, and temporally coordinate their use during writing by means of executive working memory routines (Berninger, 1999; Berninger et al., 1995; Kellogg, 1996; McCutchen et al., 1994). If spelling is not automated, the writer has to consciously monitor the ongoing spelling process, which can take working memory capacity away from the other writing processes. For example, planning content or translation of ideas into language may be disrupted if the writer has to continuously stop to figure out, phoneme by phoneme, which letters comprise the word (e.g., Graham, 1999). In their review, Graham and Harris (2000) concluded that skilled writers tend to master transcription processes (spelling and handwriting) better than less skilled writers, while individual differences in transcription skills seem to predict writing achievement. Further evidence of the assumed causal connection between spelling skills and writing has been offered by Graham, Harris, and Fink Chorzempa (2001), who found that extra spelling instruction enhanced second-grade students' writing fluency. Because of the disadvantage that poor spelling can cause throughout the writing process, effective and early interventions are indispensable for students with severe spelling difficulties.

Aims of the Present Spelling Intervention

In his comprehensive review, Graham (1999) specified several approaches used in previous studies to teach spelling to students with learning disabilities (LD). He concluded that relying on incidental learning approaches, that is, immersing students in a literacy-rich environment believing they will pick up spelling skills while reading and writing for real purposes, is not enough for students with LD. Rather, spelling skills must be explicitly taught by considering the technical aspects of handwriting and letter formation, the spelling processes, as well as spelling as part of the writing process (Graham, 1999). The choices concerning the content of the spelling intervention, the application of a specific instructional method, and the learning environment must be interrelated. Methods and learning environments form alternative ways of dealing with the individual needs of a particular student and the content of the spelling instruction. On the basis of the review by Graham (1999), several activities can be successfully used to teach spelling in general. However, we are far from knowing which combinations of content and methods are most effective for students with severe writing difficulties.

The intervention presented in this study was designed to promote the spelling skills of elementary school students with severe difficulties in producing even short text units (words, sentences and paragraphs). We chose to apply cognitive strategy instruction as the instructional method, because a substantial body of research indicates that a prerequisite for skilled academic performance is the ability to use strategies and to regulate one's own strategy use (e.g., Snyder & Pressley, 1995). In the case of writing skills, pioneering work on strategy instruction has been done by Harris and Graham and their colleagues (see Harris & Graham, 1996, 1999). Their Self-Regulated Strategy Development (SRSD) approach to writing instruction helps students to

   ... understand how and when to apply a strategy; independently produce,
   evaluate, and modify a strategy effectively; recognize meaningful
   improvement in skills, processes, and products; gain new insights regarding
   strategies and their own strategic performance; improve their expectations
   of themselves as writers; and maintain and generalize strategic
   performance. (Harris & Graham, 1996, pp. 25-26; see also Englert, 1990;
   Pressley, McGoldrick, Cariglia-Bull, & Symons, 1995)

Writing strategies in SRSD are taught in six stages: developing background knowledge, discussing the strategy, modeling the strategy, memorizing the strategy, supporting/scaffolding the strategy use, and encouraging independent strategy use. The SRSD approach considers the composition process as a whole and, thus, treats spelling only as one aspect to be considered when composition is revised and edited. In our opinion, the principles of SRSD may also be applied specifically to spelling instruction.

In the present intervention, general guidelines from the Self-Regulated Strategy Development (Harris & Graham, 1996) were applied to teach spelling and spelling revision skills. The intervention had several aims. First, the students were introduced to an elementary composing strategy to help them understand the role of spelling and spelling revision in the writing process. Second, the students were taught specific strategies to produce spellings. These strategies represent the so-called phonics approach to spelling instruction (see Graham, 1999, for a review). We included a "listen-segment-code" strategy for producing phoneme-grapheme associations, and specific strategies for using capitalization and punctuation, as well as for producing spellings for the o-phoneme and double consonants. According to Graham (1999), teaching phonics to students with LD has been an understudied aspect of spelling instruction. He summarized only two previous studies (Darch & Simpson, 1990; Henry, Calfee, & LaSalle, 1989) that have reported the positive effects on spelling development of teaching LD students, for example, phoneme-grapheme associations and morpheme patterns.

The third aim of our intervention was to teach the students how to use accurate decoding as a strategy needed in spelling revision. The revision process during composing can include revising on several textual levels, and this should be reflected by the instruction. For example, Graham and MacArthur (1988) found that instruction in revising the clarity and cohesiveness of written essays, with spelling revision considered only as a somewhat more secondary goal, did not lead to enhanced spelling revisions among elementary school children. Therefore, we instructed spelling revision explicitly. The students were instructed to decode the text they had written as accurately as possible with the help of self-statements. Furthermore, computer-assisted instruction was used to help students to develop a more accurate decoding strategy for finding spelling errors in texts. Emphasis was put on encouraging the students to read their spellings both during the initial text production and during a separate revising phase.

The fourth aim was to enhance independent regulation of spelling and spelling revision strategies during composing (cf. Englert, 1990; Englert, Raphael, Fear, & Anderson, 1988; Harris & Graham, 1996). Self-regulation refers to children's capacity to plan, guide, and monitor their behavior according to

changing circumstances (Brown, 1987). Previous studies indicate that poor writers tend to lack self-regulation during writing, while enhanced writing-related self-regulation seems to result in improved writing performance (see Graham & Harris, 2000, for a review). In our intervention, self-regulation was supported with the following three steps: (a) The students' awareness of their spelling problems was facilitated with a "list of my problems." (b) The regulation of the writing process was facilitated with "writing instructions." These lists were made up through reciprocal discussion and by analyzing the students' written products and videotaped writing processes. And (c), the regulation of the student's writing process was gradually shifted from other-regulation to self-regulation (Tharp & Gallimore, 1988).

In sum, we applied the general instructional stages of SRSD (Harris & Graham, 1996) combined with procedural facilitation and computer-assisted instruction to teach spelling knowledge, spelling processes, and spelling revision skills. To be able to evaluate the durability of the intervention effects and the generalization of skills to other learning environments, we extended the follow-up period to six months, collected spelling samples also in the emphasized transfer setting, and included in the intervention explicit steps that are expected to promote transfer.

Some Remarks About Finnish as the Spelling Language

The vast majority of studies on spelling instruction have been carried out in so-called irregular languages where the words are infrequently written as they are pronounced. As a result, words include morphological units that cannot be produced according to a certain letter-sound connection or spelling rule, and the spelling instruction has to focus on a certain spelling vocabulary. In regular languages, such as Finnish, almost all words are pronounced as they are written. The alphabet principle of written Finnish represents 25 phonemes (8 vowels and 17 consonants) with 23 letters (not including letters, a, c, q, w, x, and z, which only appear in loan words) (Karlsson, 1983). The relationship between spelling unit and phoneme is very simple and almost symmetric. Each phoneme is represented by one letter with only two exceptions: the n-phoneme (spelled either as "nk" of "ng") and the s-phoneme (spelled as "sh"). Thus, spelling instruction in a regular language can include explicit rules about how to connect spoken words with letters. As a result, students need not learn a specific spelling vocabulary (see Graham, 1999, for a review), but a spelling strategy that enables them to write any word needed. Because of these vast differences in language systems themselves, every spelling instruction method must be tested within the language in which it is intended to be used.

METHOD

Participants

The participants were selected from a sample of children referred to the Centre for Learning Research (CLR) at the University of Turku, Finland. The CLR is a research and rehabilitation unit that combines psychological and educational expertise. Teachers, psychologists of physicians can refer children with severe learning difficulties to the CLR for diagnosis, rehabilitation and consultation. For this study, we chose two 10-year-old boys, Ville and Kalle, who had severe writing difficulties. Both students were from white Finnish-speaking families. Regarding the students' socioeconomic status, Ville's mother had graduated from university, whereas Kalle's mother had completed compulsory schooling plus further vocational education. Ville and Kalle were assessed before the intervention in the following: linguistic scale of individually administered intelligence test (WISC-R; Wechsler, 1984), phonological working memory, visual-motor skills of the writing hand, phonological processing, rapid naming, oral fluency (NEPSY; Finnish standardized neuropsychological test for children by Korkman, Kirk, & Kemp, 1997; see also Korkman, 1998), and decoding and reading comprehension (ALLU; Finnish standardized reading test by Lindeman, 1998) (see Table 1).

The intervention took place in the year when Ville was in the third grade and Kalle in the fourth grade of elementary school. In Finland, children enter the first grade in August of the year they turn seven years old. The students' schools followed the mainstream curriculum of Finnish general education, including teaching in the skills that were the targets of this study (decoding, spelling, and composition).

Ville was 10 years and 1 month of age at the start of the study and had been held back in the first grade. His psychologist in the local family counseling center referred him to the CLR. In the CLR, Ville reached an average level on the linguistic scale of the WISC-R (Table 1). In the neuropsychological test battery for children (NEPSY), Ville's score on phonological processing was three years, and on visual-motor accuracy, four years below his age level (on the other applied subtests of NEPSY he reached average levels) (Table 1). His decoding score was one year below his grade level, and he obtained an average level reading comprehension score on the ALLU. According to his classroom teacher, Ville often had severe attention problems in class.

Kalle was 10 years and 4 months of age. He was referred to the CLR by a clinical neuropsychologist in the local University Central Hospital. Kalle reached an average level on the linguistic scale of the WISC-R (Table 1). In a neuropsychological test battery for children (NEPSY), his scores on phonological processing and visual-motor accuracy were three years below his age level (on the other applied subtests of NEPSY he reached average levels) (Table 1). Kalle had a decoding score two years below his grade level and an average level reading comprehension score on the ALLU.

Ville and Kalle already experienced problems with learning to read and write in the first term of the first grade. Missing or incorrect letters, problems with syllabification and sentence structure, illegible handwriting, as well as scarce production, were typical features of their writing. During the first two school years, both students had gradually become reluctant to practice writing. Homework that included writing was difficult, and parents' constant attention was needed. The students were not motivated to practice writing and did not use their writing skills voluntarily in everyday life.

At school, Ville and Kalle had received remedial teaching for two or three hours a month from the second term of first grade. This had been delivered in a small group of three to five students who had various difficulties in reading and writing. A few occasional individual sessions with a special education teacher had also been provided. Despite this special attention, the students' writing skill development had been at a standstill since the second grade. During the intervention study, Ville and Kalle did not attend remedial classes.

Procedure

The students participated individually in 18 (Ville) and 21 (Kalle) weekly intervention sessions of 90 minutes, outside regular school hours, in the CLR. The interventions took place between September 1999 and April 2000. All sessions were videotaped. The first author, who was also responsible for planning the instructional procedures, carried out the training and data collection. She has several years' experience in cognitive strategy instruction as a clinical psychologist in the CLR. The first author and a trained research assistant scored spelling, spelling revision, composition, and decoding (see Instruments).

Baseline. Before the intervention, both students participated in the baseline measurement phase where decoding (accuracy and speed), spelling on a spelling test, and composition skills (spelling, coherence, and length) were measured once a week for five (Kalle) and eight (Ville) weeks without intervention. Spelling revision skills were assessed every other week. Furthermore, the students, their parents, and classroom teachers were interviewed before the intervention. The baseline phase started the same week for both students.

Training. Ville's intervention was started three weeks later than Kalle's. All intervention sessions consisted of a 30-minute assessment period and a 45-minute instruction period. Between the periods there was a break of about 15 minutes during which the student could drink some juice and discuss his school day of hobbies with the instructor. The assessment always preceded the instruction in order to keep the weekly measurements independent of the immediate instruction. In each of the assessment periods, decoding (accuracy and speed), spelling on a spelling test, and composition (spelling, coherence, and length) were assessed. The order of these tasks was altered for each successive session. In every other session, the result of the first phase of the computer game Aleksis[R]2 Virhejahti (in English: Aleksis[R]2 Error Chase) (Jokela, Kajamies, Kinnunen, Maki, & Poskiparta, 2000) was recorded. Ville's and Kalle's teachers and parents met the researcher/instructor in the CLR after the tenth session and were asked to apply defined procedures (see Instructional Procedures) in guiding the students' writing processes at home and in school.

Post-treatment measurements. During the first five weeks immediately following the termination of treatment, decoding (accuracy and speed), spelling on a spelling test, and composition skills (spelling, coherence, and length) were assessed weekly for both students. Spelling revision skills were assessed twice. The students, their parents, and classroom teachers were also interviewed.

Maintenance measurements. For both students, one maintenance assessment of decoding (speed and accuracy), spelling on the spelling test, composition skills (spelling, coherence, and length), and spelling revision took place six months after the termination of the treatment.

Transfer measurements. During the study, monthly samples of the students' spelling in school assignments were collected with their parents' consent and sent to the CLR by the classroom teachers.

Instructional Procedures

General instructional stages of the Self-Regulated Strategy Development (SRSD) (Harris & Graham, 1996, pp. 26-33) were applied as described. The intervention consisted of the following procedures (description of the detailed instructional procedures is available upon request from the first author).

1. Analysis of writing problems and goal setting (sessions 1-3). During the initial three sessions, background knowledge was developed about composition processes and the concept of self-monitoring (SRSD stage 1: develop background knowledge about the strategy). The instructor aimed to help the student to become aware of what was wrong with his writing. Both the process and the outcome were considered. The instructor invited the student to analyze his writings produced during the baseline phase. In addition, writing processes were analyzed from videotapes. On the basis of these discussions, so-called "list of my problems" and "writing instructions" were created on colorful cards (SRSD stage 2: discuss the strategy). The "list of my problems" included concrete remarks about the student's typical spelling errors or writing process flaws (e.g., "missing periods, forgot to read the text, missing double consonants'). The "writing instructions' consisted of a simplified list of necessary steps to be kept in mind while writing a short passage ("think what you want to say--translate it into written language--write it--read what you wrote--revise--evaluate the passage: did you manage to express what you planned"). The writing instructions were treated as an elementary composing strategy. Next, the instructor modeled how to utilize such checklists during writing (SRSD stage 3: model the strategy), and the student had some practice using them. The written checklists were used in every practice session thereafter.

2. Practice of the composing strategy (sessions 4-7). Writing short passages with a text processor practiced elements of planning, translating, and revising. The goal was to ensure that the student learned the role of the revising process during composing. The student created short stories or descriptions that were based on animal pictures. The instructor guided the student to utilize the checklists while writing (SRSD stage 5: support strategy use). Skillful and poor writing processes, as well as self-regulation with the help of the checklists, were discussed and modeled (SRSD stage 3: model the strategy). The student was repeatedly instructed to memorize the three writing processes as a composing strategy (SRSD stage 4: memorize the strategy). Instruction consisted of scaffolding procedures such as modeling, feeding-back, instructing, questioning, and cognitive structuring (Gallimore & Tharp, 1990; Roehler & Cantlon, 1997; Tharp & Gallimore, 1988).

3. Practice of spelling and spelling revision strategies (sessions 8-15). Spelling strategies were practiced more thoroughly than any other writing process. First of all, the student was taught the so-called "listen-segment-code" strategy. That is, he was encouraged to pronounce the word to be spelled either vocally or subvocally, to segment it into phonemes, and to mark each phoneme with the corresponding letter. Specific strategies were practiced for applying capitalization and punctuation rules, as well as for the n-phoneme (always write the letter "n" and then either "k" or "g," depending on whether you feel the phoneme in your throat or in your mouth) and double consonants (pronounce the word with one and with two consonants --decide which one sounds right). The teaching of these strategies again included developing background knowledge about the strategy, as well as discussing, modeling, and supporting memorization of the strategy. The application of the spelling strategies was practiced with paper-and-pencil tasks (SRSD stage 5: support strategy use). Spelling revision was taught by explaining that it is possible to read differently depending on the purpose, and that special attention must be given to decoding accuracy when revising one's own spelling by reading. Two rules were discussed for reading one's own text: "do not guess" and "suspect every word." Spelling revision and accurate decoding to do it were practiced with the Aleksis[R]2 Virhejahti computer game (Jokela et al., 2000). In every other session, the student worked with a text that included several types of spelling errors. In the remaining sessions, the student worked with texts that included the spelling errors that were most difficult for him to correct. Again, the instruction was based on the scaffolding procedures described above.

4. Practice of individual regulation of the writing processes (sessions 16-20). As the student showed mastery of spelling and spelling revision strategies, the instructor faded out her role as advisor, and encouraged the student to take more responsibility for his own writing process (SRSD stage 6: support independent performance). In the beginning, the instructor guided the student through the writing process by telling him what to do and think. Next, she helped him guide himself through the writing process by scaffolding. Gradually, the student was encouraged to perform individually. The student was advised to help himself through the writing process by repeating to himself the advice that had formerly been given by the instructor and by reading the checklists. This was practiced while writing short passages with a text processor. The instructor helped the student to realize what he had already learned and encouraged him to apply these skills independently.

5. Facilitation of transfer (sessions 11-20). Transfer of spelling skills from the intervention setting to school and the home environment was facilitated by involving the teachers and the parents in the rehabilitation process. They were informed about the aims of the intervention already during the baseline phase but were requested to refrain from modifying their usual instruction. The parents and the teachers were consulted again after 10 intervention sessions. They were then asked to (a) help the student stick the written checklists to their desks in school and at home, (b) help the student in writing tasks by asking questions that guide the student through the writing process (e.g., what should you do/think first), and (c) give feedback primarily on the writing process and only secondarily on the writing product. The parents and the teachers were given written instructions on these steps.

6. Feedback discussions (sessions 1-20). Immediate, concrete and continuous feedback was offered regularly. The student was encouraged to compare his performance and his outcomes to previous ones, and thus to evaluate his own progress. The instructor invited the students to compare their current writing process and products to the ones recorded in the beginning of the intervention. The observed changes were discussed.

Instruments

Decoding. The first 2,250 words from the Finnish translation of the book Le Petit Prince (de Saint-Exupery, 1944/1986) were randomly ordered in clusters of 75 words and used as 30 versions of the decoding test. In each version, the words were placed in lines so that they produced a text including words but with no meaning at the sentence or paragraph level. The students were asked to read the text aloud as accurately and as quickly as possible. Decoding performance was expressed as the total reading time and as the number of correctly read words.

Composition coherence. The students were given a series of four pictures (Der Kleine Herr Jakob, SCHUBI Lehrmittel GmbH) and asked to arrange them in order and to write a short story on the basis of them. They were asked to write as skillfully as possible, "by thinking about everything you have learned about writing." There were no time restrictions for this task. In the assessment period of intervention sessions 4-20, the students were able to see the written checklists that had been created together during sessions 1-3. However, no instructions were given on how to use the checklists. Composition coherence was assessed on an eight-point scale (for more details, see Maki et al., 2001). In our study, coherence is defined at the textual macro level as the organization of the text under a structural schema of script.

Composition length. All compositions were scored in terms of words written. The total number of words included all written words, regardless of spelling, that represented a spoken word. Compound words were counted as one word.

Spelling in composition task. All compositions were scored in terms of spelling. Missing, wrong or added letters, as well as punctuation and capitalization errors, were counted. The total number of these spelling errors was divided by the number of words in the story.

Spelling test. A test was designed that consisted of altogether 42 syllables that formed 13 words (nouns in nominative or genitive singular), five nonwords, and two sentences (formed with nine words). This list of words, nonwords, and sentences included five double vowels, five double consonants, five diphthongs, and three n-phonemes. Only one of these linguistic features could appear in one word. Altogether 30 parallel versions of this spelling test were prepared. These versions were matched for surface and lemma frequencies of the words, and for bigram frequency of the nonwords. All the frequencies were based on the Laine and Virtanen (1999) corpus. The experimenter read the items aloud one by one and the students were asked to write the words down as they heard them. In the assessment period of intervention sessions 4-20, the students were able to see the written checklists that had been created together during sessions 1-3. However, no instructions were given on how to use the checklists. Missing, wrong of added letters, as well as punctuation and capitalization errors were counted and divided by the number of words in the test.

Spelling in school assignment. The classroom teachers sent copies of the students' assignments (compositions or other productive writing tasks) by mail to the CLR. During the time that corresponded to intervention sessions 10-20, the students were able to see, in their classrooms, the written checklists that had been created together during sessions 1-3 in the CLR. However, no instructions were given on how to use the checklists. Missing, wrong or added letters, as well as punctuation and capitalization errors were counted and divided by the number of words in the writing sample.

Revision of spelling errors. A computer game, Aleksis[R]2 Virhejahti (in English: Aleksis[R]2 Error Chase), was designed (Jokela et al., 2000). This game includes 24 different texts (12 narrative and 12 expository with approx. 80-120 words) that have several versions with various spelling errors included. The task is to revise the text by seeking and correcting spelling errors. Students proceed through three stages where, if requested, gradually focusing help is provided by the program. During phase one, no help is available. The game gives points for correctly revised spelling errors. The score depends on the phase in which the student was able to find and correct the error: 20, 10 or 5 points, respectively, for phases one, two, and three. The game was used as one means of instruction, and the score of phase one in every second session was used in the analyses. In these sessions, we used text versions that included several spelling, punctuation, and capitalization errors and the student worked through the first phase without supervision.

Student interview. Ville and Kalle were interviewed three times about their attitudes towards writing, their knowledge about writing processes and spelling rules, and their understanding of their own writing problems. The interview consisted of six questions that were discussed during the first baseline session, during the first post-treatment session, and once again during the maintenance session (see Appendix). The interview was carried out after the students had completed the composition task.

Parent and teacher interviews. The students' parents and teachers were also interviewed separately in their own sessions. The interviews consisted of four questions that dealt with parents' and teachers' views of Ville's and Kalle's writing development, ways of instructing the boys in writing tasks, and the ways in which the requested intervention procedures had been applied at home and in school (see Appendix).

Reliability of measures. To assess rater agreement, 20% of the tasks measuring spelling, decoding, composition coherence, and composition length were scored by the instructor and a trained research assistant who was unaware of the purpose and design of the study. The following agreement percentages were obtained: composition coherence 85%, composition length 92%, spelling in a composition task 92%, spelling test 92%, spelling in a school assignment 73%, and decoding accuracy 90%.

Data Analysis

The data are described by means of measures of central tendencies and time-series figures with trend lines. The means and standard deviations were calculated on the basis of the last four sessions of the baseline and treatment phases in order to evaluate improvement at the termination of the instructional phase. The means and standard deviations of post-treatment measurements were also based on the last four sessions of the post-treatment phase. The index of change in behavior was defined as the effect-size d, with the formula d = ([M.sub.Treatment] - [M.sub.Baseline]) / S[D.sub.Baseline] (Kromrey & Foster-Johnson, 1996), and effect-size for correlated measures ([d.sub.c]), with the formula ([d.sub.c] = [t.sub.c][[2(1 - r) / n].sup.1/2] (see Dunlap, Cortina, Vaslow, & Burke, 1996). With the latter formula, the d is corrected for the correlation between the measures in baseline and training phases. Without this correction, the effect sizes might be overestimated, because the correlation between baseline and end-of-treatment measures reduces the standard error of the difference between the means (Dunlap et al., 1996). Effect sizes were calculated for both participants separately on all dependent variables that were continuous and had more than two measuring points during baseline, end-of-treatment, and post-treatment phases. Cohen's (1988) definition of the magnitude of the effect-size was used as follows: .20 in absolute value is a small, .60 is a moderate, and .80 is a large effect-size.

RESULTS

Spelling

Both students made many spelling errors in the spelling test and in the composition task during the baseline phase (see Figures 1 and 2). In the spelling test, the students spelled wrongly, on average, every second word, and in the composition task, every third word. Kalle's spelling errors per word from baseline to the end-of-treatment phase decreased by an average of 92% in the spelling test, and by 81%) in the composition task (see Table 2). Ville's spelling errors decreased on average by 74% in the spelling test, and by 23% in the composition task. The effect sizes indicated large intervention effects in spelling behavior between baseline and the end-of-treatment phase (see Table 2), with one exception. The effect on Ville's spelling in the composition task was only moderate. For both students, the effect sizes were higher for the spelling test than for the composition task. The standard deviations show that Kalle's spelling performance stabilized during the intervention.

[FIGURES 1-2 OMITTED]

According to the post-treatment measurements, both students' results on the spelling test had remained at about the end-of-treatment level (see Figures 1 and 2). In the composition task, Kalle's spelling errors increased slightly, but still remained far below baseline level. However, Ville's spelling errors in the composition task returned to baseline level during the posttreatment phase. Six months after the termination of treatment, both students' results on the spelling test had remained at the end-of-treatment level. In the composition task, Kalle still made about 50%) fewer spelling errors than at baseline.

A decreasing trend in spelling errors was also found in both students' transfer measurements from baseline to end-of-treatment phase (see Table 2). Kalle's spelling errors in school assignments dropped by 33%, and Ville's spelling errors by 53%. Kalle's spelling errors in school assignments seemed to further decrease from the end-of-treatment to the post-treatment phase, whereas Ville's spelling errors remained at end-of-treatment level.

Decoding

Decoding accuracy. At baseline, Ville decoded wrongly, on average, 19%, and Kalle 14% of the words in the decoding test. A gain in decoding accuracy was found for both students from baseline to the end-of-treatment phase (see Figure 3). Ville's wrongly decoded words decreased by 62%, and Kalle's by 35%. Kalle's decoding accuracy seemed to stabilize during the latter half of the treatment phase. Thus, the effect sizes indicated large intervention effects for both students in decoding accuracy (see Table 2). In the posttreatment and maintenance measurements, both students' decoding accuracy remained at about the end-of-treatment level.

[FIGURE 3 OMITTED]

Decoding speed. At baseline, both students' average word decoding time was at about the average level of Finnish third graders (estimated from the word reading subtest of LUKILASSE by Hayrinen, Serenius-Sirve, & Korkman, 1999). Ville decoded words too quickly as reflected in his error rates. Ville's decoding speed decreased by, on average, 18% (from on average 1.04 to 1.23 seconds per word), whereas Kalle's decoding speed increased by 6% (from on average 1.66 to 1.55 seconds per word) from baseline to end-of-treatment phase (see Table 2 and Figure 4). The effect size was large for Ville, and only small for Kalle (see Table 2). Both students' decoding speed remained at end-of-treatment level in the post-treatment phase. Kalle's decoding speed further increased from the post-treatment phase to the maintenance measurement six months after termination of the treatment. Ville's decoding speed had returned to baseline level at the maintenance measurement.

[FIGURE 4 OMITTED]

Spelling Revision

At baseline, Ville corrected, on average, 16% and Kalle 43% of the spelling errors in the Aleksis 2 Virhejahti computer game (see Table 2). From baseline to end of treatment, the percentage of corrected spelling errors increased by 140% for Ville and by 69% for Kalle. The results remained at the end-of-treatment level during the post-treatment and six-month maintenance phases. On average, Kalle was able to correct more spelling errors than Ville during all phases of the study.

Composition Coherence and Length

No remarkable change occurred in scores on composition coherence from baseline to end of treatment, post-treatment, or maintenance (see Table 2). On average, Kalle's stories were somewhat more coherent than Ville's. Ville wrote either descriptions of the pictures with semantically connected propositions, of short simple stories with serially presented actions but no explicit background idea (scores 3 and 4 on a scale from 1 to 8). He wrote using mainly principal clauses. Kalle wrote many logical stories that had a fairly clear theme. Usually, some main idea was introduced in the beginning of the story and the end was connected explicitly to this idea (scores 5 and 6 on a scale from 1 to 8). He was also able to use subordinate clauses in his writings.

In composition length, there was a decreasing trend from baseline to the end-of-treatment phase for both students (see Table 2). Composition length remained at end-of-treatment level in the post-treatment phase, but returned to baseline level at the maintenance measurement six months after the termination of treatment. On average, Kalle wrote longer stories than Ville.

Student Interviews

As evidenced in the interviews, some positive changes in the students' attitudes towards writing, knowledge about writing processes and spelling rules, and understanding of their own writing problems seemed to have taken place during the intervention. The students seemed to like writing a little more and considered writing somewhat easier after the intervention. At baseline, neither of the students did any writing outside of schoolwork. However, after the intervention, Kalle had started to use writing more in his everyday life (for example, he wrote shopping lists and notes for his mother). Ville still avoided writing anything other than schoolwork.

At baseline, neither of the students could describe what was wrong with their writing. After the intervention, Kalle could repeat all the items and Ville some of the items in his "list of my problems." Both boys were able to verbally explain strategies for producing the n-phoneme in spelling and for deciding whether a word contained a double consonant or a double vowel. They were also able to apply these strategies in spelling when requested. Kalle especially had even started to apply these strategies independently in the spelling and composition tasks. Ville applied the spelling strategies occasionally, largely depending on his prevailing attention and interest.

In the baseline interview, when asked to describe his thoughts during a composition task, Kalle described the following steps: "I thought what I could write about, I wrote what is on the paper, I thought what could happen." Ville described how "I came up with one sentence for each picture, I tried to be careful, I thought I don't want to do this." Both students' knowledge of writing processes developed during the intervention. In the post-treatment interview, Kalle described the following steps: "I thought what to write, I wrote it, I was careful with spelling, I read the text and wondered if it was okay, I read syllable by syllable." Ville described how "I thought what to write, I wrote, I read it slowly and syllable by syllable if needed."

Six months after the termination of treatment, both students were still able to describe the steps of a writing process, and describe as well as apply the specific spelling strategies they had learned during the intervention.

Parent and Teacher Interviews

After consultation during the tenth intervention session, the parents and the teachers had applied the requested intervention procedures to a varying degree. The parents of both boys had helped their son stick the "writing instructions" and "list of my problems" to their desks at home. Guidance in using the instructions was given only a couple of times. The parents had not used process-oriented feedback.

Both teachers had decided to make individual problem lists for every student in their classroom. Furthermore, they had prepared writing instructions on a large board and hung it on the classroom wall for all the students to see. Kalle's teacher encouraged Kalle to guide the others on how to use such lists when they were writing. Both teachers encouraged their students a couple of times a week to use the checklists during writing. The teachers reported that, in the beginning, they had had difficulty finding time for discussion about the writing process and the checklists. Later, they felt that it became easier. Both teachers also reported that they could have utilized the checklists more regularly if requested. Process-oriented feedback was given only occasionally, since the teachers felt that it was difficult to observe one student in a large classroom so intensively that process-oriented feedback could be given.

In the post-treatment interview, both teachers reported some gain in the students' writing skills. Kalle especially had begun to be more active in writing tasks. He also developed a more positive attitude towards writing. In school he expressed pleasure when he did well in spelling tests. He reached the average level of spelling among his classmates. He also started to concentrate better in composition tasks, and tried to control his handwriting legibility. Ville continued to dislike writing, and his handwriting remained poor, but the teacher had noticed some reduction in spelling errors. Ville also seemed to concentrate better in some of the composition tasks at school. However, the teacher noted that attention problems continued to disrupt Ville's writing quite often.

DISCUSSION

The aim of this study was to test an intervention procedure designed to promote the spelling skills of elementary school students with severe writing difficulties. The intervention was based on strategy instruction with guidelines from Self-Regulated Strategy Development (Harris & Graham, 1996), accompanied by procedural facilitation, computer-assisted instruction, and steps to improve transfer to classroom learning.

What Improved?

Both students' spelling development had been almost nonexistent during the year before the intervention. Overall, the applied strategy instruction during the training phase of this intervention improved the students' spelling skills as measured by the spelling tests and composition tasks. The intervention effect showed more strongly in the spelling test. We argue that this is because the spelling test allowed the students to concentrate on translating more clearly than did the composition tasks. The latter also demanded planning and idea generation, as well as monitoring of sentence and discourse structure, processes that were not taught in this intervention. However, the students' spelling skills also improved in the composition task, indicating that spelling instruction without intensive instruction in the other writing subprocesses can be beneficial for students with writing disabilities. Further experiments are needed to show whether spelling skills, especially in composition tasks, may be further facilitated by providing instruction in the other writing subprocesses as well, thus reducing the cognitive load for spelling. This kind of multicomponent writing instruction has also been suggested by Berninger and others (1995), who reported positive effects on spelling of a writing intervention that combined instruction in low-level orthographic and phonological coding, as well as practice in high-level composition skills.

Our intervention also seemed to improve the students' spelling revision skills. Compared to baseline, both students were able to find and correct more preset spelling errors in the Aleksis 2 Virhejahti computer game after the intervention. In most cases, the students were able to correct an error if they found it. This finding is different from that of MacArthur, Graham, Haynes, and DeLaPaz (1996). They found middle-school students able to find about 30%, but able to correct only about 9% of spelling errors in their own compositions. This difference may be due to the differences in the spelling systems of the Finnish and English languages. Because in the Finnish language each phoneme is represented by one definitive letter (with only few exceptions), there is practically only one spelling for each word. As a result, students need not select the correct spelling among several possibilities, but can connect each phoneme with the corresponding letter using the listen-segment-code strategy.

In the present study, the students' difficulty in finding spelling errors in the text resulted at least partly from their inability to decode words accurately. Ville's decoding was also too hasty. Since we were able to demonstrate some improvement in decoding accuracy (and decrease in decoding speed for Ville) during the intervention, we can assume that more accurate decoding facilitated spelling revision to some degree. Furthermore, we assume that the developing spelling revision skill resulted in more accurate spelling in spelling tests and composition tasks. However, we are unable to make claims about a causal connection between decoding accuracy, spelling revision, and spelling on the basis of the current data. A longer follow-up period with more measurement points would allow the application of sophisticated time-series analysis, where the time-dependent systematic relationship between the scores of decoding accuracy, spelling revision, and spelling could be traced. It would also be important to collect more detailed data on how effectively students actually apply the "read what you wrote' strategy while they spell in the measurement tasks. Our students were frequently rather unwilling to read their own texts in the training sessions, paralleling the observations made by Graham and MacArthur (1988) in their study about improving LD students' essay revision skills. It may be that students with writing difficulties feel rather distressed about exploring their own texts because it means a growing awareness of their difficulties and a longer time spent on a, perhaps, disliked task.

We did not include instruction in planning, idea generation or revision of discourse structure in the intervention. Thus, as expected, there was no improvement in composition coherence. Composition length decreased, but returned to baseline level at the maintenance measurement six months after the termination of treatment. This most likely indicates that the students got tired of the demanding composition task that was repeated 30 times during the study. The fact that the quality of written compositions did not get better although the students learned to spell more effectively is important. Enhanced composition quality along with better spelling skills would have been a plausible expectation on the basis of previous assumptions claiming that nonautomated spelling could disrupt other writing processes (Berninger, 1999; Kellogg, 1996; McCutchen et al., 1995). Our present findings parallel those of Graham et al. (2001), who found that extra spelling instruction enhanced only writing fluency, not the quality of compositions among second-grade students.

Further studies are needed to conclude what level of automation in spelling is needed to achieve "more capacity" for other writing processes and, perhaps, enhanced writing quality. Furthermore, since students with writing disabilities often also have difficulties in self-regulation and other writing processes than transcription, mere spelling instruction may not be enough to bring about change in composition quality. What we need is a thorough knowledge of the writing-process flaws of subjects in spelling intervention studies. Then, it would be possible to determine more accurately what kind of writing difficulties can be facilitated with spelling instruction.

Who Benefited?

Overall, we were able to improve the spelling skills of both students. However, the results were more encouraging for Kalle than for Ville, who continued to have marked difficulties in spelling when measured by the composition task. This is most likely due to Ville's difficulties in attention and self-regulation during complex writing processes. Ville was able to concentrate on tasks that were highly structured and relatively quickly accomplished, such as the decoding tasks and spelling tests. However, he was only occasionally able to utilize his developing spelling strategies during the composition tasks.

Learning disabilities and emotional-behavioral problems tend to have high co-morbidity (e.g., Rock, Fessler, & Church, 1997). Mayes, Calhoun, and Crowell (2000) reported on their 8- to 16-year-old subjects that, of those diagnosed with attention deficit hyperactivity disorder (ADHD), 65% had disabilities in written expression, compared to 27% of those without ADHD. Correspondingly, the disability rate for spelling was 30% versus 6%. The mechanisms that connect emotional-behavioral and learning problems are not clear, however. Concurrent learning and behavior problems are assumed to multidirectionally affect several domains: cognitive processing, social-emotional adjustment, behavioral functioning, academic performance, language functioning, and executive functioning (Rock et al., 1997). Thus, students with LD and emotional-behavioral problems may face difficulties in accessing instruction, adapting to their inability, tolerating frustration during practice, and maintaining self-worth (Rock et al., 1997).

Did We Gain Long-Lasting Effects?

Overall, both students were able to maintain their results at the end-of-treatment level during the post-treatment phase. Only Ville's results in spelling in the composition task seemed to return, on average, to baseline level. We found it encouraging that the intervention effects still existed six months after the termination of treatment. On some measures, there was even slight progress during the six-month maintenance period. This progress was welcome, compared to the period of approximately one year before the intervention, when both students' writing skill development had been practically nonexistent.

We assume that this further progress resulted from several factors. The teachers may have continued to apply spelling strategy instruction to some degree even after the intervention study. Furthermore, when the students learned new spelling strategies, they were better equipped to practice spelling, and thus perhaps more motivated. A similar encouraging pattern of the persisting nature of tutorial assistance in spelling skills over a six-month follow-up was found in the study of Berninger and others (1995). They offered ongoing consultation for teachers and parents after the intervention, which probably supported the further spelling development of some of their subjects.

Did We Gain Transfer?

We aspired to facilitate transfer of spelling skills from the intervention setting to classroom performance by involving the teachers and the parents in the intervention process. The students were helped to understand that the spelling strategies they learned in the CLR were supposed also to be applied in other learning environments. Thus, on the basis of the spelling in their school assignments, both students also progressed in spelling outside the CLR. However, we are aware that a good intervention should be as parsimonious as possible. More experiments are needed to show whether transfer of spelling skills can be achieved without some of the steps applied here.

The students' parents were highly motivated to participate in helping their son. They had voluntarily applied for assessment and rehabilitation of their child, and volunteered to take part in the intervention study after they had been informed about their role as parent-tutors. Despite their motivation, however, they found it hard to apply the requested instructional steps at home. Although we made an effort to describe the steps in everyday language and to explain why their application would be beneficial, the parents found tutoring difficult. Therefore, we recommend that if parents are included, they be given plenty of opportunities to discuss their role as tutors. Furthermore, it might be beneficial to let parents and children practice the tutoring process together first in the intervention setting.

The classroom teachers and Kalle's special education teacher seemed to interpret their participation as an opportunity to learn. They were active in asking questions about the intervention steps, and both classroom teachers independently applied some of the spelling strategy instruction procedures to their classroom teaching even with their other students. The main problem seemed to be the allocation of time for one student with special needs. On the basis of these observations, it would be beneficial to design, together with teachers, how to apply the strategy instruction to whole-classroom teaching in such a way that the target student, as well as the other students in the class, can learn new strategies together.

Overall Evaluation of the Intervention Study

According to our preliminary findings, the general instructional procedures of Self-Regulated Strategy Development that have previously been applied to the instruction of composition processes (Harris & Graham, 1996) seemed effective also in the spelling instruction of young students with severe writing difficulties. This intervention gave us an excellent opportunity to closely analyze the complete learning processes in a given time window, and to detect possible reasons for the failure to benefit from instruction, as described above. However, the design demanded many measurement points, and this can lead to procedural and motivational problems. We were, however, able to handle the comparability of the test versions and the regularity of the measurement intervals. On the other hand, 30 parallel measurements led to lack of motivation in some tasks. Nevertheless, only the length of the compositions seemed to reflect this motivational problem. We assume that lack of motivation would lead to poorer overall performance, and, thus, work against the intervention effects.

We managed to promote several cognitive and metacognitive skills. However, as shown in many previous intervention studies, there is great individual variability in the response to treatment due to motivational, emotional and behavioral variables (e.g., Berninger et al., 1995; Vauras, Rauhanummi, Kinnunen, & Lepola, 1999). To be able to help students with motivational-emotional-behavioral learning problems, the intervention procedures presented in this study should be developed further. Procedures might include (a) creating subjectively more meaningful writing tasks, (b) allocating more time to discussions of self-concept as a writer, (c) developing more rewarding means of concrete feedback, and (d) inviting students to help in planning their own intervention. The latter could be achieved by means of goal setting based on the students' developing insight into the skilled writing process and knowledge about the subskills they still need to learn.

APPENDIX

The Interviews

Student Interview

1. How do you use writing in your everyday life? (Baseline and post-treatment)

2. How easy was this writing task for you (estimation on a four-point scale from very easy to extremely difficult)? (Baseline and post-treatment)

3. How much do you like such writing tasks (estimation on a four-point scale from very much to not at all)? (Baseline and post-treatment)

4. Describe what kinds of problems you have in writing? (Baseline, post-treatment and maintenance)

5. Think about the composing task you completed a few minutes ago. Describe how you did it. What did you think and do while you were writing? (Baseline, post-treatment and maintenance)

6. Describe the spelling rules for the n-phoneme, double consonant, and double vowel. (Baseline, post-treatment and maintenance)

Parent Interview

1. How would you describe your child's writing development? (In the post-treatment interview: How does your child write these days?) (Baseline and post-treatment)

2. How well does your child concentrate on writing tasks at home? (Baseline and post-treatment)

3. How have you tried to help your child with writing? (Baseline)

4. Describe how you managed, in tutoring your child, to apply the procedures we discussed in the consultation after 10 intervention sessions? (Post-treatment)

Teacher Interview

1. How would you describe Ville's/Kalle's writing development? (In the last interview: How does Ville/Kalle write these days?) (Baseline and post-treatment)

2. How well does Ville/Kalle concentrate on writing tasks at school? (Baseline and post-treatment)

3. Has Ville/Kalle received remedial teaching because of his writing difficulty? (Baseline and post-treatment)

4. Describe how you have managed, in teaching Ville/Kalle, to apply the procedures we discussed in the consultation after ten intervention sessions? (Post-treatment)

Table 1
Background Data on Students

                                               Ville   Kalle

Verbal IQ (a)                                   101     91
NEPSY (b)
  Oral fluency                                   10      8
  Phonological processing                         2      1
  Rapid naming                                   11     11
  Visual-motor accuracy                           1      4
  Phonological working memory for words           9      8
  Phonological working memory for sentences      12     11
ALLU (c)
  Decoding                                        3      2
  Reading comprehension                           4      5

(a) Wechsler Intelligence Scale for Children-Revised.

(b) NEPSY [A Neuropsychological Assessment for Children]. Standard
scores of the subtests (min 1, max 19).

(c) Finnish Standardized Reading Test for Grades 1-6. Reading skill
groups as follows: 1-3 poor, 4-6 average, and 7-9 skilled.
Table 2
Changes in Dependent Variables Between Baseline, Training,
Post-Treatment, and Maintenance Phases

Measure                        Baseline (d)      Training (d)

                                M       SD       M       SD

Ville
Decoding accuracy              14.50    1.29     5.50    2.65
Decoding speed                 78.25    8.54    92.50   10.25
Spelling test                    .50     .05      .13     .05
Spelling (composition task)      .30     .03      .23     .17
Spelling (school ass.) (a)       .38              .18
Composition coherence (b)       3.00     .00     3.50     .58
Composition length             25.75    4.57    17.50    4.36
Spelling revision (c)           2.50     .71     6.00    1.41

Kalle
Decoding accuracy              10.75    5.25     7.00     .82
Decoding speed                124.50   20.07   116.50    1.29
Spelling test                    .47     .25      .03     .02
Spelling (composition task)      .37     .15      .07     .08
Spelling (school ass.) (a)       .33     .22      .14
Composition coherence (b)       5.50    1.00     4.50     .58
Composition length             51.50   22.23    29.25    7.27
Spelling revision (c)           6.50    2.12    11.00    1.41

Measure                       Post-treatment (d)   Maintenance (e)

                                  M       SD            score

Ville
Decoding accuracy                 7.75    3.40            4.00
Decoding speed                   90.75   10.28           74.00
Spelling test                      .10     .05             .04
Spelling (composition task)        .29     .10             .26
Spelling (school ass.) (a)         .19
Composition coherence (b)         3.25    0.96            4.00
Composition length               17.25    5.32           27.00
Spelling revision (c)             5.00    1.41            7.00

Kalle
Decoding accuracy                 4.75     .96            6.00
Decoding speed                  116.50   14.27          109.00
Spelling test                      .06     .05             .04
Spelling (composition task)        .12     .08             .17
Spelling (school ass.) (a)
Composition coherence (b)         4.50    1.00            6.00
Composition length               32.25    9.74           59.00
Spelling revision (c)            11.50    0.71           10.00

Measure
                                d     d (c f)

Ville
Decoding accuracy             -6.98     -4.29
Decoding speed                 1.67      1.49
Spelling test                 -7.40     -7.41
Spelling (composition task)   -2.33      -.53
Spelling (school ass.) (a)
Composition coherence (b)
Composition length            -1.81      -.57
Spelling revision (c)

Kalle
Decoding accuracy              -.71      -.83
Decoding speed                 -.40      -.55
Spelling test                 -1.76     -1.79
Spelling (composition task)   -2.00     -1.48
Spelling (school ass.) (a)
Composition coherence (b)
Composition length            -1.00      -.56
Spelling revision (c)

Note. For decoding accuracy and spelling measures, a negative
effect-size indicates a positive change in skill.

(a) Only one measurement during the last four baseline, training,
and post-treatment sessions; effect-size not calculated.

(b) Measured in ordinal scale; effect-size not calculated.

(c) Only two measurements during the last four baseline, training,
and post-treatment sessions; effect-size not calculated.

(d) The last four measurements.

(e) Only one measurement.

(f) Effect-size for correlated measures based on the change in
behavior between the last four baseline and training measurements.

REFERENCES

Berninger, V. W. (1999). Coordinating transcription and text generation in working memory during composition: Automatic and constructive processes. Learning Disability Quarterly, 22, 99-112.

Berninger, V. W., Abbott, R. D., Whitaker, D., Sylvester, L., & Nolen, S. (1995). Integrating low- and high-level skills in instructional protocols for writing disabilities. Learning Disability Quarterly, 18, 293-309.

Bradley, L., & Bryant, P. (1983). Categorizing sounds and learning to read--a causal connection. Nature, 301, 419-421.

Brown, A. (1987). Metacognition, executive control, self-regulation, and other more mysterious mechanisms. In F. Weinert & R. Kluwe (Eds.), Metacognition, motivation and understanding (pp. 65-116). Hillsdale, NY: Erlbaum.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Darch, C., & Simpson, R. (1990). Effectiveness of visual imagery versus rule-based strategies in teaching spelling to learning disabled students. Research in Rural Education, 7, 61-70.

Dunlap, W. P., Cortina, J. M., Vaslow, J. B., & Burke, M. J. (1996). Meta-analysis of experiments with matched groups or repeated measures designs. Psychological Methods, 1, 170-177.

Englert, C. S. (1990). Unraveling the mysteries of writing through strategy instruction. In T. E. Scruggs & B.Y.L. Wong (Eds.), Intervention research in learning disabilities (pp. 186-223). New York: Springer-Verlag.

Englert, C. S., Raphael, T. E., Fear, K. L., & Anderson, L. (1988). Students' metacognitive knowledge about how to write informational texts. Learning Disability Quarterly, 11, 18-46.

Gallimore, R., & Tharp, R. (1990). Teaching mind in society: Teaching, schooling, and literate discourse. In L. C. Moll (Ed.), Vygotsky and education, Instructional implications and applications of sociohistorical psychology (pp. 175-205). Cambridge, UK: Cambridge University Press.

Graham, S. (1999). Handwriting and spelling instruction for students with learning disabilities: A review. Learning Disability Quarterly, 22, 78-98.

Graham, S., & Freeman, S. (1986). Strategy training and teacher-vs. student-controlled study conditions: Effects on LD students' spelling performance. Learning Disability Quarterly, 8, 267-274.

Graham, S., & Harris, K. (2000). The role of self-regulation and transcription skills in writing and writing development. Educational Psychologist, 35, 3-12.

Graham, S., Harris, K., & Fink Chorzempa, B. (2001). Is spelling causally related to learning to write and read? Treatment of spelling problems in beginning writers. Manuscript submitted for publication.

Graham, S., & MacArthur, C. (1988). Improving learning disabled students' skills at revising essays produced on a word processor: Self-instructional strategy training. Journal of Special Education, 22, 133-152.

Hacker, D. J., Plumb, C., Butterfield, E. C., Quathamer, D., & Heineken, E. (1994). Text revision: Detection and correction of errors. Journal of Educational Psychology, 86, 65-78.

Harris, K. R., & Graham, S. (1996). Making the writing process work: Strategies for composition and self-regulation. Cambridge, MA: Brookline Books.

Harris, K. R., & Graham, S. (1999). Programmatic intervention research: Illustrations from the evolution of self-regulated strategy development. Learning Disability Quarterly, 22, 251-262.

Hayes, J. (1996). A new framework for understanding cognition and affect in writing. In C. M. Levy & S. Ransdell (Eds.), The science of writing. Theories, methods, individual differences, and applications (pp. 1-27). Mahwah, NJ: Erlbaum.

Hayes, J., & Flower, L. (1980). Identifying the organization of writing processes. In L. Gregg & E. R. Steinberg (Eds.), Cognitive processes in writing (pp. 3-30). Hillsdale, NJ: Erlbaum.

Hayes, J., Flower, L., Schriver, K., Stratman, J., & Carey, L. (1987). Cognitive processes in revision. In S. Rosenberg (Ed.), Advances in applied psycholinguistics: Vol. 2 (pp. 176-240). Cambridge, UK: Cambridge University Press.

Henry, M., Calfee, R., & LaSalle, R. (1989). A structural approach to decoding and spelling. In S. McCormick & J. Zutell (Eds.), Cognitive and social perspectives for literacy research and instruction (pp. 155-163). Chicago: National Reading Conference.

Hayrinen, T., Serenius-Sirve, S., & Korkman, M. (1999). LUKILASSE [A test of decoding, reading comprehension, spelling, and basic mathematical skills]. Helsinki: Psykologien Kustannus Oy.

Jokela, J., Kajamies, A., Kinnunen, R., Maki, H., & Poskiparta, E. (2000). Aleksis[R]2 Virhejahti [Aleksis[R]2 Error Chase]. Centre for Learning Research, University of Turku, Finland.

Karlsson, F. (1983). Suomen kielen aanne-ja muotorakenne [Phonology and Morphology of Finnish]. Juva: WSOY.

Kellogg, R. T. (1996). A model of working memory in writing. In C. M. Levy & S. Ransdell (Eds.), The science of writing. Theories, methods, individual differences, and applications (pp. 57-71). Mahwah, NJ: Erlbaum.

Korkman, M. (1998). NEPSY--An adaptation of Luria's investigation for young children. The Clinical Neuropsychologist, 2, 375-392.

Korkman, M., Kirk, U., & Kemp, S. L. (1997). NEPSY--Lasten neuropsykologinen tutkimus [A Neuropsychological Assessment for Children]. Helsinki: Psykologien Kustannus Oy.

Kromrey, J. D., & Foster-Johnson, L. (1996). Determining the efficacy of intervention: The use of effect sizes for data analysis in single-subject research. Journal of Experimental Education, 65, 73-93.

Laine, M., & Virtanen, P. (1999). WordMill Lexical Search Program. Center for Cognitive Neuroscience, University of Turku, Finland.

Lindeman, J. (1998). Ala-asteen lukutesti ALLU [Reading test for elementary grades 1-6]. Centre for Learning Research, University of Turku, Finland.

MacArthur, C. A., Graham, S., Haynes, J. B., & DeLaPaz, S. (1996). Spelling checkers and students with learning disabilities: Performance comparisons and impact on spelling. Journal of Special Education, 30, 35-57.

Mayes, S. D., Calhoun, S. L., & Crowell, E. W. (2000). Learning disabilities and ADHD: Overlapping spectrum disorders. Journal of Learning Disabilities, 33, 417-424.

McCutchen, D., Covill, A., Hoyne, S. H., & Mildes, K. (1994). Individual differences in writing: Implications of translating fluency. Journal of Educational Psychology, 86, 256-266.

Maki, H., Voeten, M., Vauras, M. & Poskiparta, E. (2001). Predicting writing skill development with word recognition and preschool readiness skills. Reading and Writing: An Interdisciplinary Journal, 14, 643-672.

Pressley, M., McGoldrick, J. A., Cariglia-Bull, T., & Symons, S. (1995). Writing. In M. Pressley & V. Woloshyn (Eds.), Cognitive strategy instruction that really improves children's academic performance (pp. 153-183). Cambridge, MA: Brookline Books.

Rock, E. E., Fessler, M. A., & Church, R. P. (1997). The concomitance of learning disabilities and emotional/behavioural disorders: A conceptual model. Journal of Learning Disabilities, 30, 245-263.

Roehler, L. R., & Cantlon, D. (1997). Scaffolding: A powerful tool in social constructivist classrooms. In K. Hogan & M. Pressley (Eds.), Scaffolding student learning. Instructional approaches and issues (pp. 6-42). Cambridge, MA: Brookline Books.

de Saint-Exupery, C. (1986). Pikku prinssi [Le Petit Prince] (I. Packalen, Trans.). Porvoo: WSOY. (Original work published 1944)

Sandler, A. D., Watson, T. E., Footo, M., Levine, M. D., Coleman, W. L., & Hooper, S. R. (1992). Neurodevelopmental study of writing disorders in middle childhood. Journal of Developmental and Behavioral Pediatrics, 13, 17-23.

Schneider, W., & Naslund, J. C. (1993). The impact of early metalinguistic competencies and memory capacity on reading and spelling in elementary school: Results of the Munich Longitudinal Study on the Genesis of Individual Competencies (LOGIC). European Journal of Psychology of Education, 8, 273-287.

Snyder, B., & Pressley, M. (1995). Introduction to cognitive strategy instruction. In M. Pressley & V. Voloshyn (Eds.), Cognitive strategy instruction that really improves children's academic performance (pp. 1-18). Cambridge, MA: Brookline Books.

Tharp, R. G., & Gallimore, R. (1988). Rousing minds to life. Teaching, Learning, and schooling in social context. Cambridge, UK: Cambridge University Press.

Vauras, M., Rauhanummi, T., Kinnunen, R., & Lepola, J. (1999). Motivational vulnerability as a challenge for educational interventions. International Journal of Educational Research, 31, 515-531.

Wechsler, D. (1984). The Wechsler Intelligence Scale for Children-Revised. Helsinki: Psykologien Kustannus Oy.

Wong, B. (1986). A cognitive approach to teaching spelling. Exceptional Children, 53, 169-173.

NOTES

This study was supported by grants from the Niilo Maki Foundation and the Finnish Cultural Foundation to the first author. We wish to thank the two boys who participated in this intervention along with their parents and teachers. Unfortunately, they must remain anonymous. We also thank Dr. Marinus Voeten, University of Nijmegen, The Netherlands, for sharing his expertise on effect-size calculation.

Correspondence concerning this article should be addressed to: Hanna Maki, Centre for Learning Research, University of Turku, FIN-20014 Turku, Finland. Electronic mail may be sent to [e-mail].

HANNA S. MAKI, Lic. Psych., is a developmental and educational psychologist, Centre for Learning Research, University of Turku, Finland.

MARJA M. S. VAURAS, Ph.D., is professor in education, Department of Teacher Education, University of Turku, Finland.

SEPPO VAINIO, M.A., is researcher, Department of Psychology, University of Turku, Finland.