LANGUAGE-BASED SPELLING INSTRUCTION: TEACHING CHILDREN TO MAKE MULTIPLE CONNECTIONS BETWEEN SPOKEN AND WRITTEN WORDS.

Abstract. Two studies addressed issues related to multiple instructional components in early intervention for at-risk spellers learning to spell polysyllabic words. The first study was a follow-up to a prior second-grade intervention. The fast responders in that study, who were monitored at the beginning and end of third grade (n=61), maintained their earlier gains during third grade when treatment was withdrawn. Thirty-two of the slower responders received continuing tutoring (12 individual tutorials over 6 to 8 weeks in late fall of third grade), which showed that children who received only alphabet principle training did as well as those who received combined alphabet principle and syllable awareness training (syllable types in English), but that these children required 24 practice trials for short-term mastery of spelling specific words. The second study with a new sample of 48 third graders also evaluated the effectiveness of alphabet principle training only versus combined alphabet principle and syllable awareness training. In these 24 individual tutorials over a 4-month period beginning in the fifth month of third grade, the combined treatment was more effective for (a) spelling untrained transfer words, (b) spelling taught polysyllabic words with a final, silent e syllable, and (c) transfer to phonological awareness. A two-tier model for early intervention to prevent spelling disabilities is proposed. In the first tier alphabet principle is taught (along with other sound-spelling connections for words including syllable awareness) and applied to practice in spelling words singly and in text (teacher-directed dictation and child-generated composition). In the second tier children are monitored in the year following early intervention and continuing tutoring is provided if necessary.

Although there has been considerable research on early intervention to prevent reading disability (e.g., Foorman, Francis, Fletcher, Schatschneider, & Mehta, 1998; Torgesen, Wagner, & Rashotte, 1997; Vellutino, Scanlon, Sipay, Small, Pratt, Chen, & Denckla, 1996), there has been relatively little research on early intervention to prevent writing disability. This lack of research on early intervention for writing is problematic for three reasons. First, the degree to which transcription skills (handwriting and spelling) are developed is the best predictor of the amount and quality of written composition in the elementary grades (Graham, Berninger, Abbott, Abbott, & Whitaker, 1997). Yet, current educational practice minimizes explicit instruction and practice of transcription skills (Graham, 1999), thus placing children who start out at the low end of the continuum for spelling and/or handwriting skill at an even greater risk for developing significant problems in written expression later in schooling (Berninger, 1999; Berninger & Swanson, 1994). Second, although reading disability may be the most common form of learning disability early in schooling, many children overcome their reading disability only to struggle with persisting spelling problems. Little is known about how to prevent or remediate spelling disabilities compared to reading disabilities. Third, some students have no difficulty in learning to read but their relatively poor spelling skills are masked or ignored until requirements for written assignments increase, usually in the upper-elementary grades. By then, it may be more difficult to overcome the spelling problems than it would have been had the student received early intervention to prevent this kind of transcription problem in the first place.

Graham (1999, Table 1) summarized five instructional approaches that have been found to be effective in teaching spelling to students with learning disabilities. What all five approaches have in common is that they meet the criteria for a structured language, multisensory approach to word learning (e.g., Clark & Uhry, 1995; Moats, 1995). That is, in contrast to a widespread myth that learning to spell is just a matter of memorizing a sequence of letters, learning to spell requires that the mind's eye, ear, mouth, and hand learn to communicate in processing and producing written words. For example, Graham and Freeman's (1986) kinesthetic method emphasizes coordination between the mind's mouth, hand, and eye at the level of the whole word. Murphy, Hem, Williams and McLaughlin's (1990) copy-cover-compare method requires coordination between the mind's eye and hand at the level of the whole word. Bradley's (1981) simultaneous oral spelling method facilitates communication among the mind's ear, mouth, hand, and eye to operate at the whole-word level. Berninger, Abbott, Whitaker, Sylvester, and Nolen's (1995) visual-imaging method coordinates the mind's eye, mouth, ear, and hand at the whole-word level. Berninger, Vaughan et al.'s (1998) multiple-connections model teaches the mind's ear, mouth, and eye to talk to each other at different units of language: (a) phoneme to letter(s), (b) whole spoken to whole written word, and (c) spoken and written onset-rimes.

Table 1
Sample Characteristics

                                            Follow-Up Study

                                     Monitor Only   Continue Tutor
                                        n = 61          n = 32

Parents' Level of Education
less than high school
mom                                       0               3
dad                                       3               4

high school
mom                                      10               4
dad                                       6               8

community college
mom                                      14               8
dad                                       6               8

college
mom                                      25              14
dad                                      21               8

graduate school
mom                                      10               2
dad                                      22               3

not available
mom                                       2               1
dad                                       3               1

Ethnicity
Asian-American                            4               1
Black-American                            0               1
Caucasian                                54              24
Hispanic                                  1               1
Native American                           2               4
not available                             0               1

Family History of Spelling Problem
no history                               32              14
some history                             27              18
unknown                                   2               0

Gender of Participants
boys                                     32              20
girls                                    29              12

Handedness of Participants
right                                    50              25
left                                      9               5
unknown                                   2               2

WISC-III Verbal IQ of Participants
M                                       107.16          103.00
SD                                       12.10           12.56

Chronological Age (1,2)
M                                       106.15          106.16
SD                                        3.83            3.86

                                      New Study

Parents' Level of Education
less than high school
mom                                       2
dad                                       3

high school
mom                                      10
dad                                      15

community college
mom                                      11
dad                                       8

college
mom                                      19
dad                                      13

graduate school
mom                                       2
dad                                       4

not available
mom                                       3
dad                                       4

Ethnicity
Asian-American                            4
Black-American                            6
Caucasian                                30
Hispanic                                  2
Native American                           2
not available                             3

Family History of Spelling Problem
no history                               21
some history                             22
unknown                                   4

Gender of Participants
boys                                     25
girls                                    22

Handedness of Participants
right                                    41
left                                      6
unknown                                   0

WISC-III Verbal IQ of Participants
M                                      109.87
SD                                      14.38

Chronological Age (1,2)
M                                      105.06
SD                                       3.63

(1) as of fall of third grade.
(2) as of winter of third grade.

Of these methods of spelling instruction, only the multiple-connections approach includes explicit instruction in alphabet principle in isolation prior to combining it with whole-word and onset-rime feedback in practicing the spelling of specific words. Explicit training in alphabet principle has been found to be effective in preventing reading disability (e.g., Foorman et al., 1998; Torgesen et al., 1997; Vellutino et al., 1996), and is likely to be effective in preventing spelling disability as well because written English is an alphabetic system. Venezky (1970) analyzed the corpus of written English words and showed that alphabet principle plus a small set of additional rules such as gemination or doubling of certain letters at the end of words, such as s, l, and f, accounted for how English orthography represents English phonology. He also demonstrated that words fall along a continuum of spelling-sound predictability and cannot be categorized neatly as purely regular or irregular in correspondence between letters and phonemes. Even though written English words vary in degree of predictability, if teachers provide explicit instruction in the patterns of correspondence, beginning readers and writers can learn to read and spell words along a continuum of predictability (e.g., Abbott, Reed, Abbott, & Berninger, 1997; Berninger, Abbott et al., 1998; Berninger, Vaughan et al., 1998). In addition to alphabet principle, the internal structure of the syllable -- its onset phoneme(s) and rime (all of the syllable except the onset) -- may also play an important role in this predictability (Treiman, 1985).

The purpose of the research reported here is threefold. First, we report the results of a two-part follow-up study in which children who were given early intervention during second grade (Berninger, Vaughan et al., 1998) were followed longitudinally into third grade. Those who were fast responders during second grade were monitored in third grade when early intervention was withdrawn; we refer to these children as the Monitor Only Group. Those who were slower responders during second grade received continuing tutoring in third grade and were also monitored; we call these children the Continue Tutor Group. Second, we compare the relative effectiveness of alphabet principle training only and combined alphabet principle and syllable awareness training in both the follow-up study (Continue Tutor Group only) and in a new study. Per the logic of Graham and Harris (1989), we compared the effect of providing two instructional components versus the effectiveness of one of those instructional components alone. Third, given that prevailing educational practice has minimized systematic use of commercially available spelling programs that provide ongoing practice with specific words (see Graham, 1999), we wondered how important continued practice with specific words would be for at-risk beginning spellers. Therefore, for the children in the follow-up study who received continuing tutoring, we also investigated how much practice with specific words was needed for these words to become represented in long-term memory. Based on Dreyer, Luke, and Melican's (1995) report that poor spellers differ from good spellers not in learning rate but in long-term retention of written spellings, repeated practice in writing words from dictation might be needed to supplement alphabet principle training with or without syllable awareness training.

STUDY 1: FOLLOW-UP STUDY

Summary of Second-Grade Intervention

From January to the end of April, 1998, 128 second graders from eight schools, who had scored at the low end of the spelling continuum in the fall of second grade, participated in 24, twenty-minute, twice-a-week dyadic tutorial sessions. Children were randomly assigned to one of seven treatment conditions or a trained contact control. All children in the treatment group received explicit training in the alphabet principle out of word context during the first layer and composed during the third layer of the intervention protocol. During the second layer they were randomly assigned to treatments for explicit training in learning specific words using one of seven approaches to making connections between spoken and written words: whole word, letter(s)-phoneme, onset-rime, each two-way combination of the three single approaches, or the three-way combination of all approaches. The contact control received training in orthographic and phonological awareness but not in making connections between spoken and written words; they also composed. Following the explicit alphabet principle training, all seven treatments resulted in better word-specific training than did the contact control, but only one treatment led to better transfer to spelling untrained words yoked to the trained words: combined whole word and onset-rime. However, following explicit training in alphabet principle out of context (letter[s]-phoneme), only the treatment that modeled the alphabet principle explicitly in word context resulted in longer compositions and more correctly spelled words in the compositions. So different kinds of modeling may lead to transfer to spelling single words versus transfer to composition skills.

Overall, evidence suggested that instructional methods that teach multiple connections between different units of spoken and written language (letter[s]-phoneme, whole word, and onset-rime) are most likely to benefit the writing system in general. However, children varied in their rate of response during the tutorial. Although most children improved, half the sample reached grade level or better by the end of the tutorial, whereas other children had improved but were not yet at grade level. During third grade we monitored both the fast responders (Monitor Only Group) and the slower responders but also provided an additional 12 individual tutorial sessions during late fall in third grade for the slower responders (Continue Tutor Group).

Monitor Only Group

Participants. Of the 64 children who were at or above grade level at the end of the second-grade intervention, 61 parents granted permission for continued participation. Three children had moved and were no longer able to participate. Characteristics of this Monitoring Only Group are reported in Table 1. Over half of the parents had graduated from college. Although ethnically diverse, the majority were Caucasian. Slightly more than half were boys and slightly more than half had no history of spelling problems in the family. Their verbal intelligence was in the upper limits of the average range, above the national mean (see Table 1). At the beginning of the monitoring study the Monitor Only Group's spelling achievement was not statistically related to the treatment group they were in during the second-grade study.

Monitoring battery. Four measures were administered in October and May of third grade: the handwriting fluency task (Berninger et al., 1997), a composition fluency task (Berninger et al., 1995), the WRAT3 spelling subtest (Wilkinson, 1993), and the WIAT spelling subtest (Wechsler, 1992).

The first task required writing the alphabet letters from memory in lowercase manuscript and in order; the score is the number of correct letters produced in the first 15 seconds. The composition task required children to complete the story starter "One day (choose person) had the (choose best or worst) day in school" and to keep writing for 5 minutes. The scores on the composition were length of composition (number of words) or percentage of words spelled correctly. (See Graham et al., [1997] for additional details about the administration, scoring, and reliability of each of these tasks.) The WRAT3 and WIAT require writing dictated words; they were administered and scored according to instructions in the test manuals. The handwriting score is the number of correctly produced alphabet letters. The WRAT3 and WIAT spelling scores are standard scores for age on a scale with a mean of 100 and a standard deviation of 15. Composition length is the number of words written with a 5-minute time limit. Percent correct spelling is based on total words and total correctly spelled words in the compositions.

Results and discussion. See Table 2 for descriptive and inferential statistics. When handwriting, spelling (from dictation or in composition), or composition length was compared from the end of second grade to the end of third grade or from the beginning to the end of third grade, the fast responders during second grade not only maintained their levels of performance during third grade but also showed significant, relative gains by the end of third grade. Therefore, we conclude that they did not lose ground when early intervention was not provided during third grade. Our design does not permit us to evaluate how much of their skill maintenance or relative gains might be due to the second-grade intervention, the third-grade regular spelling program (which varied greatly across the sample), or self-teaching (Jorm & Share, 1983) once these children had some knowledge of alphabet principle. On both nationally normed measures of spelling, they were in the average range for age.

Table 2 Longitudinal Assessment of Writing Skills in the Monitor Only Group (n=61) and Continue Tutor Group (n=32)

                     End       2nd      Begin   3rd     End

                      M        SD         M      SD      M
Monitor Only

Handwriting          8.1       2.6       9.5    3.1    12.4
WRAT-3 Spelling      96.3      6.3      93.6    6.6    97.8
WIAT Spelling        96.2      7.4      93.8    6.8    98.3

Composition
Length               37.2     14.7      44.1    16.7   55.6

% Correct Spelling
in Composition       81.1      9.8      82.1    9.7    88.6

Continue Tutor

Handwriting          6.0       2.2       7.7    2.6     9.0
WRAT-3 Spelling      83.8      5.6      84.1    6.0    87.9
WIAT Spelling        85.1      5.7      84.9    5.8    86.1

Composition
Length               30.0     14.2      35.2    13.8   38.2

% Correct Spelling
in Composition       69.6     17.5      69.6    15.4   76.9

                     3rd    End 2nd to End 3rd  Begin to End 3rd

                      SD          t(60)=             t60)=
Monitor Only

Handwriting          4.6         -8.19(1)(***)      -5.26(2)(***)
WRAT-3 Spelling      6.1         -2.13(*)           -5.65(***)
WIAT Spelling        7.9         -2.40(***)         -6.07(***)

Composition
Length               19.3        -7.57(***)         -5.79(***)

% Correct Spelling
in Composition       8.0         -5.93(***)         -5.54(***)

Continue Tutor                     t(31)=             t(31)=

Handwriting          2.7         -4.77(**)          -2.17(3)(*)
WRAT-3 Spelling      5.4         -4.21(***)         -5.22(***)
WIAT Spelling        7.5         -0.9               -1.0

Composition
Length               21.2        -2.48(*)           -0.9

% Correct Spelling
in Composition       17.6        -2.47(*)           -2.43(*)

(1) df=59 because complete data missing for one child.

(2) df=58 because complete data missing for two children.

(3) df=30 because complete data missing for one child.

(*) = p [is less than or equal to] .05.

(**) = p [is less than or equal to] .01.

(***) = p [is less than or equal to] .001.

Continue Tutor Group

Participants and assignment to conditions. Parents of the 64 children who were still not at grade level after the early intervention for spelling in second grade, even though they had made progress during second grade, were invited to allow their children to continue receiving tutoring in third grade. Children of the first 32 parents who granted informed consent were given 12 additional individual tutoring sessions in late fall of third grade. Because of holidays and inservices, the tutoring was completed over a 6- to 8-week period. Characteristics of this sample are reported in Table 1. In contrast to the children in Study 1, fewer parents had graduated from college, relatively more boys than girls were affected, and relatively more children had some family history of spelling problems. Like the faster responders, the slower responders had a higher incidence of left-handers than the 8 to 10% incidence in the general population, and had verbal intelligence solidly in the average range.

At the beginning of the continuing tutoring, the Continue Tutor Group's spelling achievement was not statistically related to the treatment group they were in during the second grade study. These 32 children were randomly assigned to the contact control in layer 2 (the alphabet principle training only column in Table 3, n=16) or to the syllable awareness training in layer 2 (the combined alphabet principle and syllable awareness training column in Table 3, n=16). Each tutor was assigned to an equal number of children in each of these conditions. The two conditions did not differ significantly in WRAT3 spelling or WIAT spelling before the tutoring began.

Table 3 Instructional Protocol for Continue Tutor Group

                    Alphabet Principle Only

Layer 1 Subword
5 minutes           Talking letters-3 blocks
                    of phoneme-spelling
                    correspondence that alternate
                    for 4 repetitions
                    across 12 lessons

Layer 2 Word
5 minutes           Trained, contact
                    control-sound play
                    with polysyllabic words
5 minutes           Word practice
                    using combined
                    whole word and
                    onset-rime modeling
                    for 10 target words

Layer 3 Text
5 minutes           Dictation-controlled
                    repetition
                    of specific words

                    Combined Alphabet Principle and
                    Syllable Awareness

Layer 1 Subword
5 minutes           Talking letters-3 blocks
                    of phoneme-spelling
                    correspondence that alternate
                    for 4 repetitions
                    across 12 lessons

Layer 2 Word
5 minutes           Explicit syllable awareness-
                    classify 6 syllable types
                    in polysyllabic words
5 minutes           Word practice
                    using combined
                    whole word and
                    onset-rime modeling
                    for 10 target words

Layer 3 Text
5 minutes           Dictation-
                    controlled repetition
                    of specific words

Assessment measures. In addition to the handwriting, spelling, and composition measures in the monitoring battery that was given to both the Monitor Only and Continue Tutor Groups at the beginning and end of third grade, the following three tasks were given just to the Continue Tutor Group right after they completed the third-grade tutorial: (a) WIAT spelling, (b) a prepublication version of Wagner and Torgesen's (1999) phonological memory task, and (c) the Word Identification subtest of the Woodcock Reading Mastery Test-Revised (WRMT-R) (Woodcock, 1987). The second task, which requires children to repeat pseudowords pronounced by the examiner, was given because this process measure was the only one that did not show improvement after a prior intervention for older students with writing disability (Brooks, Vaughan, & Berninger, 1999), and we wondered if it would improve after early spelling intervention with younger children. The score was the number of spoken pseudowords correctly repeated. The third task, which requires children to read (pronounce) real words on a list, was administered and scored according to procedures in the test manual. Both of these tasks had also been given at the end of second grade, so we could evaluate whether any gains in spelling resulted in transfer to untrained phonological memory or word recognition. At issue was not whether our intervention caused gains in these skills but rather if gains in spelling (whatever their source) were accompanied by transfer to other skills that were not directly trained.

Intervention protocol. Table 3 summarizes the three layers of the intervention protocol for both treatment conditions. Appendix A contains the syllable types that were trained in layer 2, and Appendix B contains the 80 words that were used in the dictation activities in layer 3. These words were selected from lists of words primary-grade children use most frequently in their writing (Graham, Harris, & Loynachan, 1993, 1994).

In contrast to the second-grade intervention delivered to dyads, the third-grade intervention was always given to individual children. Also, whereas the spelling training in second grade only included monosyllabic words, the training in third grade included only polysyllabic words. Twelve 20-minute individual tutorials occurred twice a week over a 6- to 8-week period, beginning in late October or early November. For theoretical reasons related to maximizing transfer among components of a functional writing system (Berninger, 1998), sessions were structured into three layers, each aimed at a different level of language (see Table 3). For ecological validity considerations (instruction in real classrooms consists of multiple components and is not a single, pure method), multiple instructional components were provided, but were held constant, except for one systematic contrast between an instructional component of interest and a trained, contact control component in layer 2. In this way we combined features of both design experiments (Brown, 1992), which include all the necessary components to achieve an instructional goal, and controlled research.

For the first five minutes children reviewed the most common phoneme-spelling correspondences underlying the alphabet principle, which had been introduced in the second-grade tutorial (Berninger, Vaughan et al.,1998). For purposes of this continuing tutoring, these subword correspondences were reviewed in three blocks, which alternated across lessons for four reviews of each. The first block contained 29 spelling units in which a single letter corresponded to a vowel or a consonant phoneme. These corresponded to the easy order of spelling-sound predictability (e.g., ball -> /b/ -> b or apple -> /short a/ -> a). The second block contained 31 spelling units in which, for the most part, a one- or two-letter spelling unit corresponded to one vowel or consonant phoneme, but in some cases a two-letter spelling unit corresponded to two phonemes (e.g., blends). These corresponded to the moderate order of difficulty (e.g., boy -> /oi/ -> oy or clock ->/cl/-> cl). The third block contained 31 spelling units in which two-letter spelling units corresponded to a single vowel or consonant phoneme (e.g., r- and 1-controlled vowels and consonant digraphs). These corresponded to the difficult order of predictability (e.g., bird ->/er/-> ir or king ->/ng/-> ng). (For more information on the easy, moderate, and difficult orders, see Berninger, Abbott et al., 1998, or Berninger, Vaughan et al., 1998.) In the first six lessons, the tutor and child chanted the correspondences among the pictured word, target phonemes in pictured words, and the name of letter(s) in the spelling unit in parallel; in the last six lessons, the tutor pointed to the pictures as a prompt and asked the child to produce the target phoneme and name the associated spelling unit.

For the next five minutes, the alphabet principle only group had a trained, contact control instructional component and the combined alphabet and syllable awareness group had training in the syllable types of English. (see layer 2 of Table 3). In the contact control training, the tutor pronounced the word syllable by syllable but artificially with a 1-second pause between the syllables. The tutor pronounced each word in two ways, placing the accent either on the first or last syllable of the word. The child was asked to pronounce the word correctly with the accent on the correct syllable. The combined treatment focused on a different syllable type in each lesson, until all six syllable types were introduced (closed, open, silent e, vowel team, r-controlled, and -le). Thereafter, all syllable types were reviewed in each lesson. Children were given cards that illustrated the syllable types in English using visual symbols (C for consonant, V for vowel, see Appendix A) and asked to categorize each syllable in a set of 10 two-syllable words, which were presented visually, on the basis of these syllable types. In each lesson the same 10 words (drawn from the high-frequency, primary-grade words in Graham et al., 1993, 1994) were used for the contrasting trained, contact control and the syllable awareness training component and the common instructional component the two treatments shared in layer 2 (Table 3).

For the last five minutes of layer 2, both groups practiced the set of 10 two-syllable words in the same way (see layer 2, Table 3). Children in both conditions practiced these polysyllabic words in the way that led to the greatest transfer for monosyllabic words in the second grade study -- combined whole-word and onset-rime feedback -- but a writing component was added. First, the tutor named the word and pointed to and named each letter in left-to-right order. The child repeated the process, naming the word and then pointing to and naming each letter in it. Next, the tutor pronounced the word by onset and rime units in each syllable and wrote these units as they were pronounced. The child repeated the word by each onset and rime unit in each syllable. Then the word was covered and the child wrote it from memory.

Finally, for the last five minutes, the child spelled in writing a set of 15 words, which were first dictated for spelling in isolation and then dictated in sentences to spell in context (see Appendix B). The rationale for the sentence dictation was that spelling practice would be more likely to generalize to one's own composition if (a) words were retrieved in the context of meaningful sentences, with semantic as well as orthographic and phonological connections; and (b) the working memory load for sentence dictation is more like that for composing than that for spelling single words from dictation. If the child misspelled the target word either in isolation or sentence context, the tutor provided feedback about the correct spelling. During this third layer in Table 3, the amount of practice with each target word was systematically varied. Of the 80 words in Appendix B, 60 appeared in only one lesson, 15 appeared in only 4 lessons, and 5 appeared in all 12 lessons. Thus, in each lesson, there was a set of five words that appeared only once in the tutorial, a set of five words that alternated four times across blocks of three lessons, and a repeating set of five constant words. However, because each of these 15 words in layer 3 was written from dictation, first in isolation and then in sentence context, some target words were practiced 24 times, others were practiced 8 times, and yet others were practiced twice over the course of the 12-lesson tutorial.

The last five minutes of a session consisted of an authentic literacy activity. Children wrote for three minutes on a different topic each time and then read their composition to the tutor, who commented on what she or he liked about it. Topics for composition included Christopher Columbus, United States, Seattle, Baseball, Halloween, Autumn, Rainy Days, Candy, Movies, Thanksgiving, Balloons, and Laughing.

Fidelity of treatment implementation was monitored through initial training sessions, weekly meetings with tutors, audiotapes of sessions, and self-monitoring checklists, which tutors completed after each session as in Berninger, Vaughan et al. (1998). By the end of the third lesson, the first two authors determined that the tutors were implementing the treatment as designed with 100% accuracy. Because these tutors had participated in the earlier intervention, they were very familiar with the procedures.

Results and discussion. We differentiate between treatment-specific effects (statistically significant greater improvement in one treatment than the other) and treatment effects (statistically significant improvement over time for both treatments). On the one hand, there were no treatment-specific effects related to the layer 2 contrast between alphabet principle only training (contact control) or combined alphabet principle and syllable awareness training on any measure given at pretest and at posttest after the treatment ended. Thus there was no advantage of explicit training in the syllable types over and beyond the benefit of alphabet principle training only. (All other aspects of the intervention protocol in Table 3 had been held constant.)

However, there were treatment effects in that both groups improved significantly from the beginning to the end of the tutorial on each of the raw scores, which were analyzed instead of standard scores for age because the tutoring only lasted two months or less: WIAT spelling, F(1,30)=8.627, p=.0063; WRMT-R Word Identification raw scores, F(1,30)=101.582, p=.0001; and phonological memory, F(1,30)=11.941, p=.0017. WIAT spelling improved from a mean of 17.47 (SD=2.98) to a mean of 20.31 (SD=3.40). WRMT-R Word Identification raw scores improved from a mean of 40.53 (SD=9.99) to a mean of 49.41 (SD=8.27). Phonological memory improved from a mean of 7.22 (SD=3.56) to a mean of 8.66 (SD=3.23). Although neither phonological memory nor word recognition was directly trained, there appeared to be some transfer of improved writing skills to these writing-related skills. In contrast, Brooks et al. (1999), who also did not train phonological memory directly, did not find transfer from improved writing skills to either phonological memory or word recognition in students in the fourth through sixth grades. However, in that study syllable awareness was trained in a more traditional way with rules for segmenting written words into syllables (i.e., a visual strategy), in comparison to the Continue Tutor Group, which received training in syllable types and categorization of these types in spoken and written words (i.e., a linguistic strategy).

The most important finding for this short-term tutorial is the interaction between amount of word-specific practice and short-term retention of the taught words. A 2-way repeated ANOVA on time of testing (before or after continued tutoring) and number of repetitions (24x or 8x or 2x) of target words yielded a significant effect for sessions, F(1,30)=434.981, p=.0001; number of repetitions, F(2,60)=92.972, p=.0001; and interaction between sessions and repetitions, F(2,60)=95.015, p=.0001. The dependent measure was the inventory of taught words given at both pretest and posttest (see Appendix B). Overall accuracy was greatest for words practiced 24 times (M=60.94%, SD=39.60), next best for words practiced 8 times (M=49.56%, SD=27.08), and worst for words practiced only 2 times (M=30.18%, SD=I 5.43). Overall level of performance improved with practice, from 28.01% (SD=18.94) at pretest to 65.88% (SD=30.99) at posttest. At pretest words for 24x (M=32.52%, SD=18.08), 8x (M=25.87%, SD=13.36) and 2x (M=25.63%, SD=23.95) were not significantly different, but by posttest accuracy was dramatically different for words for 24x (M=96.25%, SD 9.42), 8x (M=66.46%, SD 24.03) and 2x (M=33.11%, SD=16.80) (see Figure 1 for a graphical portrayal of this interaction).

[Figure 1 ILLUSTRATION OMITTED]

The significance of this finding is best understood in the context of current instructional practice, which has downplayed the importance of not only explicit instruction in spelling but also repeated practice in writing words from dictation (see Graham, 1999). One rationale for minimizing the importance of repeated practice in writing words from dictation is that such practice does not generalize to spelling in composition. This assumption has not been subjected to the scrutiny of controlled research, however. Although the present research design does not allow us to examine this assumption directly, we do note that the Continue Group did improve from the beginning to the end of third grade in spelling in their composition (see Table 2). Our design does show that amount of practice in writing words from dictation was related to retention by the end of the tutorial for words written from dictation. We speculate that for children at risk for spelling problems or with learning disabilities, repeated practice in writing words from memory may be essential for their short-term retention of specific word spellings and that such practice will not impair transfer of spelling to written composition.

Although the fast responders (Monitor Only Group) and slower responders (Continue Tutoring Group) did not differ in Verbal IQ, t(91)=-1.56, p=.123, by the end of third grade, the groups differed on all measures in Table 2: Alphabet Task, t(90)=-3.75, p [is less than] .001; WRAT3 spelling, t(91)=-7.76, p [is less than] .001; WIAT spelling, t(91)=7.24, p [is less than] .001; length of composition, t(91)=-3.98, p [is less than] .001; and percentage correctly spelled words in composition, t(91)=-4.39, p [is less than] .001. Thus, the slower responders seemed relatively more impaired across the board in writing and not just in spelling -- but not in general verbal aptitude. The fast responders may just have needed a jump start or early boost to develop grade-appropriate spelling during second grade; they had been at risk but with early intervention did not look like they would have writing disabilities. The slower responders, in contrast, needed early intervention of longer duration and may have been at greater risk for writing disabilities.

Accordingly, we examined phonological, orthographic, and reading measures at the end of second grade for clues to the processing problems that might differentiate the fast and slower responders and possibly place the slower responders at greater risk for spelling disability. Although the groups did not differ on phonological segmentation (requiring that all the sounds in a word be pronounced separately), or in reported family history of spelling problems, they did differ on (a) the Wagner and Torgesen phonological memory task [t(91)=-2.17, p=.32], (b) the original Rosner phoneme deletion task [t(91)=-4.38, p [is less than] .001], (c) the handwriting fluency task [t(91)=-3.84, p [is less than] .001], (d) the letter-cluster coding task (requiring decisions about whether a multiletter unit was in a previously briefly presented word) [t(91)=-5.18, p [is less than] .001], (e) the orthographic choice task (requiring choice of the correct spelling for a word and pseudohomonym pair pronounced the same) [t(91)=-5.27, p [is less than] .001], and (f) WMRT-R Word Identification [t(91)=-5.81, p [is less than] .001] (see Berninger, Vaughan et al., 1998, for procedural details for all tasks). We then entered the measures on which they differed (phonological memory, phonological deletion, handwriting fluency, letter-cluster coding, orthographic choice, and word recognition [reading single words]) into a multiple regression that included children in both the Monitor Only and Continue Tutor Groups to identify those measures that uniquely predicted spelling achievement in children at risk for spelling problems. At the end of third grade, the regressions were all significant at p [is less than] .0001. Word recognition and handwriting fluency uniquely predicted achievement on both the WRAT3 (word recognition, t=3.928, p=.0002; handwriting fluency, t=2.280, p=.0251) and WIAT (word recognition, t=5.263, p [is less than] .0001; handwriting fluency, t=2.598, p=.0110) spelling subtests. Letter-cluster coding uniquely predicted spelling achievement on the WRAT3 spelling subtest (t=3.486, p=.0008).

Ability to read words orally appears to be a constraint in how easily a child learns to spell a word. Not only did word recognition consistently contribute unique variance to spelling achievement over and beyond its shared variance with handwriting in the multiple regressions, but the fast responders (Study 1) started out higher on the WRMT-R Word Identification subtest (M = 92.8, SD = 9.6) than the slower responders (M = 83.2, SD = 9.6). Likewise, Brooks et al. (1999) found that older students' progress in spelling during a writing tutorial was related to level of beginning reading achievement. Ability to retrieve alphabet letters from memory and produce them in writing quickly and accurately also appears to differentiate the fast and slower responders, illustrating that handwriting ability can exert an effect on spelling ability (Berninger, Abbott et al., 1998). Ability to encode multiletter spelling units may also play a role in spelling (Berninger, Yates, & Lester, 1991) because application of the alphabet principle depends on linking phonemes with spelling units that are often larger than a single letter (Berninger, Vaughan et al., 1998).

Taken together, these findings suggest the following. First, in planning a prevention or remedial spelling program for a student, reading as well as spelling skills should be assessed; students with both word recognition and spelling problems should receive explicit instruction and practice in both reading and spelling single words, whereas students with only spelling problems might respond to explicit instruction in spelling without a word recognition component. Second, handwriting automaticity training and looking games (Berninger, 1998) may help children develop letter-processing skills that facilitate spelling acquisition. Third, children might be given guided assistance in parsing written words into the spelling units that correspond to phonemes (Berninger, 1998; Venezky, personal communication, November 3, 1997).

STUDY 2: FURTHER COMPARISON OF ALPHABET PRINCIPLE ALONE TO COMBINED ALPHABET PRINCIPLE AND SYLLABLE TRAINING

Rationale

A second study was conducted to evaluate more precisely whether training in syllable awareness adds value over and beyond explicit training in alphabet principle alone. A new third-grade sample was recruited to address this issue without the confounding past history of treatment, which the Continue Tutor sample had already received. Also, the treatment in the second study was carried out for a longer duration -- about twice as long, and later in the school year.

Participants and Assignment to Conditions

The WRAT3 spelling and WIAT spelling subtests were group-administered to third-grade classrooms. To qualify, a child had to be, on average, at least 1/3 standard deviation below the mean in spelling (i.e., an average standard score of 95 or less on the WRAT3 and WIAT spelling subtests). Also, the classroom teacher had to confirm that selected participants had significant difficulty with spelling in the regular program (on tests and/or written assignments). A prorated WISC-3 Verbal IQ (information, similarities, vocabulary, and comprehension) was also given; a student had to have at least an 80 Verbal IQ to be included in the study.

Characteristics of the sample are reported in Table 1. The new sample of 48 third graders was comparable to both groups in the first study in age and verbal intelligence. Handedness was representative of the general population; both gender and absence or presence of a family history of spelling problem was about equally distributed. Again the sample was ethnically diverse but primarily Caucasian. Less than half the parents had graduated from college. Of the 48 children, who were recruited to create a balanced design of each of four tutors providing treatment for six children in each of two conditions, half (n = 24) were randomly assigned to each condition. Only 47 children completed the 24, 20-minute, twice-weekly tutorials between January and April of the school year; one child moved before he could be posttested. None of the children was receiving special education services.

Procedures

Assessment battery. The pretreatment battery consisted of the handwriting fluency task, the WRAT3 and WIAT spelling subtests, the inventory of taught words, the original Rosner test of phoneme and syllable detection (Rosner & Simon, 1971), phonological segmentation and memory (Wagner & Torgesen, 1999), letter-cluster coding (Berninger et al., 1991), orthographic choice (Olson, Kliegl, Davidson, & Foltz, 1985), and WRMT-R Word Identification. Of these measures, WRAT3 spelling, WIAT spelling, and the inventory of taught words were administered again at midtest after 12 tutoring sessions. The entire battery was read-ministered at posttest. The inventory was scored separately for total accuracy and for each syllable type (closed, open, vowel team, r-, controlled, le). In addition, a transfer test was constructed with 14 untrained words (two exemplars containing each of six syllable types plus the schwa syllable type). This test was given at midtest and posttest only.

Instructional protocol. Table 4 summarizes the instructional components in the intervention, which was, for theoretical reasons (Berninger, 1998), aimed at all levels of language -- subword (layer 1), word (layer 2), and text (layer 3). All instructional components were held constant for the two groups except that one group (column headed Alphabet Principle Only in Table 4) received only one kind of subword training whereas the other group (column headed Combined Alphabet Principle and Syllable Awareness in Table 4) received two kinds of subword training. In addition, during the word training the combined 24, 20-minute individual tutorials took place about twice a week, beginning in January and ending in late April or early May.

Table 4 Instructional Protocol for Second Study

                   Alphabet Principle Only

Layer 1 Subword
5 minutes          Talking letters-3 blocks
                   of phoneme-spelling
                   correspondence that alternate
                   for 8 repetitions across
                   24 lessons
2 minutes          Trained, contact control-
                   keyboard training

Layer 2 Word
3 minutes          Keyboard training
                   continues
5 minutes          Word practice using combined
                   whole word + phoneme-letter(s)
                   to model 10 target words

Layer 3 Text
5 minutes          Child-generated composition
                   on teacher-given topic

                   Combined Alphabet Principle
                   and Syllable Awareness

Layer 1 Subword
5 minutes          Talking letters-3 blocks
                   of phoneme-spelling
                   correspondence that alternate
                   for 8 repetitions across

2 minutes          24 lessons
                   Explicit training in syllable types
                   with 10 spoken words

Layer 2 Word
3 minutes          Explicit syllable awareness-classify 6
                   syllable types in 10 written words
5 minutes          Word practice using combined
                   whole word + phoneme-letter(s)
                   to model 10 target words

Layer 3 Text       Child-generated composition
5 minutes          on teacher-given topic

For the first five minutes children in both conditions were introduced to the subword phoneme-spelling correspondences underlying the alphabet principle using the same procedures as in the first study and in Berninger, Vaughan et al. (1998). The same blocks as used in the first study alternated every three lessons, for a total of eight repetitions of each block across the 24 lessons. For the first 12 lessons, the tutor modeled saying the name of the pictured word, sound of the target phoneme(s), and name of each letter in the spelling unit; and the child imitated the sequence. For the last 12 lessons, the tutor pointed to the picture, but the child produced the association of word name, target phoneme, and name of letters in spelling unit on his or her own; tutors provided assistance only as needed.

For the last two minutes of layer 1 and first three minutes of layer 2, children differed as to the treatment they received. While the students in the combined treatment began the instructional sequence below, the children in the alphabet principle only treatment received a contact control treatment of keyboard training (see Table 4), which is described later. To equate time for the two conditions, these two contrasting treatments continued into the first three minutes of layer 2 (see Table 4) for a total of five minutes for the systematic comparison. In the last two steps of the sequence, the combined condition introduced (first six lessons) a different one of the six syllable types (closed, open, final silent e/long vowel, le, r-controlled, or vowel team) or reviewed the syllable types already presented (last 18 lessons). The polysyllabic words used for this purpose were drawn from frequent words primary-grade children use in their writing (Graham et al., 1993, 1994).

Tutor: Says word.

Child: Says word syllable by syllable.

Tutor: Says first syllable in word.

Child: Places colored discs for each phoneme in the syllable on the desktop.

Tutor: Says second syllable in word.

Child: Places colored discs for each phoneme in the syllable on the desktop.

Tutor: Shows child card with written word segmented by syllables

Child: Uses cards with graphic cues for syllable types (see Appendix A) to classify each syllable in the word by type.

All the steps in this sequence were completed for each word before proceeding to the other words in the 10-word set in each lesson, which was also used for the word practice each treatment had in common during the second part of layer 2 (see Appendix C). Each word was recycled four times each in the 24 lessons. Although the focus in each lesson was on one type of syllable, children also classified other syllables in the word that varied in syllable type. Two additional categories were introduced, as needed, to prepare children for syllables with reduced vowels (schwas) and syllables best treated as morphemes (affixes, such as -ing, -ed, -s); these could also be used to classify syllables if appropriate.

Because spoken words can be segmented at different boundaries, depending on the timing of production, syllable boundaries in spoken words can be arbitrary and not match the visual segmentation of written words into syllables (Venezky, personal communication, November 3, 1997). For this reason, tutors were told to focus more on the number of syllables and the syllable type than on the exact boundary and to adjust the number of phonemes in the subsequent step as necessary. For example, depending on rate of production, the first syllable in table may end with/long a/or/b/. We avoided canned rules for visual segmentation of written words into syllables (e.g., vc/c or v/cv) because the resulting visual units (e.g., bun ny) do not always correspond to the aural units in the spoken word (e.g., bun y). This potential mismatch between aural and visual syllable segmentation can be confusing to beginning readers and spellers; besides, syllable segmentation for written words in the dictionary may not correspond to syllable boundaries in spoken English. Thus, the focus was on the relationship between vowels and consonants within syllables rather than on syllable boundaries, which can vary depending on temporal parameters of word production. However, we did show children each written word segmented by spoken syllable boundaries and asked them to classify the syllables by types of syllables in English (Appendix A).

The contact control condition did not receive any syllable awareness training. Instead, they received keyboard training. To equate time for the contact control and syllable awareness training, the contact control took place during the last two minutes of layer 1 and the first three minutes of layer 2. Three keyboard training games were played as time permitted using a laminated card with a graphic representation of each letter key in standard format on a keyboard. For letter find children touched the letter dictated by the tutor. For letter bingo, the child found the dictated letter and covered it with a square; the game ended when all letters in a row were covered. Finally, for letter concentration, all letters were covered with counters. The tutor dictated letters and the child picked up a counter where the letter was thought to be; if correct, the child kept the counter. The game proceeded until all keys or keys in a row were uncovered.

For the last five minutes of layer 2, both treatment groups practiced each of the same 10 words, which the combined group had already classified into syllables, in the same way. The tutor pointed to the word on a card that showed each spelling unit in a different color, named the word, and then each of its letters. The child repeated this process. The tutor produced the phoneme for each of the color-coded spelling units in a left-to-right sequence. The child repeated the sequence of phonemes. The tutor turned the card over and asked the child to spell it orally.

For the last five minutes (layer 3), children in both conditions used their spelling in composing at the text level. They were taught the strategy "What I think I can say; and what I can say, I can write" and how to use pictured cues for phoneme-spelling correspondence as taught in layer 1, if needed, to find a spelling unit for sounds in the words they said out loud while composing. Children wrote compositions on these topics: My School, My Teacher, My Principal, The School Bus, Recess, Valentines, My Family, My Name, My Favorite Toy, My Favorite Sport, My Favorite Food, Pets, Television, Physical Education, Art, Music, Lunch, Pizza, Ice Cream, My Best Friend, Rain, Spring, Weekends, and Vacation.

Fidelity of treatment implementation. A four-pronged approach was used to ensure that tutors, who were graduate students in educational psychology with school psychology specialization and prior teaching or tutoring experience, delivered the intervention in a standard, consistent manner. First, the different segments of the intervention were timed to ensure that tutors were devoting the same amount of time to each component of the treatment. Second, ongoing consultation with tutors at weekly group meetings was provided to monitor and discuss any problems that arose, either management or instructional, so that they would be handled the same way by all tutors. An instructional manual containing all the procedures was used for initial training and periodic review. Third, tutors were asked to review a self-monitoring checklist of procedures on a regular basis. Fourth, tutors audiotaped a session of each of the two treatments.

To assess fidelity of treatment, the audiotapes were reviewed by the research coordinator. The 10-item checklist assessed the tutor's exact use of procedures, stimulus materials, and time allotments for the different instructional segments of the session, as well as child-management techniques. A score of 10 on the checklist indicates perfect adherence to procedures in a well-managed session. The review indicated that tutors were implementing treatments in a standard fashion. All sessions received a score of 10.

Results and Discussion

Mixed ANOVAs, with repeated measures on sessions (either pretest, midtest, and posttest, or pretest and posttest) and treatments as a between-subjects variable, yielded no treatment-specific effects or treatment x intervention sessions for handwriting fluency, WRAT3 spelling, WIAT spelling, total accuracy on inventory of taught words, WRMT-R word identification, phonological segmentation, phonological memory, letter cluster coding, or orthographic choice.

The only main effects for treatment were for the transfer test and length of composition. Consistently at midtest and posttest the group that received combined alphabet principle and syllable training did better than those who received only alphabet principle training on the transfer test (see Table 5); this test was not given at pretest, and midtest scores probably reflect changes from pretest. Consistently at pretest and posttest the groups that received combined alphabet principle and syllable training wrote more than the group receiving only alphabet principle training. However, the combined group also wrote longer compositions at pretest, t(44)=-2.21, p [is less than].03, so we are reluctant to attribute this effect for composition to treatment.

Table 5 Treatment, Treatment x Session, and Session Effects in Second Study (N=48)

                                   Combined Alphabet Principle
                                   and Syllable Training (n=24)
Treatment Effects

Transfer Test(a)
F(1,45)=13.22, p=.001
                                   midtest    M=12.43    SD=1.67
                                   posttest   M=12.65    SD=1.37

Composition Length
F(1,44)=9.85, p=.003
                                   pretest    M=36.05    SD=14.19
                                   posttest   M=57.96    SD=17.91

Treatment x Session Effects

Silent e/long vowel syllables(b)
F(2,90)=3.64, p=.030
                                   pretest    M 3.48     SD 2.15
                                   midtest    M=5.70     SD=1.82
                                   posttest   M=7.09     SD=1.93

Phoneme deletion(c)
F(1,45)=7.70, p=.008
                                   pretest    M=18.96    SD=5.86
                                   posttest   M=22.39    SD=5.56

Syllable deletion(d)
F(1,45)=14.66, p= .059
                                   pretest    M=4.41     SD=2.13
                                   posttest   M=6.18     SD=2.32

                                   Alphabet Principle Training Only
                                   (n=24)

Treatment Effects

Transfer Test(a)
F(1,45)=13.22, p=.001
                                   midtest    M=10.54    SD=2.28
                                   posttest   M=11.38    SD=1.71

Composition Length
F(1,44)=9.85, p=.003
                                   pretest    M=27.29    SD=12.62
                                   posttest   M=43.17    SD=16.58

Treatment x Session Effects

Silent e/long vowel syllables(b)
F(2,90)=3.64, p=.030
                                   pretest    M 3.79     SD 2.19
                                   midtest    M=5.08     SD=2.26
                                   posttest   M=5.91     SD=2.41

Phoneme deletion(c)
F(1,45)=7.70, p=.008
                                   pretest    M=17.83    SD=5.30
                                   posttest   M=17.87    SD=5.55

Syllable deletion(d)
F(1,45)=14.66, p=.059
                                   pretest    M=3.71     SD=1.97
                                   posttest   M=4.29     SD=2.07

                                                pretest
                                             SD         M

Session Effects

Handwriting Fluency(e)
F(1,43)=6.02, p=.018                        9.27       3.40

WIAT (raw)
F(2,90)=50.05, p [is less than] .001       19.43       3.26

WRAT3 (raw)
F (2,90)=28.63, p [is less than] .001      22.94       1.86

WRAT3 (standard scores)(f)
F(2,90)=14.2, p [is less than] .001        89.28       6.49

WI (raw)
F(1,45)=73.37, p [is less than] .001       55.57       8.69

Composition Length
F(1,44)=50.02, p [is less than].001        31.48      13.96

Phoneme Deletion(c)
F(1,45)=21.27, p=.001                      17.85       5.36

Syllable Deletion(d)
F(1,45)=14.66, p [is less than] .001        4.00       2.05

Phoneme Segmentation(g)
F(1,45)=61.30, p [is less than] .001        9.06       3.91

Phonological Memory(h)
F(1,45)=6.09, p=.017                       15.96       3.46

Letter-Cluster Coding(i)
F(1,45)=48.45, p [is less than] .001       80.10       9.94

Orthographic Choice(i)
F(1,45)=11.23, p=.002                      95.74       5.80

Target Words
Closed Syllables(b)
F(2,90)=49.59, p [is less than] .001        5.81       2.52

Open Syllables(b)
F(2,90)=69.29, p [is less than] .001        3.17       2.71

Vowel Team(b)
F(2,90)=86.12, p [is less than] .001        3.51       2.60

Silent e/long vowel(b)
F(2,90)=54.10, p [is less than] .001        5.21       2.06

R-controlled(b)
F(2,90)=82.91, p [is less than] .001        3.60       2.46

Final-le(b)
F(2,90)=58.01, p [is less than] .001        3.64       2.15

                                              midtest
                                           M          SD
Session Effects

Handwriting Fluency(e)
F(1,43)=6.02, p=.018

WIAT (raw)
F(2,90)=50.05, p [is less than] .001

WRAT3 (raw)
F (2,90)=28.63, p [is less than] .001

WRAT3 (standard scores)(f)
F(2,90)=14.2, p [is less than] .001

WI (raw)
F(1,45)=73.37, p [is less than] .001

Composition Length
F(1,44)=50.02, p [is less than].001

Phoneme Deletion(c)
F(1,45)=21.27, p=.001

Syllable Deletion(d)
F(1,45)=14.66, p [is less than] .001

Phoneme Segmentation(g)
F(1,45)=61.30, p [is less than] .001

Phonological Memory(h)
F(1,45)=6.09, p=.017

Letter-Cluster Coding(i)
F(1,45)=48.45, p [is less than] .001

Orthographic Choice(i)
F(1,45)=11.23, p=.002

Target Words
Closed Syllables(b)
F(2,90)=49.59, p [is less than] .001
                                            7.19       2.34
Open Syllables(b)
F(2,90)=69.29, p [is less than] .001
                                            5.30       3.08
Vowel Team(b)
F(2,90)=86.12, p [is less than] .001
                                            5.79       2.94
Silent e/long vowel(b)
F(2,90)=54.10, p [is less than] .001
                                            6.83       1.95
R-controlled(b)
F(2,90)=82.91, p [is less than] .001
                                            6.09       2.76
Final-le(b)
F(2,90)=58.01, p [is less than] .001
                                            5.38       2.06

                                               posttest
                                             SD         M
Session Effects

Handwriting Fluency(e)
F(1,43)=6.02, p=.018
                                           10.85       3.21
WIAT (raw)
F(2,90)=50.05, p [is less than] .001
                                           24.26       3.86
WRAT3 (raw)
F (2,90)=28.63, p [is less than] .001
                                           26.19       3.49
WRAT3 (standard scores)(f)
F(2,90)=14.2, p [is less than] .001
                                           94.74       8.79
WI (raw)
F(1,45)=73.37, p [is less than] .001
                                           59.43       8.08
Composition Length
F(1,44)=50.02, p [is less than].001
                                           50.40      18.62
Phoneme Deletion(c)
F(1,45)=21.27, p=.001
                                           20.64       5.91
Syllable Deletion(d)
F(1,45)=14.66, p [is less than] .001
                                            5.17       2.35
Phoneme Segmentation(g)
F(1,45)=61.30, p [is less than] .001
                                           15.00       4.87
Phonological Memory(h)
F(1,45)=6.09, p=.017
                                           17.11       3.55
Letter-Cluster Coding(i)
F(1,45)=48.45, p [is less than] .001
                                           87.45       8.04
Orthographic Choice(i)
F(1,45)=11.23, p=.002
                                           98.38       2.59
Target Words
Closed Syllables(b)
F(2,90)=49.59, p [is less than] .001
                                            8.15       1.89
Open Syllables(b)
F(2,90)=69.29, p [is less than] .001
                                            7.09       2.71
Vowel Team(b)
F(2,90)=86.12, p [is less than] .001
                                            6.98       2.68
Silent e/long vowel(b)
F(2,90)=54.10, p [is less than] .001
                                            7.77       1.90
R-controlled(b)
F(2,90)=82.91, p [is less than] .001
                                            6.85       2.75
Final-le(b)
F(2,90)=58.01, p [is less than] .001
                                            6.49       2.24

(a) Maximum score=14, (b) Maximum score=10, (c) Maximum score=20, (d) Maximum score=10, (e)Maximum score=26, (f) Standard score for age (M=100, SD=15), (g) Maximum score= 25, (h) Maximum score= 25, (i) Percentage correct.

The only treatment x session interactions, indicating treatment-specific effects, were for the silent e/long vowel syllables and for phoneme segmentation on the original Rosner. The interaction was marginally significant for syllable segmentation on the original Rosner (see Table 5). For each significant interaction, there was no difference between treatments at pretest, but by posttest the treatment that combined alphabet principle and syllable awareness training was superior. The explicit syllable training may have helped children differentiate the silent e syllables, which often drop the final e when adding an affix (e.g., saving) from an open syllable that also has a VCV pattern. Although only the combined treatment received explicit training in phonological awareness, both groups increased over the course of the tutorial in phonological awareness and memory skills (see treatment effects). Naming the pictured word and then the target phoneme in that spoken word during the alphabet principle training may have contributed to developing phonological awareness in both groups. However, the session x treatment interaction qualified this session effect for phonological skills, demonstrating that children who also received syllable awareness training developed phonological awareness to a greater extent than those who only received alphabet principle training.

There were many treatment effects in the form of significant session effects, indicating that both treatment groups improved: on alphabet fluency, WIAT spelling (raw), WRAT3 spelling (raw and standard scores), composition length, Word Identification (raw), phoneme deletion, syllable deletion, phoneme segmentation, phonological memory, letter-cluster coding, orthographic choice, and each of the syllable types on the inventory of taught words (see Table 5). Although word recognition was not directly trained, it appears that improved spelling may be accompanied by transfer to improved word recognition as well; however, this transfer is modest and occurred only for raw scores, not for standard scores for age. Processes related to spelling that were not directly trained in either treatment -- letter-cluster coding and representation of specific written words in memory -- improved over the course of the tutorial. The two-letter spelling units in the alphabet principle training may have improved letter cluster coding. Practice with taught words might have improved procedures for creating word specific representations in long-term memory.

In sum, children in both treatment groups improved in their handwriting, spelling, composition, orthographic, phonological, and word recognition skills. The advantage of the combined alphabet principle and syllable awareness training was demonstrated only for spelling a specific syllable type (final silent e syllable) in taught words, spelling untrained transfer words for all six syllable types, and phonological awareness.

GENERAL DISCUSSION

Proposed Two-Tier Early Intervention

Some children just need a boost or jump start to launch their spelling acquisition. This first-tier intervention should begin by second grade for children who seem to be getting off to a slow start in spelling. Those children who were brought to grade level with intervention during second grade and could be located in third grade not only maintained their relative gains but continued to make small relative gains on standardized tests with age norms after early intervention was terminated. At the beginning of second grade, these fast responders had been struggling primarily in spelling and not in word recognition.

However, other children may require more intensive spelling instruction of longer duration and need both first- and second-tier interventions. These slower-responding children also improved in spelling from the beginning of second grade to the end of third grade but took longer and made smaller gains. At the beginning of second-grade these children had been struggling in both word recognition and spelling. What differentiated them from the students who were fast responders to the second graded spelling intervention was not Verbal IQ or family history of spelling problems. Rather, the fast and slow responders differed in three skills that also contributed uniquely to spelling achievement for all the children who participated in the second-grade spelling intervention: word recognition in reading, fluency in producing alphabet letters, and accuracy in encoding multiletter spelling units in briefly presented written words. These findings suggest that early intervention for children with both word recognition and spelling problems might teach the alphabet principle in both directions, from spelling to phonemes and from phonemes to spelling, and not just in one direction as we did, from phoneme to spelling unit. In addition, these "slower-responding" children may also benefit from handwriting automatically training to improve fluency of retrieving and producing alphabet letters and practice in underlining the one- and two-letter spelling units in written words that correspond to the spelling units in the alphabet principle. Finally, they may benefit from daily practice with teacher-directed dictation of specific words (as in the first study) and with child-generated composition (as in the second study). Structured language, multisensory approaches emphasize the former and process writing emphasizes the latter. Both kinds of text writing may benefit at-risk spellers.

Regardless of whether children are fast responders during the first tier of intervention or require continued intervention in a second tier, it is essential to monitor spelling achievement after tutoring is terminated. Although the group of fast responders did not lose relative ground when tutoring was eliminated, some individual children may and continuing tutoring should be reinstated.

Multiple Connections Between Spoken and Written Words

Of the many ways children can forge connections between spoken and written words, the alphabet principle is powerful. For third graders given 12 additional tutorial sessions, adding syllable awareness training to alphabet principle training did not confer an advantage on any learning outcome. However, for third graders given 24 initial tutorial sessions, there was justification for adding syllable awareness training to alphabet principle training on three learning outcomes (target words with silent e pattern, transfer test, and phonological awareness). For this reason, the current version of Talking Letters that we use to train alphabet principle presents vowel units within the syllable types of English to help children integrate the alphabet principle and awareness of syllable types (Berninger, 1998). Although syllable training is often advocated as a necessary and important component of structured language, multisensory methods, there is little empirical research on the effectiveness of these methods in general (e.g., Clark & Uhry, 1995), or syllable training in particular. Future research should examine whether there is an optimal time and way to introduce syllable training into the instructional program for preventing and remediating spelling (and reading) problems. Future research should also address the nature of the syllable training provided. Brooks et al. (1999) did not find that rules for visual segmentation of words into syllables were effective, whereas Study 2 found that syllable awareness training with spoken words before syllable awareness training for types of syllables in written words can be effective.

Word-specific practice in writing words from dictation, singly and in sentence context, appears to be an important supplement to alphabet principle training. The slow responders given continuing tutoring in the first study only began to approach short-term mastery of spelling specific words when they had practiced writing them 24 times over a 6- to 8-week period (see Figure 1). Early intervention for prevention of spelling disabilities for these students will probably also require frequent practice in making connections between the spoken and written form of specific words. Incidental instruction during the "teachable moments," weekly spelling tests, and use of personal dictionaries alone are unlikely to be sufficient for these students.

Instead, we recommend that (a) a systematic spelling curriculum be used at all grade levels (for continuity of curriculum from year to year), with students placed at their instructional (not grade) level each year (see Moats, 1995; Morris, Blanton, Blanton, & Pernay, 1995; (b) daily writing from dictation, child-generated composing, and reading; and (c) explicit instruction and feedback at multiple levels (corresponding units of spoken and written words) -- especially for the alphabet principle (phoneme-letter[s]) but also for the whole word, onset-rime, and syllable structure types. The goal of syllable training should be to increase awareness of the syllable structure of spoken words and the syllable types of English rather than to apply canned rules for visual segmentation of written words into syllable segments as defined by the dictionary.

In conclusion, there are multiple ways to teach the mind's ear, mouth, eye, and hand to talk to each other to process and produce written words -- at the phoneme, word, syllable, and onset-rime levels. All may contribute to learning to spell (and read). The results of the research reported here for poor spellers in third grade suggest that alphabet principle alone can lead to gains in spelling achievement but that repeated practice in spelling dictated words in writing is also needed to reach short-term mastery levels for specific words, and explicit training in syllabic awareness may also be helpful and refine spelling skill. Although some initially low-achieving spellers respond quickly to early intervention for spelling, others respond more slowly and will need more sustained, long-term instructional intervention to optimize their writing achievement.

REFERENCES

Abbott, S., Reed, L., Abbott, R., & Berninger, V. (1997). Year-long balanced reading/writing tutorial: A design experiment used for dynamic assessment. Learning Disability Quarterly, 20, 249-263.

Berninger, V. (1998). Process assessment of the learner: Guides for intervention. Austin, TX: The Psychological Corporation.

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

Berninger, V., Abbott, R., Rogan, L., Reed, E., Abbott, S., Brooks, A., Vaughan, K., & Graham, S. (1998). Teaching spelling to children with specific learning disabilities: The mind's ear and eye beats the computer or pencil. Learning Disability Quarterly, 21, 106-122.

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

Berninger, V., & Swanson, H. L. (1994). Modifying Hayes & Flower's model of skilled writing to explain beginning and developing writing. In E. Butterfield (Ed.), Children's writing: Toward a process theory of development of skilled writing (pp. 57-81). Greenwich, CT: JAI Press.

Berninger, V., Vaughan, K., Abbott, R., Abbott, S., Brooks, A., Rogan, L., Reed, E., & Graham, S. (1997). Treatment of handwriting fluency problems in beginning writing: Transfer from handwriting to composition. Journal of Educational Psychology, 89, 652-666.

Berninger, V., Vaughan, K., Abbott, R., Brooks, A., Abbott, S., Reed, E., Rogan, L., & Graham, S. (1998). Early intervention for spelling problems: Teaching spelling units of varying size within a multiple connection framework. Journal of Educational Psychology, 90, 587-605.

Berninger, V., Yates, C., & Lester, K. (1991). Multiple orthographic codes in reading and writing acquisition. Reading and Writing: An Interdisciplinary Journal, 3, 115-149.

Brooks, A., Vaughan, K., & Berninger, V. (1999). Tutorial interventions for writing disabilities: Comparison of transcription and text generation processes. Learning Disability Quarterly, 22, 183-191.

Bradley, L. (1981). The organization of motor patterns for spelling: An effective remedial strategy for backward reading. Developmental Medicine and Child Neurology, 23, 83-91.

Brown, A. (1992). Design experiments: Theoretical and methodological challenges in creating complex interventions in classroom settings. The Journal of the Learning Sciences 2, 141-178.

Clark, D., & Uhry, J. (1995). Dyslexia: Theory and practice of remedial instruction (2nd ed.). Baltimore: York Press.

Dreyer, L., Luke, S., & Melican, E. (1995). Children's acquisition and retention of word spellings. In V. W. Berninger (Ed.), The varieties of orthographic knowledge: II. Relationships to phonology, reading, and writing (pp. 291-320). Dordrecht, The Netherlands: Kluwer Academic.

Foorman, B., Francis, D., Fletcher, J., Schatschneider, C., & Mehta, P. (1998). The role of instruction in learning to read: Preventing reading failure in at-risk children. Journal of Educational Psychology, 90, 37-55.

Graham, S. (1991). Handwriting and spelling instruction for students with learning disabilities: A review. Learning Disability Quarterly, 22, 78-98.

Graham, S., Berninger, V., Abbott, R., Abbott, S., & Whitaker, D. (1997). The role of mechanics in composing of elementary school students: A new methodological approach. Journal of Educational Psychology, 89, 170-182.

Graham, S., & Freeman, S. (1986). Strategy training and teacher-vs, student-controlled study conditions: Effects on learning disabled students' spelling performance. Learning Disability Quarterly, 9, 42-52.

Graham, S., & Harris, K. (1989). Component analysis of cognitive teaching strategy instruction: Effects on learning disabled students' compositions and self-efficacy. Journal of Educational Psychology, 81, 353-361.

Graham, S., Harris, K., & Loynachan, C. (1993). The basic spelling vocabulary list. Journal of Educational Research, 86, 363-368.

Graham, S., Harris, K., & Loynachan, C. (1994). The spelling for writing list. Journal of Learning Disabilities, 27, 210-214.

Jorm, A., & Share, D. (1983). Phonological recoding and reading acquisition. Applied Psycholinguistics, 4, 103-147.

Moats, L. (1995). Spelling: Development, disability, and instruction. Baltimore: York Press.

Murphy, J., Hem, C., Williams, R., & McLaughlin, T. (1990). The effects of the copy, cover, compare approach in increasing spelling accuracy with learning disabled students. Contemporary Educational Psychology, 15, 378-386.

Olson, R., Kliegl, R., Davidson, B., & Foltz, G. (1985). Individual and developmental differences in reading disability in G. E. MacKinnon & T. G. Waller (Eds.), Reading research: Advances in theory and practice (Vol. 4, pp. 1-64). New York: Academic Press.

Rosner, J., & Simon, D. (1971). The Auditory Analysis Test: An initial report. Journal of Learning Disabilities, 4, 40-48.

Torgesen, J., Wagner, R., & Rashotte, C. (1997). Prevention and remediation of severe reading disabilities: Keeping the end in mind. Scientific Studies of Reading, 1, 217-234.

Treiman, R. (1985). Onsets and rimes are units of spoken syllables: Evidence from children. Journal of Experimental Child Psychology, 39, 161-181.

Vellutino, F., Scanlon, D., Sipay, E., Small, S., Pratt, A., Chen, R., & Denckla, M. (1996). Cognitive profiles of difficult-to-remediate and readily remediated poor readers: Early intervention as a vehicle for distinguishing between cognitive and experiential deficits as basic causes of specific reading disability. Journal of Educational Psychology, 88, 601-638.

Venezky, R. (1970). The structure of English orthography. The Hague, The Netherlands: Mouton.

Wagner, R., & Torgesen, J. (1999). The Comprehensive Test of Phonological Awareness. Austin, TX: Pro-Ed.

Wechsler, D. (1992). Wechsler Individual Achievement Test. San Antonio, TX: The Psychological Corporation.

Wilkinson, G. (1993). Wide Range Achievement Test (3rd ed.). Wilmington, DE: Wide Range, Inc.

Woodcock, R. (1987). Woodcock Reading Mastery Test-Revised. Circle Pines, MN: American Guidance Service.

APPENDIX A
Syllable Types Trained in Layer 2

Each of the graphic representations for the syllable
types in English appeared on a separate 3x5 card.
Parentheses = optional

1. Closed syllable             (c) vc

2. Open syllable               (c) v
3. Silent e syllable           (c) vce
                               cvc + suffix
4. le syllable                 (c) le
5. r-controlled syllable       (c) vr (c)
6. vowel team                  (c) vv (c)
7. Schwa syllable              pictorial representation of above
                               (a bird flying above a tree)
8. affixes (morphemes)         ing
                               ed
                               s

APPENDIX B
Dictation Words for Continue Tutor Group in First Study (Layer 3)

        Words That Repeated in Each Lesson
color
because
rocket
spelling
funny

   Blocks of Five Words Words That Recycled Every Three Lessons
letter                      paper                          cabin
hello                       table                          along
happy                       window                         dishes
getting                     also                           birthday
airplane                    working                        having

           Words That Appeared in Only One Lesson
1                           5                             9
basket                      little                        without
better                      people                        sometime
children                    able                          everyone
cotton                      cattle                        anything
picture                     middle                        afternoon

2                           6                              10
begin                       winter                         already
later                       weather                        always
music                       water                          above
taken                       under                          above
lady                        teacher                        afraid

3                           7                              11
wagon                       flying                         early
seventh                     sitting                        family
never                       loving                         lovely
living                      making                         nearly
given                       running                        only

4                           8                              12
maybe                       busy                           kittens
below                       penny                          animals
fellow                      party                          inches
away                        story                          goes
today                       many                           hurried

APPENDIX C
Words Used for Syllable Awareness Training (Combined Treatment
Only) and for Spelling Training (Both Treatments) in Second Study,

Closed Syllables (Lessons 1, 7, 13, 19)
win                         =3 (closed)                    dow
bask                        =4 (closed)                    et
win                         =3 (closed)                    ter
un                          =2 (closed)                    der
hun                         =3 (closed)                    dred
un                          =2 (closed)                    til
sis                         =3 (closed)                    ter
num                         =3 (closed)                    ber
Sun                         =3 (closed)                    day
twen                        =4 (closed)                    ty

Open Syllable (Lesson 2, 8, 14, 20)
be                          =2 (open)                      gin
mu                          =3 (open)                      sic
la                          =2 (open)                      dy
be                          =2 (open)                      low
pa                          =2 (open)                      per
Fri                         =3 (open)                      day
al                          =2 (closed)                    so
may                         =2 (vowel team)                be
a                           =1 (schwa)                     go
pho                         =2 (open)                      to

Silent e (Lessons 3, 9, 15, 21)
air                         =1 (r-controlled)              plane
a                           =1 (schwa)                     lone
in                          =2 (closed)                    side
a                           =1 (schwa)                     while
ex                          =3 (closed)                    cuse
mak                         =3 (silent e)                  ing
brok                        =4 (silent e)                  en
tak                         =3 (silent e)                  en
lat                         =3 (silent e)                  er
writ                        =3 (silent e)                  ing

-le Syllable (Lessons 4, 10, 16, 22)
ta                          =2 (open)                      ble
peo                         =2 (vowel team)                ple
a                           =1 (open)                      ble
un                          =2 (closed)                    cle
sim                         =3 (closed)                    ple
cir                         =2 (r-controlled)              cle
ea                          =1 (vowel team)                gle
pur                         =2 (r-controlled)              ple
ma                          =2 (open)                      ple
can                         =3 (closed)                    dle

r-controlled (Lessons 5, 11, 17, 23)

bark                        =3 (r-controlled)              ing
larg                        =3 (r-controlled)              est
teach                       =3 (vowel team)                er
cov                         =3 (closed)                    er
birth                       =3 (r-controlled)              day
thirst                      =4 (r-controlled)              y
morn                        =3 (r-controlled)              ing
or                          =1 (r-controlled)              der
hurt                        =3 (r-controlled)              ing
re                          =2 (open)                      turn

Team Syllable (Lessons 6, 12, 18, 24)
paint                       =4 (vowel team)                ed
a                           =1 (schwa)                     round
al                          =2 (closed)                    ways
beach                       =3 (vowel team)                es
draw                        =3 (vowel team)                ing
flow                        =3 (vowel team)                er
coach                       =3 (vowel team)                ing
sleep                       =4 (vowel team)                er
en                          =2 (closed)                    joy
fool                        =3 (vowel team)                ish

Closed Syllables (Lessons 1, 7, 13, 19)
win                            =2(vowel team)
bask                           =2 (closed)
win                            =2(r-controlled)
un                             =2(r-controlled)
hun                            =4 (closed)
un                             =3 (closed)
sis                            =2(r-controlled)
num                            =2(r-controlled)
Sun                            =2(vowel team)
twen                           =2 (open)

Open Syllable (Lessons 2, 8, 14, 20)
be                             =3 (closed)
mu                             =3 (closed)
la                             =2 (open)
be                             =2 (vowel team)
pa                             =2 (r-controlled)
Fri                            =2 (vowel team)
al                             =2 (open)
may                            =2 (open)
a                              =2 (open)
pho                            =2 (open)

Silent (Lessons 3, 9, 15, 21)
air                            =4 (silent e)
a                              =3 (silent e)
in                             =3 (silent e)
a                              =3 (silent e)
ex                             =4 (silent e)
mak                            =2 (other)
brok                           =2 (closed)
tak                            =2 (closed)
lat                            =1 (r-controlled)
writ                           =2 (other)

-le Syllable (Lessons 4, 10, 16, 22)
ta                             =3 (-le)
peo                            =3 (-le)
a                              =3 (-le)
un                             =3 (-le)
sim                            =3 (-le)
cir                            =3 (-le)
ea                             =3 (-le)
pur                            =3 (-le)
ma                             =3 (-le)
can                            =3 (-le)

r-controlled (Lessons 5, 11, 17, 23)
bark                           =2 (other)
larg                           =3 (closed)
teach                          =1 (r-controlled)
cov                            =1 (r-controlled)
birth                          =2 (vowel team)
thirst                         =1 (open)
morn                           =2 (other)
or                             =2 (r-controlled)
hurt                           =2 (other)
re                             =3 (r-controlled)

Team Syllable (Lessons 6, 12, 18, 24)
paint                          =2 (closed)
a                              =4 (vowel team)
al                             =3 (vowel team)
beach                          =2 (closed)
draw                           =2 (other)
flow                           =1 (r-controlled)
coach                          =2 (other)

sleep                          =1 (r-controlled)
en                             =2 (vowel team)
fool                           =3 (closed)

(1) The syllable and phoneme segmentation was based on two raters, one of whom has had 20 years' experience in phonetic transcription and one of whom has had as much experience in teaching phonics sounds. The occasional differences were discussed until agreement was reached.

NOTES

This research was supported by National Institute of Child Health and Human Development Grants HD 25858-09 and P5033812-03.

We thank the participating children and staff at Briarcrest, Ridgecrest, and Meridian Park, Shoreline, Washington; at Enatai Elementary, Bellevue, Washington; at Island Park and West Mercer, Mercer Island, Washington; St. Luke's in the Archdiocese of Seattle, Washington; and Adams, Montlake, Sacajawea, and West Woodland, Seattle. We also thank anonymous reviewers for helpful comments on an earlier version and Deborah McCutchen for help with the interrater reliability for items in Appendix C.

Correspondence should be addressed to first author at: 322 Miller, Box 353600, University of Washington Seattle, WA 98195-3600. <[e-mail]>

VIRGINIA W. BERNINGER, Ph.D., is professor, Educational Psychology, University of Washington.

KATHERINE VAUGHAN, Ph.D., is research coordinator, Learning Disability Center, University of Washington.

ROBERT D. ABBOTT, Ph.D., is chair, Educational Psychology, University of Washington.

ALLISON BROOKS, M. Ed., is research assistant, Multidisciplinary Learning Disabilities Center, University of Washington.

KRISTIN BEGAY is a master's degree candidate, School Psychology, University of Washington.

GERALD CURTIN, M. Ed., is a school psychologist, Marysville Schools, Washington.

KRISTINA BYRD, M. Ed., is a school psychologist, Bellevue Schools, Washington.

STEVE GRAHAM, Ph.D., is professor, Special Education, University of Maryland.