Handbook of Learning Disabilities

278 EFFECTIVE INSTRUCTION

tional evidence from controlled comparative demographic and verbal IQ differences were
intervention studies has demonstrated that, statistically controlled. This constitutes a
with focused and systematic intervention, particularly important qualification to this
measurable progress in phonological read- study because random assignment to instruc-
ing skills can be achieved throughout the el- tional condition was not feasible. The syn-
ementary school years even with the most thetic phonics subjects did demonstrate
severely disabled readers of a clinical sample greater gains in phonological analysis ability
(Lovett & Steinbach, 1997; Lovett, Stein- relative to the other two groups, but this ad-
bach, & Frijters, 2000). vantage did not generalize to superior per-
formance on measures of word identification
Significant questions remain, particularly skill.
regarding the normalization of phonologi-
cally based and fluent reading skills in Reports by other researchers have also fo-
reading disabled individuals (Foorman & cused on deficient phonological processing
Torgesen, 2001; Torgesen, Rashotte, & and the phonologically based reading
Alexander, 2001; Wolf & Katzir-Cohen, deficits of older reading disabled children.
2001). Progress has been made, however, in Olson, Wise, and their colleagues (Olson et
identifying some of the ingredients of effec- al., 1997; Wise & Olson, 1995; Wise, Ring,
tive remediation. A recent empirical review & Olson, 2000; Wise et al., 1997) conduct-
concludes that effective reading interven- ed an intervention study in which they com-
tions with struggling readers must include bined features of an oral–motor program
“(a) phonological awareness training, (b) for training phonological awareness, read-
systematic phonics instruction that is linked ing, and spelling skills (the Lindamood
to spelling, and (c) oral reading practice ADD Program; Lindamood & Lindamood,
with decodable texts (i.e., texts that include 1975, cited in Torgesen et al., 1999) with
only words using the accumulating set of their own computer-based reading training
letter–sound correspondences that have program (Reading with Orthographic and
been taught” (Rayner et al., 2001, p. 45). Speech Support, or ROSS). Two groups of
These conclusions emphasize the necessity children with reading disabilities from
of direct remediation of phonological grades 2 through 5 received phonological
awareness deficits, systematic and explicit decoding and digitized speech support to
instruction in letter- and letter cluster– help them read unknown words in story
sound mappings, and reinforcement of reading on the computer; with this training
word identification learning through ample experience, one group received additional
text reading practice using controlled de- phonological awareness training using the
codable reading vocabulary. ADD oral–motor methods and the other
group received extra training in reading
Within the past decade, there have been comprehension strategies. There was no
mixed results and a range of outcomes from control group in this study. The group that
remediation studies with readers who are received extra phonological awareness (PA
more severely disabled and with older chil- or ADD) training was superior in phonolog-
dren. Foorman and colleagues (1997) com- ical awareness and phonological decoding
pared the relative efficacy of three remedia- skill both right after training and at 1-year
tion programs for second- and third-graders follow-up testing. The PA group’s superiori-
with identified reading disorders; the pro- ty on phonological measures did not, how-
grams differed in instructional approach and ever, result in superior word recognition
in the size of the print-to-sound unit empha- performance at 1- or 2-year follow-up as-
sized in reading instruction. In analytic sessments. These negative findings 1 and 2
phonics, words were segmented into onset years after remediation were contrary to the
and rime units, in synthetic phonics into let- expectation that the PA group’s advantage
ter–sound units, and in sight word ap- in phonological awareness and nonword de-
proaches instructed as whole words. After coding skills would be associated with
the instruction, the synthetic phonics group greater word recognition development over
appeared superior to the other two groups time (Olson et al., 1997; Share, 1995). Wise
on testing, but their treatment advantage and colleagues (2000) concluded that “this
was not confirmed when the influences of

Word Identification and Decoding Difficulties 279

amount of this type of (PA) training has not ing. . . . We believe that future research
been sufficient for children to transfer the should include longer training times and in-
improved phonological skills to indepen- vestigate supplementary methods to increase
dent accurate reading compared to less in- automaticity and application, to help children
tensive phonologically supported instruc- to transfer improved phonological skills to
tion” (p. 202). Individual differences were similarly improved reading even after training
apparent however: Younger, lower-perform- has ceased.
ing readers were found to gain more from
intervention and to benefit more from Generalization in Disabled Readers’
phonological training instruction than were Response to Remediation: Components
older, higher-performing children (Wise et of Effective Instruction for Children
al., 2000). with Reading Disabilities

Generalization of phonological process- Torgesen and colleagues (1997) have sug-
ing gains and improvements in nonword de- gested that the generalization problem in
coding skills to other aspects of reading skill these intervention studies may reflect in part
development has posed a hurdle for which the complexity of the processing deficits
relatively few remediation programs have seen in readers with more severe disabilities.
proved effective. Despite the gains observed In fact, it is now acknowledged that the core
in children’s phonologically based word at- processing impairments of many children
tack and decoding skills, many investigators with reading disabilities extend beyond the
have found that these often sizable gains do realm of phonological awareness to other
not reliably generalize to other dimensions domains of function. Wolf and Bowers have
of reading development. Children who, af- identified the deficit in naming speed as po-
ter intervention, could “sound out” new tentially causal to developmental reading
words or nonwords were not always im- disability (Bowers & Wolf, 1993; Wolf,
proved relative to other instructional groups 1991; Wolf & Bowers, 1999, 2000). Both
in their word recognition, text reading, or phonological awareness and naming speed
reading comprehension skills. Generaliza- deficits impede reading acquisition at the
tion failures have been reported for many lexical level and the existence of both
intervention methods evaluated in the litera- deficits in combination are known to consti-
ture (e.g., Olson et al., 1997; Torgesen et tute a risk factor for more severe forms of
al., 1997) and for interventions that we as- reading disability than demonstration of ei-
sessed in our early remediation studies at ther deficit separately (Lovett, Steinbach, &
The Hospital for Sick Children in Toronto Frijters, 2000; Wolf, Bowers, & Biddle,
(Lovett, Ransby, Hardwick, Johns, & Don- 2000).
aldson, 1989; Lovett, Warren-Chaplin,
Ransby, & Borden, 1990). As Moats and Many of the generalization failures expe-
Foorman (1997) note in their summary of rienced by children with reading disabilities
several notable intervention studies with could also be attributed to a more general
positive findings, difficulty with acquiring effective, flexible
word identification strategies and with
These gains are to be celebrated although . . . monitoring and evaluating the effectiveness
generalization and transfer of decoding profi- of their strategic efforts. Difficulties with
ciency to fluent word recognition and better specific aspects of executive functioning and
reading comprehension was not automatic strategy learning appear to exist in readers
and constitutes a next phase of remediation with learning disabilities independent of
that needs . . . additional study. (p. 188) their phonological processing difficulties
(Levin, 1990; Swanson, 1999a; Swanson &
Wise and colleagues (2000, p. 231) con- Alexander, 1997). Recent research has also
clude their report with a parallel observa- highlighted the specificity of the transfer-of-
tion: learning deficit, however: Transfer-of-learn-
ing problems for children with reading dis-
Researchers must try to specify how best to abilities appear specific to printed language
achieve transfer from improved phonolog- learning and are not evident on other learn-
ical skills to greater long-term growth in read-

280 EFFECTIVE INSTRUCTION

ing tasks with similar cognitive demands 1986). WIST adapts the Benchmark Pro-
but no phonological processing require- gram’s dialogue structure for strategy in-
ments (Benson, 2000; Benson, Lovett, & struction, their keywords, and their Com-
Kroeber, 1997). pare/Contrast strategy (a strategy of “Word
Identification by Analogy”). WIST differs
Our clinical research program at The from the Benchmark Program in its inclusion
Hospital for Sick Children in Toronto has a of three additional word identification
long history of specific interest in general- strategies (“Vowel Variation”—trying vari-
ization and transfer-of-learning questions. ant vowel pronunciations in order of the fre-
Difficulties in achieving generalization of in- quency with which they occur in spoken
tervention gains have motivated much of English; “Peeling Off” prefixes and suffixes
our recent remediation research with severe- in a multisyllabic word; and “Spy”spying the
ly reading disabled children referred to our part(s) of a word you already know), its di-
laboratory classrooms. We have described rect training focus on the subskills necessary
research in which children were randomly for strategy implementation, and its provi-
assigned to one of two remedial reading sion of a metacognitive “Game Plan” to
programs or to an active control treatment train flexibility in strategy choice and evalu-
that worked on helping them acquire better ation of the success of those choices.
study, organizational, and problem-solving
skills (CSS for Classroom Survival Skills Both PHAB/DI and WIST recognize the
Program; Lovett et al., 1994; Lovett & need for subsyllabic segmentation during
Steinbach, 1997; Lovett, Steinbach, & Fri- word identification learning and its impor-
jters, 2000). Both of the reading interven- tance to attaining transfer of learning to
tions targeted the problem of generalization uninstructed words (Lovett, 1991; Lovett et
of instructional gains in word identification al., 1990). The PHAB/DI and WIST pro-
learning, but they addressed this problem grams, however, work on subsyllabic seg-
with quite different remedial approaches mentation using subword units of different
and at different levels of print-to-sound seg- size: PHAB/DI emphasizes the smallest
mentation. spelling-to-sound units (letter–sound) and
WIST focuses on recognition of larger sub-
The PHAB/DI (Phonological Analysis and syllabic units, particularly the rime. Both re-
Blending/Direct Instruction) Program con- medial programs attempt to promote gener-
sisted of lessons from the direct instruction alization of word identification skills in
decoding programs developed by Engelmann different ways—PHAB/DI through intensive
and his colleagues at the University of Ore- remediation of core sound analysis and
gon; these programs train phonological blending deficits in the context of systematic
analysis, phonological blending, and let- decoding training, WIST by teaching a set
ter–sound association skills in the context of of flexible and effective word identification
intensive systematic word recognition and strategies and the specific skills and content
decoding instruction (see Reading Mastery required to implement them successfully.
I/II Fast Cycle, and Corrective Reading Pro- Every lesson of the PHAB/DI and WIST
gram: Engelmann & Bruner, 1988; Engel- programs includes practice using new word
mann, Carnine, & Johnson, 1988; Engel- identification skills in context—by reading
mann et al., 1988). The other remedial connected text with controlled vocabulary
reading program was called WIST (Word and at carefully selected levels of difficulty.
Identification Strategy Training). WIST has a
strong metacognitive focus, instructs In a controlled research design, both the
through a teacher-led dialogue, and teaches PHAB/DI and the WIST programs proved
the children how to use and monitor the ap- far more effective than our previous inter-
plication of four metacognitive decoding vention programs (Lovett, Ransby, & Bar-
strategies. The WIST program was devel- ron, 1988; Lovett et al., 1989, 1990) from
oped in our laboratory classrooms at The the perspective of achieving generalization
Hospital for Sick Children and is based in of remedial gains: PHAB/DI- and WIST-
part on the original Benchmark School Word trained children were reliably improved on
Identification/Vocabulary Development Pro- several standardized and experimental test
gram developed by Irene Gaskins and her measures and demonstrated significant gen-
colleagues (Gaskins, Downer, & Gaskins, eralization on word reading measures in-

Word Identification and Decoding Difficulties 281

cluding transfer probes varying in their dis- In a subsequent study, we used a sequen-
tance from instructed target words included tial crossover design to address the question
in the PHAB/DI and WIST lessons (Lovett of whether phonologically based remedia-
et al., 1994; Lovett & Steinbach, 1997; tion is sufficient to achieve the best remedial
Lovett, Steinbach, & Frijters, 2000). Al- outcomes for children with severe reading
though both programs were associated with disabilities, or whether a combination of
large positive effects, different patterns of phonological and strategy training ap-
transfer were observed following the two proaches would produce superior outcomes
programs on some measures, confirming the (Lovett, Lacerenza, Borden, et al., 2000).
existence of some treatment-specific effects. The efficacy of a combination of the
The phonological program, PHAB/DI, was PHAB/DI and WIST programs was com-
associated with broader-based and deeper pared to that of longer-term intervention
generalization specifically within the phono- with each approach separately. Eighty-five
logical skill domain (speech- and print- children, 7–13 years of age and severely
based), and the WIST program, with its reading disabled, were randomly assigned
strategy training focus, resulted in broader- to 70 hours of remedial instruction in one of
based generalization for real English words five program sequences: PHAB/DI Ǟ WIST;
(i.e., generalization was observed on regular WIST Ǟ PHAB/DI; PHAB/DI × 2; WIST ×
and exception words). 2; or CSS Ǟ MATH (Classroom Survival
Skills Ǟ Mathematics, a control treatment
Three findings from this line of research offering study skills and then math instruc-
were particularly encouraging: (1) the tion). Each child’s skills were assessed at
demonstration of generalization within the five time points: before, at three points dur-
domain of word identification learning, (2) ing, and following 70 hours of intervention;
the finding that the phonological skills and this testing schedule allowed a closer exami-
decoding performance of children with se- nation of the time course of remedial gains
vere reading disabilities could be improved and their generalization within the different
with focused intensive remediation of this instructional conditions. Generalized treat-
type, and (3) that positive effects were ment effects were demonstrated on stan-
achieved even with later (e.g., grades 5–6) in- dardized measures of word recognition,
tervention (Lovett & Steinbach, 1997) and passage comprehension, and nonword read-
even for the most disabled children of a very ing, confirming the effectiveness of the pre-
reading disabled sample (Lovett, Steinbach, sent instructional programs on multiple in-
& Frijters, 2000). The phonological process- dices of reading skill acquisition.
ing and reading deficits of these children
were not ameliorated after this short-term The most critical findings of this research
intervention, but both print- and speech- were the demonstration of superior out-
based phonological skills were reliably im- comes and steeper learning curves for those
proved and moved closer to age-appropriate children who had received a sequential com-
expectations. After only 35 hours of instruc- bination of the phonological and strategy-
tion, PHAB/DI and WIST graduates typical- based instruction. A combination of
ly were not reading at grade level; however, PHAB/DI and WIST proved superior to ei-
they demonstrated markedly improved let- ther intervention alone on measures of
ter–sound knowledge, better decoding abili- phonological reading skill (nonword read-
ties, and more accurate word identification ing), tests of letter–sound and keyword
skills. Before their remedial interventions, knowledge, and three word identification
these children with severe reading disabilities measures (two of near and far transfer
were incorrectly identifying one-syllable words and the third of uninstructed multi-
words such as way, left, and put: After syllabic challenge words). These results pro-
PHAB/DI or WIST instruction, many of vide evidence of the separate instructional
these children were able to decode accurate- contributions of both the phonological and
ly (although often slowly) challenging multi- the strategy-based methods. More impor-
syllabic words such as unintelligible, mistak- tant, these findings demonstrate that gener-
enly, and disengaged. (These examples are alization from nonword decoding to other
taken from the test protocols of a 10-year- reading measures can be best achieved with
old girl with a reading disability.) a combination of effective remedial compo-

Was there transfer to uninstructed words?
(bake Ǟ babe, fake, bike, baker; grab Ǟ graft, stab, grub . . .)

(a)
Was there transfer to uninstructed words?

Treatment vs. Control Contrast

(b)
Was there transfer to uninstructed words?
Combined vs. Single Treatment Contrast

(c)
FIGURE 17.1a–c. Learning curves for disabled readers receiving a sequential combination of phonolog-
ical and strategy-based instruction (PHAB/DI Ǟ WIST; WIST Ǟ PHAB/DI); phonological (PHAB/DI) or
strategy-based instruction alone (WIST); or a control treatment program (CSS Ǟ MATH). From Lovett,
Lacerenza, Borden, Frijters, et al. (2000). Copyright 2000 by the American Psychological Association.
Reprinted by permission.

Word Identification and Decoding Difficulties 283

nents. Figure 17.1 summarizes representa- for struggling readers. Called the PHAST
tive learning curves and average group out- Track Reading Program (PHAST for
comes on a transfer test of word identifica- Phonological and Strategy Training), this
tion skill. new program begins with PHAB/DI’s pro-
gram of phonological and letter–sound
The results of this intervention research training and uses it as a framework on
suggest that phonologically based ap- which each of the four WIST strategies are
proaches alone are not sufficient for achiev- introduced and scaffolded. A detailed de-
ing optimal remedial outcomes with indi- scription of the instructional design of the
viduals with reading disabilities. While PHAST Track Reading Program, of its com-
systematic phonologically based, deficit- ponent instructional parts (PHAB/DI and
directed interventions appear necessary to WIST), of the dialogue structure used for
achieve gains, generalization of those gains the acquisition and monitoring of each
is more probable if a multidimensional ap- word identification strategy, and of sample
proach to core reading-related deficits is lesson materials developed for each pro-
used. Our findings indicate that faster learn- gram section has been published (Lovett,
ing and superior outcomes are attained Lacerenza, & Borden, 2000).
when a broader-based intervention ap-
proach is adopted—particularly one com- This part of the PHAST program covers
bining direct and dialogue-based instruc- lessons 1–70 and focuses on the develop-
tion, explicitly teaching children different ment of the five PHAST word identification
levels of subsyllabic segmentation, and strategies and acquisition of the prerequisite
training them in the acquisition and effec- skills and knowledge needed to implement
tive use of multiple decoding strategies. The them successfully. The five PHAST word
central importance of strategy instruction identification strategies are sounding out
and the promotion of a flexible approach to (the left-to-right letter–sound decoding
word identification and text reading tasks strategy of PHAB/DI), rhyming (word iden-
cannot be overemphasized when it comes to tification by analogy, using a bank of
achieving generalization of remedial gains. known keywords to decode an unknown
word with the same spelling pattern), peel-
Conclusions from this research are com- ing off (stripping prefixes and suffixes from
patible with recent findings from a meta- a multisyllabic word to get a smaller root
analysis of the treatment outcome literature word which can then be decoded using one
on learning disabilities: Swanson and of the other strategies), vowel alert (trying
Hoskyn (1998) conducted a rigorous review variable vowel sounds in the order in which
of 180 intervention studies; they concluded they occur in spoken English to see which
that the optimal approach in instructing pronunciation gives a word the child
children with learning disabilities was a knows), and I spy (seeking smaller parts of a
combined intervention model that included word that the child already knows). Every
both direct instruction and strategy instruc- lesson includes some instructional time ac-
tion methods. Additional work by Swanson quiring the skills and knowledge to execute
(1999b) has focused specifically on meta- the strategies successfully. For sounding out,
analytic analyses of remedial reading inter- the child learns the constituent letter–
ventions. Again, Swanson has demonstrated sounds of words in an order specified by the
in this report that a combined direct instruc- Reading Mastery I/II Fast Cycle or the Cor-
tion/strategy training approach is particu- rective Reading programs. To use the
larly important when outcomes in the do- rhyming strategy, the child learns a corpus
main of reading comprehension and text of keywords, which represent 120 high-
reading are considered. frequency spelling patterns in the English
language (adapted from the original Bench-
The PHAST Track Reading Program mark School Word Identification/Vocabu-
lary Development Program [Gaskins et al.,
Based on the results previously summarized, 1986]). Figure 17.2 lists the PHAST key-
our research team at The Hospital for Sick words. Peeling off requires the recognition
Children worked to integrate these two ap- of affixes whch are taught in a specific order
proaches into a single intervention program dictated by frequency of occurrence (e.g.,

284 EFFECTIVE INSTRUCTION

FIGURE 17.2. The PHAST Keyword Bank: A physical organization of keyword spelling patterns by
vowel and rime units. From the Benchmark School Word Identification/Vocabulary Development Pro-
gram by Gaskins, Downer, and Gaskins (1986). Copyright 1986 by Benchmark School. Adapted by
permission. Figure from Lovett, Lacerenza, and Borden (2000). Copyright 2000 by PRO-ED, Inc.
Reprinted by permission.

pre-, re-, un-, -ing, -ly, and -ment). Vowel a particular unknown word and monitors
alert requires the learning of short and long the effectiveness of strategy application in
pronunciations for single vowels and in- decoding the word sucessfully. The game
struction in vowel combinations and other Pyramid Challenge provides an attractive
variants (ea, oo, ie, ow). gameboard on which readers navigate a
pyramid’s different levels and tunnels by cor-
PHAST is designed to teach the child con- rectly decoding difficult challenge words.
text-free word identification skills and The challenge games present the readers—
strategies and to promote their immediate now equipped with a treasure chest of sever-
and effective application to the desired al effective decoding strategies—a playing
goal—reading connected text for meaning. field on which to practice various decoding
The 70 daily PHAST lessons include many plans and further refine their skills and
opportunities for the children to practice ap- strategies. Figure 17.3 summarizes the
plying a metacognitive Game Plan in regular PHAST Game Plan and Table 17.1 presents
story reading and in various challenge word sample dialogue for application of the Game
games. In the Game Plan, children choose, Plan in decoding an unknown word.
use, check, and score their application of the
five PHAST strategies. The Game Plan dia- The PHAST Track Reading Program is
logue both guides their implementation of intended to be situated as part of a linguisti-
the different strategies they have selected for cally enriched and literature-based balanced

Word Identification and Decoding Difficulties 285

FIGURE 17.3. The Game Plan: A formula for metacognitive instructional focus and the
strategy selection, application, monitoring, and dialogue structure of the PHAST Track
evaluation. Lovett, Lacerenza, and Borden Reading Program are compatible with dia-
(2000). Copyright 2000 by PRO-ED, Inc. logue-based approaches to text comprehen-
Reprinted by permission. sion and writing training. Ideally, full imple-
mentation of the PHAST Program in a
literacy program. PHAST is an intensive classroom setting would place it as part of a
program to teach word identification and highly integrated approach to reading,
word attack skills and to promote the devel- spelling, and writing instruction—an ap-
opment of the basic skills necessary to per- proach that allows intense instruction in
mit independent decoding and reading for spelling-to-sound and sound-to-spelling
meaning, information, or pleasure. The analysis at the subsyllabic, lexical, and con-
nected text levels. Our goal is to foster the
development of reading, spelling, compre-
hension, and literacy skills in a manner that
would equip the child for a world where the
demands for fluent, effective, multifaceted
literacy capabilities are ever increasing.

The PHAST Track Reading Program was
developed for struggling readers in our own
laboratory classrooms, but it is adaptable to
the needs of precocious and average readers
in the early elementary years. Current devel-
opments and extensions of the PHAST Pro-
gram include a PHAST Reading Compre-
hension Track (Lessons 71–140) that builds
on and follows the word identification
lessons in Lessons 1–70. Supplementary
spelling and writing tracks are also being pi-
loted to accompany the decoding and text
comprehension segments of the PHAST
Track Reading Program. Finally, a new
adaptation of PHAST for older individuals
with reading disorders is in development

TABLE 17.1. A Sample of Strategy Execution Using the Metacognitive Dialogue “GAME Plan” from
the PHAST Track Reading Program

The following is an example of the prompts and the dialogue of Game Plan as it is used to decode the
word unstacking:

Step 1. CHOOSE:
“My Game Plan is to first use Peeling Off. Then I am going to use the Rhyming Strategy and look for
the spelling patterns I know.”

Step 2. USE:
“I am Peeling Off un and ing. My next Game Plan is Rhyming. I see the spelling pattern a-c-k. The key
word is pack. If I know pack, then I know stack.”

Step 3. CHECK:
“I have to stop and think about whether I’m using the strategy(ies) properly. Is it working? Yes, I’ll
keep on going. I will put all the parts together—un-stack-ing.”

Step 4. SCORE/RE-CHOOSE:
“The word is unstacking. I scored! I used Peeling Off and Rhyming to help me figure out this word
and they worked.” (If the strategy did not result in a real word, the child begins again at Step 1, and
chooses another strategy to try.)

Note. From Lovett, Lacerenza, and Borden (2000). Copyright 2000 by PRO-ED, Inc. Reprinted by permission.

286 EFFECTIVE INSTRUCTION

(the PHAST PACES Program for struggling comprehension only, however. The subjects
high school readers and young adults). remained 2 full standard deviations below
age-level expectations on measures of read-
The PHAST Track Reading Program ing rate at follow-up; this deficit persisted
(Lessons 1–70) has been evaluated in a large despite an overall increase in the number of
multisite intervention study funded by words per minute read at follow-up and
NICHD: In this project, Robin Morris, substantial improvement in the children’s
Maryanne Wolf, and the first author are basic reading skills. Torgesen and his col-
evaluating the effectiveness of two different leagues speculate that to “close the gap” for
dual-focus, deficit-directed intervention older children with reading disorders, they
programs for young children with reading would have to add sight words (words that
disabilities (PHAST Lessons 1–70 [Lovett, can be recognized immediately, without
Lacerenza, & Borden, 2000] and PHAB/DI phonologically mediated decoding) to their
+ RAVE-O [Wolf, Miller, & Donnelly, lexicons at a rate that exceeded that of their
2000]) against both an alternative treatment agemates without disabilities. These investi-
control program (Classroom Survival Skills gators also acknowledge that some of the
+ Math [see Lovett, Lacerenza, Borden, Fri- older children with reading disabilities may
jters, et al., 2000]) and a phonological treat- have severe constraints upon levels of pro-
ment control program (PHAB/DI + CSS). A cessing speed, a deficit perhaps indexed by
second NICHD grant to our multisite group their slowed naming speeds on rapid autom-
is supporting evaluation of longer multi- atized naming tasks (Wolf, Bally, & Morris,
focus intervention programs based on the 1986; Wolf & Bowers, 1999).
original aforementioned 70-hour programs.
A 125-hour PHAST Track Reading Pro- There has been recent recognition that
gram is being evaluated as part of this de- extended models of remedial intervention
sign and includes both integrated Decoding need to be developed if children with read-
and Reading Comprehension tracks. The ing disabilities are to make gains in phono-
reader is referred to the published papers logical processing skills, word identification
for more detailed descriptions of the inter- and text reading accuracy and in word
ventions described in these pages. identification speed and text reading fluen-
cy with intervention. In fact, there remains
Beyond the Acquisition of Basic little consensus in definitions of reading flu-
Reading Skills: The Attainment of ency (Meyer & Felton, 1999; Wolf &
Reading Fluency in Individuals with Katzir-Cohen, 2001). Meyer and Felton
Developmental Reading Disorders (1999) describe fluency as “the ability to
read connected text rapidly, smoothly, ef-
In a recent study assessing two intensive and fortlessly, and automatically with little con-
individualized remediation programs for scious attention to the mechanics of reading
8–10-year-old children with reading disabil- such as decoding” (p. 284), a definition
ities, Torgesen, Alexander, and colleagues similar to LaBerge and Samuels’ (1974) ear-
(2001) found significant improvement in ly model of reading fluency. LaBerge and
generalized reading skills that were main- Samuels suggested that fluency results from
tained over a 2-year follow-up period but the automatization of component processes
little impact in terms of their subjects’ con- which allows attentional resources to be re-
tinued impairment in reading rate. Extend- allocated to other levels of processing.
ed individual instruction in both remedial Berninger, Abbott, Billingsley, and Nagy
programs (each 67½ hours and differing in (2001) propose a “systems” conceptualiza-
the pedagogical focus of the phonological tion of reading fluency, based on their re-
training) was associated with good growth search investigating components of reading,
in these children’s reading skills, and gradu- spelling, and written language development
ates of the two programs scored on average in children. Berninger and her colleagues
within the age-appropriate range of reading suggest that fluency depends on three main
achievement 2 years after intervention. sources of influence: characteristics of the
These improvements were demonstrated for informational input (visual or speech), the
measures of word identification and passage efficiency and automaticity of component
language processing systems, and coordina-

Word Identification and Decoding Difficulties 287

tion by an executive function (management) nent processes; (2) to facilitate the develop-
system. ment of fluency in word identification,
word attack, and text reading and compre-
Wolf and Katzir-Cohen (2001) propose a hension processes; and (3) to retrain the at-
developmental and component-based titudes and affect of disabled readers in
framework for understanding and promot- their approach to words and written lan-
ing the acquisition of reading fluency. They guage. RAVE-O encourages the children to
define reading fluency as learn to play with the English language
through animated computer games (e.g.,
the product of the initial development of ac- Speed Wizards; Wolf & Goodman, 1996),
curacy and the subsequent development of au- the building of imaginative word webs, in-
tomaticity in underlying sublexical processes, struction in systematic yet playful word-
lexical processes, and their integration in sin- retrieval strategies, and the reading of
gle-word reading and connected text. These minute mystery stories.
include perceptual, phonological, orthograph-
ic, and morphological processes at the letter, Like the PHAST Track Reading Program
letter-pattern, and word level, as well as se- described earlier, the RAVE-O program in
mantic and syntactic processes at the word combination with the previously described
level and connected text level. After it is fully PHAB/DI Program (Engelmann et al., 1988;
developed, reading fluency refers to a level of Lovett et al., 1994) is being evaluated in the
accuracy and rate where decoding is relatively same multisite intervention study funded by
effortless; where oral reading is smooth and NICHD. In this 5-year, three-city project,
accurate with correct prosody; and where at- Morris, Wolf, and the first author have been
tention can be allocated to comprehension. (p. investigating the effectiveness of different
219) multicomponent intervention programs for
young children with reading disabilites
It should be recognized, however, that (PHAB/DI + RAVE-O; and PHAB/DI +
much remains unknown about the nature WIST—i.e., the PHAST Track Reading Pro-
and reciprocity of the relationships between gram [Lovett, Lacerenza, & Borden, 2000])
reading fluency and text comprehension. It against both an alternative treatment con-
has long been assumed that increased effi- trol program (Classroom Survival Skills +
ciency and automatization of word recogni- Math [see Lovett et al., 2000]) and a phono-
tion and decoding processes result in a free- logical treatment control program
ing up of attentional resources for text (PHAB/DI + CSS). Again, the reader is re-
comprehension processes; however, whether ferred to the published papers for a more
direct reciprocal relationships between flu- complete description of the PHAST and
ency and comprehension exist has not been RAVE-O interventions described in these
established (Meyer & Felton, 1999; Wolf & pages. In a second 5-year intervention study
Katzir-Cohen, 2001). by the same investigators and funded by
NICHD, we are extending the intervention
From their developmental model of read- period and focus to evaluate the effective-
ing fluency, Wolf and her colleagues at ness of extended PHAST and RAVE-O Pro-
Tufts University have developed an experi- grams (125 hours) and to develop and eval-
mental reading intervention program called uate a triple-focus program that integrates
RAVE-O to facilitate its development the PHAST Track Reading (Decoding and
(Wolf, Miller, & Donnelly, 2000). RAVE-O Reading Comprehension Tracks) Program
is taught in combination with a systematic and RAVE-O.
phonologically based intervention program
that teaches letter–sound knowledge, de- Final Comments
coding, and word identification skills while
remediating the speech-based phonological New conceptualizations of developmental
processes of children with reading disabili- reading disabilities, the core underlying
ties. In combination with this decoding pro- learning impairments, and their effective re-
gram, RAVE-O (for Retrieval, Automatici- mediation place us in a better position to
ty, Vocabulary elaboration, Engagement undertake rigorous and productive research
with language, and Orthography) is de-
signed (1) to promote accuracy and auto-
maticity in reading subskills and compo-

288 EFFECTIVE INSTRUCTION

on reading disorders, their prevention in at- circumstances of millions of at-risk chil-
risk populations, their optimal remediation, dren.
and long-term outcome. Recent develop-
ments will allow us to refine our method- Acknowledgments
ologies for the measurement of change, its
evaluation and analysis, and now place us Preparation of this chapter was supported by a Na-
in a position to study individual and devel- tional Institute of Child Health and Human Devel-
opmental variation within a richer and opment Grant (HD30970-01A1) to Georgia State
more coherent theoretical context and with University, Tufts University, and The Hospital for
significantly improved research tools. Ques- Sick Children/University of Toronto. The remedia-
tions about what mediates, moderates, and tion research reported here was supported, in addi-
predicts change for children with develop- tion, by operating grants to the authors from the
mental reading disorders and what treat- Ontario Mental Health Foundation, the Velleman
ment or treatment combinations will best Foundation, and the Social Sciences and Humani-
serve them can now be productively ad- ties Research Council of Canada.
dressed. Advances in the cognitive neuro-
science of reading and in functional neu- We gratefully acknowledge the conceptual contri-
roimaging techniques allow additional butions of our collaborators and colleagues over the
evaluation of whether the functional and years, particularly those of Robin Morris and
neurobiological substrates of reading in the Maryanne Wolf. We also gratefully acknowledge
brain change with effective remediation of the intellectual contributions of senior members of
reading problems (Simos et al., 2002). the Learning Disabilities Research Program (LDRP)
at The Hospital for Sick Children—Karen A. Stein-
Within the past 5 years, there has been bach, Maria De Palma, Jan C. Frijters, Meredith
growing recognition of the need to study the Temple, Léa Lacerenza, Denis Murphy, Gail Mark-
remedial response of individuals with read- son, Jody Chong, and the whole LDRP staff who
ing disabilities along a continuum of skill have contributed so much to past and current inter-
development—addressing accuracy and rate vention studies.
criteria of skill acquisition in each reading-
related domain (Lovett, 1984; 1987). With Correspondence regarding this paper may be di-
recognition of the multidimensional nature rected to the authors at The Hospital for Sick Chil-
of the learning involved, we will be better dren, 555 University Avenue, Toronto, Ontario,
able to assess not only what components of CANADA M5G 1X8 or via e-mail at mwl@sick-
effective intervention allow children with kids.ca. Inquires about the Learning Disabilities Re-
reading disabilities to acquire basic decod- search Program may be directed to [email protected].
ing and word identification skills but also
what types of training allow them to consol- References
idate and automatize newly acquired skills
to the point that word identification and de- Barron, R. W. (1998). Proto-literate knowledge: An-
coding processes are relatively automatized tecedents and influences on phonological aware-
and that different types of text are read flu- ness and literacy. In C. Hulme & R. M. Joshi
ently and with good comprehension (Foor- (Eds.), Reading and spelling: Development and
man & Torgesen, 2001; Torgesen, Alexan- disorders (pp. 153–173). Mahwah, NJ: Erlbaum.
der, et al., 2001; Torgesen, Rashotte, &
Alexander, 2001; Wolf & Katzir-Cohen, Benson, N. J. (2000). Analysis of specific deficits:
2001). Foorman and Torgesen (2001) claim Evidence of transfer in disabled and normal read-
that if current research findings on effective ers following oral–motor awareness training.
classroom reading instruction were imple- Journal of Educational Psychology, 92(4),
mented, meeting the additional needs of the 646–658.
at-risk child for effective, intense, and ex-
plicit individual or small-group instruction, Benson, N. J., Lovett, M. W., & Kroeber, C. L.
the literacy needs of all children could be (1997). Training and transfer-of-learning effects
met. This is a case in which advances in the in disabled and normal readers: Evidence of spe-
science of reading disorders and interven- cific deficits. Journal of Experimental Child Psy-
tion research, if brought to the front line of chology, 64(3), 343–366.
educational practice, could change the life
Berninger, V. W., & Abbott, R. D. (1994). Redefin-
ing learning disabilities: Moving beyond apti-
tude–achievement discrepancies to failure to re-
spond to validated treatment protocols. In G. R.
Lyon (Ed.), Frames of reference for the assess-
ment of learning disabilities: New views on mea-
surement issues (pp. 163–183). Baltimore:
Brookes.

Berninger, V. W., Abbott, R. D., Billingsley, F., &

Word Identification and Decoding Difficulties 289

Nagy, W. (2001). Processes underlying timing What’s gone wrong in America’s classrooms (pp.
and fluency of reading: Efficiency, automaticity, 49–90). Stanford, CA: Hoover Institution Press.
coordination, and morphological awareness. In Fletcher, J. M., Shaywitz, S. E., Shankweiler, D. P.,
M. Wolf (Ed.), Dyslexia, fluency, and the brain Katz, L., Liberman, I. Y., Stuebing, K. K., Fran-
(pp. 383–413). Timonium, MD: York Press. cis, D. J., Fowler, A. E., & Shaywitz, B. A.
Bishop, D. V., & Adams, C. (1990). A prospective (1994). Cognitive profiles of reading disability:
study of the relationship between specific lan- Comparisons of discrepancy and low achieve-
guage impairment, phonological disorders and ment definitions. Journal of Educational Psychol-
reading disabilities. Journal of Child Psychology ogy, 86(1), 6–23.
and Psychiatry, 31(7), 1027–1050. Foorman, B. R., Francis, D. J., Fletcher, J. M.,
Bowers, P. G., & Wolf, M. (1993). Theoretical links Schatschneider, C., & Mehta, P. (1998). The role
between naming speed, precise mechanisms, and of instruction in learning to read: Preventing
orthographic skill in dyslexia. Reading and Writ- reading failure in at-risk children. Journal of Ed-
ing: An Interdisciplinary Journal, 5, 69–85. ucational Psychology, 90(1), 37–55.
Brady, S. A. (1997). Ability to encode phonological Foorman, B. R., Francis, D. J., Winikates, D.,
representations: An underlying difficulty of poor Mehta, P., Schatschneider, C., & Fletcher, J. M.
readers. In B. A. Blachman (Ed.), Foundations of (1997). Early interventions for children with
reading acquisition and dyslexia (pp. 21–47). reading disabilities. Scientific Studies of Reading,
Hillsdale, NJ: Erlbaum. 1(3), 255–276.
Bruck, M. (1992). Persistence of dyslexics’ phono- Foorman, B. R., & Torgesen, J. (2001). Critical ele-
logical awareness deficits. Developmental Psy- ments of classroom and small-group instruction
chology, 28, 874–886. promote reading success in all children. Learning
Bruck, M. (1998). Outcomes of adults with child- Disabilities Research and Practice, 16(4),
hood histories of dyslexia. In C. Hulme & R. M. 203–212.
Joshi (Eds.), Reading and spelling: Development Francis, D. J., Shaywitz, S. E., Stuebing, K. K.,
and disorders (pp. 179–200). Mahwah, NJ: Erl- Shaywitz, B. A., & Fletcher, J. M. (1996). Devel-
baum. opmental lag versus deficit models of reading dis-
Cattell, M. (1886). The time it takes to see and ability: A longitudinal, individual growth curves
name objects. Mind, 2, 63–85. analysis. Journal of Educational Psychology,
Cunningham, A. E., & Stanovich, K. (1997). Early 88(1), 3–17.
reading acquisition and its relation to reading ex- Gaskins, I. W., Downer, M. A., & Gaskins, R. W.
perience and ability. Developmental Psychology, (1986). Introduction to the Benchmark School
33, 934–945. Word Identification/Vocabulary Development
Elbro, C., Borstrom, I., & Petersen, D. K. (1998). program. Media, PA: Benchmark School.
Predicting dyslexia from kindergarten: The im- Gathercole, S. E., & Baddeley, A. D. (1987). The
portance of distinctness of phonological repre- processes underlying segmental analysis. Euro-
sentations of lexical items. Reading Research pean Bulletin of Cognitive Psychology, 7,
Quarterly, 33(1), 36–60. 462–464.
Engelmann, S., & Bruner, E. C. (1988). Reading Gittelman, R. (1983). Treatment of reading disor-
Mastery I/II Fast Cycle: Teacher’s guide. Chicago: ders. In M. Rutter (Ed.), Developmental Neu-
Science Research Associates. ropsychiatry (pp. 520–541). New York: Guilford
Engelmann, S., Carnine, L., & Johnson, G. (1988). Press.
Corrective reading: Word attack basics, decoding Goswami, U., & Bryant, P. E. (1990). Phonological
A. Chicago: Science Research Associates. skills and learning to read. Hove, East Sussex,
Engelmann, S., Johnson, G., Carnine, L., Meyer, L., UK: Erlbaum.
Becker, W., & Eisele, J. (1988). Corrective read- Grigorenko, E. L. (2001). Developmental dyslexia:
ing: Decoding strategies, decoding B1. Chicago: An update on genes, brains, and environments.
Science Research Associates. Journal of Child Psychology and Psychiatry,
Felton, R. H., & Brown, I. S. (1990). Phonological 42(1), 91–125.
processes as predictors of specific reading skills Grigorenko, E. L., Wood, F. B., Meyer, M. S., Hart,
in children at risk for reading failure. Reading L. A., Speed, W. C., Shuster, A., & Pauls, D. L.
and Writing: An Interdisciplinary Journal, 2, (1997). Susceptibility loci for distinct compo-
39–59. nents of developmental dyslexia on chromosomes
Filipek, P. A. (1995). Neurobiological correlates of 6 and 15. American Journal of Human Genetics,
developmental dyslexia: What do we know of 60, 27–39.
how dyslexics’ brains differ from those of normal Hanushek, E. A., Kain, J. F., & Rivkin, S. G.
readers? Journal of Child Neurology, 10(Suppl.), (1998). Does special education raise academic
S62–S69. achievement for students with disabilities? Na-
Filipek, P. A. (1999). Neuroimaging in the develop- tional Bureau of Economic Research, Working
mental disorders: The state of the science. Jour- Paper No. 6690, Cambridge, MA.
nal of Child Psychology and Psychiatry, 40(1), Hewison, J. (1982). The current status of remedial
113–128. intervention for children with reading problems.
Fletcher, J. M., & Lyon, G. R. (1998). Reading: A Developmental Medicine and Child Neurology,
research-based approach. In W. M. Evers (Ed.), 24, 183–186.

290 EFFECTIVE INSTRUCTION

Hinshelwood, J. (1917). Congenital word blindness. Vol. 8, Part II, pp. 255–300). Amsterdam: Elsevi-
London: H. K. Lewis. er Science.
Lovett, M. W., & Barron, R. W. (2002). The search
Huey, E. B. (1908). The psychology and pedagogy. for individual and subtype differences in reading
New York: MacMillan. disabled children’s response to remediation. In D.
L. Molfese & V. J. Molfese (Eds.), developmental
Juel, C., & Minden-Cupp, C. (2000). Learning to variations in learning: Applications to social, ex-
read words: Linguistic units and instructional ecutive function, language, and reading skills (pp.
strategies. Reading Research Quarterly, 35(4), 309–337). Mahwah, NJ: Erlbaum.
458–492. Lovett, M. W., Borden, S. L., DeLuca, T., Laceren-
za, L., Benson, N. J., & Brackstone, D. (1994).
LaBerge, D., & Samuels, S. J. (1974). Toward a the- Treating the core deficits of developmental
ory of automatic information processing in read- dyslexia: Evidence of transfer-of-learning follow-
ing. Cognitive Psychology, 6, 293–323. ing phonologically- and strategy-based reading
training programs. Developmental Psychology,
Levin, B. E. (1990). Organizational deficits in 30(6), 805–822.
dyslexia: Possible frontal lobe dysfunction. De- Lovett, M. W., Hinchley, J., & Benson, N. J. (1997).
velopmental Neuropsychology, 6, 95–110. Assessing the remedial gains of disabled learners:
Conceptual, measurement, and statistical consid-
Levy, B. A., Abello, B., & Lysynchuk, L. (1997). erations in the evaluation of remedial outcome.
Transfer from word training to reading in con- Unpublished manuscript.
text: Gains in reading fluency and comprehen- Lovett, M. W., Lacerenza, L., & Borden, S. L.
sion. Learning Disability Quarterly, 20(3), (2000). Putting struggling readers on the PHAST
173–188. track: A program to integrate phonological and
strategy-based remedial reading instruction and
Liberman, A., & Mattingly, I. G. (1985). The motor maximize outcomes. Journal of Learning Disabil-
theory of speech perception revised. Cognition, ities, 33(5), 458–476.
21, 1–36. Lovett, M. W., Lacerenza, L., Borden, S. L., Frijters,
J. C., Steinbach, K. A., & De Palma, M. (2000).
Liberman, I. Y., Shankweiler, D., & Liberman, A. Components of effective remediation for devel-
M. (1989). The alphabetic principle and learning opmental reading disabilities: Combining phono-
to read. In D. Shankweiler & I. Y. Liberman logical and strategy-based instruction to improve
(Eds.), Phonology and reading disability: Solving outcomes. Journal of Educational Psychology,
the reading puzzle (pp. 1–33). Ann Arbor: Uni- 92(2), 263–283.
versity of Michigan Press. Lovett, M. W., Ransby, M. J., & Barron, R. W.
(1988). Treatment, subtype, and word type ef-
Lindamood, C. H., & Lindamood, P. C. (1975). fects in dyslexic children’s response to remedia-
Auditory discrimination in depth. Allen, TX: tion. Brain and Language, 34, 328–349.
DLM Teaching Resources. Lovett, M. W., Ransby, M. J., Hardwick, N., Johns,
M. S., & Donaldson, S. A. (1989). Can dyslexia
Lovett, M. W. (1984). A developmental perspective be treated? Treatment-specific and generalized
on reading dysfunction: Accuracy and rate crite- treatment effects in dyslexic children’s response
ria in the subtyping of dyslexic children. Brain to remediation. Brain and Language, 37, 90–121.
and Language, 22, 67–91. Lovett, M. W., & Steinbach, K. A. (1997). The ef-
fectiveness of remedial programs for reading dis-
Lovett, M. W. (1987). A developmental approach to abled children of different ages: Is there de-
reading disability: Accuracy and speed criteria of creased benefit for older children? Learning
normal and deficient reading skill. Child Devel- Disability Quarterly, 20(3), 189–210.
opment, 58, 234–260. Lovett, M. W., Steinbach, K. A., & Frijters, J. C.
(2000). Remediating the core deficits of develop-
Lovett, M. (1991). Reading, writing, and remedia- mental reading disability: A double deficit per-
tion: Perspectives on the dyslexic learning disabil- spective. Journal of Learning Disabilities, 33(4),
ity from remedial outcome data. Learning and 334–358.
Individual Differences, 3, 295–305. Lovett, M. W., Warren-Chaplin, P. M., Ransby, M.
J., & Borden, S. L. (1990). Training the word
Lovett, M. W. (1992). Developmental dyslexia. In I. recognition skills of reading disabled children:
Rapin & S. J. Segalowitz (Eds.), Handbook of Treatment and transfer effects. Journal of Educa-
neuropsychology, Vol. 7: Child neuropsychology tional Psychology, 82, 769–780.
(pp. 163–185). Amsterdam: Elsevier Science. Lyon, G. R. (1995). Research initiatives in learning
disabilities: Contributions from scientists sup-
Lovett, M. W. (1997). Developmental reading dis- ported by the National Institute of Child Health
orders. In T. E. Feinberg & M. J. Farah (Eds.), and Human Development. Journal of Child Neu-
Behavioral neurology and neuropsychology (pp. rology, 10(1), S120–S126.
773–787). New York: McGraw-Hill. Lyon, G. R. (1996). Learning disabilities. In E. J.

Lovett, M. W. (1999). Defining and remediating the
core deficits of developmental dyslexia: Lessons
from remedial outcome research with reading
disabled children. In R. Klein & P. McMullen
(Eds.), Converging methods for understanding
reading and dyslexia. Language, speech, and
communication (pp. 111–132). Cambridge, MA:
MIT Press.

Lovett, M. W., & Barron, R. W. (2003). Neuropsy-
chological perspectives on reading development
and developmental reading disorders. In S. J.
Segalowitz & I. Rapin (Eds.), Handbook of neu-
ropsychology. Child neuropsychology (2nd ed.,

Word Identification and Decoding Difficulties 291

Marsh & R. A. Barkley (Eds.), Child psycho- Specific reading disability: A view of the spec-
pathology (pp. 390–435). New York: Guilford trum (pp. 75–119). Timonium, MD: York Press.
Press. Shanahan, T., & Barr, R. (1995). Reading Recov-
Lyon, G. R., & Moats, L. C. (1997). Critical con- ery: An independent evaluation of the effects of
ceptual and methodological considerations in an early instructional intervention for at-risk
reading intervention research. Journal of Learn- learners. Reading Research Quarterly, 30(4),
ing Disabilities, 30(6), 578–588. 958–996.
Mann, V. (1986). Why some children encounter Share, D. L. (1995). Phonological recoding and self-
reading problems. In J. Torgesen & B. Wong teaching. Sine qua non of reading acquisition.
(Eds.), Psychological and educational perspec- Cognition, 55, 151–218.
tives on learning disabilities (pp. 133–159). New Shaywitz, S. E., Fletcher, J. M., Holahan, J. M.,
York: Academic Press. Schneider, A. E., Marchione, K. E., Stuebing, K.
Meyer, M. M., & Felton, R. H. (1999). Repeated K., Francis, D. J., Pugh, K. R., & Shaywitz, B. A.
reading to enhance fluency: Old approaches and (1999). Persistence of dyslexia: The Connecticut
new direction. Annals of Dyslexia, 49, 283–306. Longitudinal Study at adolescence. Pediatrics,
Moats, L. C., & Foorman, B. R. (1997). Introduc- 104(6), 1351–1359.
tion to special issue of SSR: Components of effec- Shaywitz, S. E., Shaywitz, B. A., Pugh, K. R., Ful-
tive reading instruction. Scientific Studies of bright, R. K., Constable, R. T., Mencl, W. E.,
Reading, 1(3), 187–189. Shankweiler, D. P., Liberman, A. M., Skudlarski,
Morgan, W. P. (1896). A case of congenital word- P., Fletcher, J. M., Katz, L., Marchione, K. E., La-
blindness. British Medical Journal, 2, 1378. cadie, C., Gatenby, C., & Gore, J. C. (1998,
Olson, R. K., Wise, B., Ring, J., & Johnson, M. March). Functional disruption in the organiza-
(1997). Computer-based remedial training in tion of the brain for reading in dyslexia. Proceed-
phoneme awareness and phonological decoding: ings of the National Academy of Sciences of the
Effects on the posttraining development of word United States of America, 95, 2636–2641.
recognition. Scientific Studies of Reading, 1(3), Simos, P. G., Fletcher, J. M., Bergman, E., Breier, J.
235–254. I., Foorman, B. R., Castillo, E. M., Davis, R. N.,
Pennington, B. F. (1997). Using genetics to dissect Fitzgerald, M., & Papanicolaou, A. C. (2002).
cognition. American Journal of Human Genetics, Dyslexia-specific brain activation profile becomes
60, 13–16. normal following successful remedial training.
Pennington, B. F. (1999). Toward an integrated un- Neurology, 58, 1203–1213.
derstanding of dyslexia: Genetic, neurological, Snowling, M., & Hulme, C. (1993). Developmental
and cognitive mechanisms. Development and dyslexia and language disorders. In G. Blanken,
Psychopathology, 11, 629–654. J. Dittmann, H. Grimm, J. C. Marshall, & C.-W.
Pugh, K. R., Mencl, W. E., Jenner, A. R., Katz, L., Wallesch (Eds.), Linguistic disorders and
Frost, S. J., Lee, J. R., Shaywitz, S. E., & Shay- pathologies: An international handbook (pp.
witz, B. A. (2001). Neurobiological studies of 724–732). New York: Walter de Gruyter.
reading and reading disability. Journal of Com- Stanovich, K. E. (1986). Matthew effects in reading:
munication Disorders, 34(6), 479–492. Some consequences of individual differences in
Pugh, K. R., Mencl, W. E., Jenner, A. J., Katz, L., the acquisition of literacy. Reading Research
Lee, J. R., Shaywitz, S. E., & Shaywitz, B. A. Quarterly, 21, 360–407.
(2000). Functional neuroimaging studies of read- Stanovich, K. E. (1991). Changing models of read-
ing and reading disability (developmental dyslex- ing and reading acquisition. In L. Rieben & C. A.
ia). Mental Retardation and Developmental Dis- Perfetti (Eds.), Learning to read: Basic research
abilities Review, 6, 207–213. and Its implications (pp. 19–31). Hillsdale, NJ:
Rayner, K., Foorman, B. R., Perfetti, C. A., Peset- Erlbaum.
sky, D., & Seidenberg, M. S. (2001). How psy- Stanovich, K. E. (1994). Annotation: Does dyslexia
chological science informs the teaching of read- exist? Journal of Child Psychology and Psychia-
ing. Psychology in the Public Interest, 2, 31–74. try, 55(4), 579–595.
Scanlon, D. M., & Vellutino, F. R. (1997). A com- Stanovich, K. E., & Siegel, L. S. (1994). Phenotypic
parison of the instructional backgrounds and performance profile of children with reading dis-
cognitive profiles of poor, average, and good abilities: A regression-based test of the phonolog-
readers who were initially identified as at risk for ical-core variable-difference model. Journal of
reading failure. Scientific Studies of Reading, Educational Psychology, 86(1), 24–53.
1(3), 191–216. Swanson, H. L. (1999a). Reading comprehension
Scarborough, H. S. (1984). Continuity between and working memory in learning-disabled read-
childhood dyslexia and adult reading. British ers: Is the phonological loop more important
Journal of Psychology, 75(3), 329–348. than the executive system? Journal of Experimen-
Scarborough, H. S. (1990). Very early language tal Child Psychology, 72, 1–31.
deficits in dyslexic children. Child Development, Swanson, H. L. (1999b). Reading research for stu-
61, 1728–1743. dents with learning disabilities: A meta-analysis
Scarborough, H. S. (1998). Early identification of of intervention outcomes. Journal of Learning
children at risk for reading disabilities. In B. K. Disabilities, 32(6), 504–532.
Shapiro, P. J. Accardo, & A. J. Capute (Eds.), Swanson, H. L., & Alexander, J. E. (1997). Cogni-

292 EFFECTIVE INSTRUCTION

tive processes as predictors of word recognition the acquisition of reading skills. Psychological
and reading comprehension in learning-disabled Bulletin, 101, 192–212.
and skilled readers: Revisiting the specificity hy- Wagner, R. K., Torgesen, J. K., & Rashotte, C. A.
pothesis. Journal of Educational Psychology, (1994). Development of reading-related phono-
89(1), 128–158. logical processing abilities: New evidence of bidi-
Swanson, H. L., & Hoskyn, M. (1998). Experimen- rectional causality from a latent variable longitu-
tal intervention research on students with learn- dinal study. Developmental Psychology, 30(1),
ing disabilities: A meta-analysis of treatment out- 73–87.
comes. Review of Educational Research, 68(3), Wise, B. W., & Olson, R. K. (1995). Computer-
277–321. based phonological awareness and reading in-
Torgesen, J. K. (2000). Individual differences in struction. Annals of Dyslexia, 45, 99–122.
response to early interventions in reading: The Wise, B. W., Ring, J., & Olson, R. K. (2000). Indi-
lingering problem of treatment resisters. Learn- vidual differences in gains from computer-assist-
ing Disabilities Research and Practice, 15(1), ed remedial reading. Journal of Experimental
55–64. Child Psychology, 77, 197–235.
Torgesen, J. K., Alexander, A. W., Wagner, R. K., Wise, B. W., Ring, J., Sessions, L., & Olson, R. K.
Rashotte, C. A., Voeller, K. K. S., & Conway, T. (1997). Phonological awareness with and with-
(2001). Intensive remedial instruction for chil- out articulation: A preliminary study. Learning
dren with severe reading disabilities: Immediate Disability Quarterly, 20(3), 211–225.
and long-term outcomes from two instructional Wolf, M. (1991). Naming speed and reading: The
approaches. Journal of Learning Disabilities, contribution of the cognitive neurosciences.
34(1), 33–58. Reading Research Quarterly, 26, 123–141.
Torgesen, J. K., Rashotte, C. A., & Alexander, A. Wolf, M., Bally, H., & Morris, R. (1986). Auto-
W. (2001). Principles of fluency instruction in maticity, retrieval processes, and reading: A lon-
reading: Relationships with established empirical gitudinal study in average and impaired readers.
outcomes. In M. Wolf (Ed.), Dyslexia, fluency, Child Development, 57(4), 988–1000.
and the brain (pp. 333–355). Timonium, MD: Wolf, M., & Bowers, P. G. (1999). The double-
York Press. deficit hypothesis for the developmental dyslexi-
Torgesen, J. K., Wagner, R. K., & Rashotte, C. A. as. Journal of Educational Psychology, 91(3),
(1997). Approaches to the prevention and reme- 415–438.
diation of phonologically-based reading disabili- Wolf, M., & Bowers, P. G. (2000). Naming-speed
ties. In B. A. Blachman (Ed.), Foundations of processes and developmental reading disabilities:
reading acquisition and dyslexia: Implications for An introduction to the special issue on the dou-
early intervention (pp. 287–304). Mahwah, NJ: ble-deficit hypothesis. Journal of Learning Dis-
Erlbaum. abilities, 33(4), 322–324.
Torgesen, J., Wagner, R., Rashotte, C., Lindamood, Wolf, M., Bowers, P. G., & Biddle, K. (2000). Nam-
P., Rose, E., Conway, T., & Garvan, C. (1999). ing-speed processes, timing, and reading: A con-
Preventing reading failure in young children with ceptual review. Journal of Learning Disabilities,
phonological processing disabilities: Group and 33(4), 387–407.
individual responses to instruction. Journal of Wolf, M., & Goodman, G. (1996). Speed wizards
Educational Psychology, 91(4), 579–593. [Computer software], Tufts University, Boston
Vellutino, F. R., Scanlon, D. M., Sipay, E. R., Small, and Rochester Institute of Technology, Rochester,
S. G., Pratt, A., Chen, R., & Denckla, M. B. NY.
(1996). Cognitive profiles of difficult-to-remedi- Wolf, M., & Katzir-Cohen, T. (2001). Reading flu-
ate and readily remediated poor readers: Early in- ency and its intervention. Scientific Studies of
tervention as a vehicle for distinguishing between Reading, 5(3), 211–239.
cognitive and experiential deficits as basic causes Wolf, M., Miller, L., & Donnelly, K. (2000). Re-
of specific reading disability. Journal of Educa- trieval, automaticity, vocabulary elaboration, or-
tional Psychology, 88(4), 601–638. thography (RAVE-O): A comprehensive, fluency-
Wagner, R. K., & Torgesen, J. K. (1987). The nature based reading intervention program. Journal of
of phonological processing and its causal role in Learning Disabilities, 33(4), 375–386.

18

Teaching Text Structure to Improve
Reading Comprehension

Joanna P. Williams

Comprehending what is read presents a gen- how to deploy them (Swanson & Hoskyn,
uine challenge for many students with learn- 1998). Students are directed to stop occa-
ing disabilities. Sometimes the difficulty sionally during their reading to monitor
stems from a lack of fluency in word recog- their understanding by asking themselves
nition. However, in many cases children questions or by trying to summarize. They
who read fluently do not understand what are taught to take steps to ensure their un-
they read because of cognitive processing derstanding by rereading, by trying to con-
problems such as working memory limita- nect the material to be learned with what
tions, lexical processing difficulties, poor in- they already know, and by using other gen-
ference making, and ineffective comprehen- eral study skills (Swanson & Hoskyn,
sion monitoring (Gersten, Fuchs, Williams, 1998). These strategies can be thought of as
& Baker, 2001). In other cases their knowl- ways to put oneself in a position to compre-
edge, either of word meanings or of a partic- hend. They might be considered remedial in
ular domain, may be minimal (Perfetti, Mar- the sense that good readers seem to use
ron, & Foltz, 1996). Of course, many of them without specific instruction and in-
these same problems are sometimes seen in deed often without even being aware that
students who do not have learning disabili- they are using them.
ties; these students, too, have trouble under-
standing what they read. Some strategies involve more than be-
coming aware of, manipulating, and
About 30 years ago, researchers began to monitoring one’s cognitive processing. Well-
investigate the nature of reading compre- structured text enhances recall and compre-
hension by analyzing the strategies used by hension, for those who have acquired sensi-
proficient readers. Today a focus on teach- tivity to structure (Pearson & Dole, 1987),
ing such strategies has become the main ap- and instruction designed to teach students
proach to comprehension instruction (Press- to recognize the underlying structure of the
ley, 2000). Most strategy instruction text that they are reading improves compre-
involves a strong emphasis on metacogni- hension. This instruction typically involves
tion; that is, instruction is geared toward an teaching students to identify the important
awareness of one’s cognitive processes and structural elements of a particular type of

293

294 EFFECTIVE INSTRUCTION

text and then to memorize a list of generic process larger units (Perfetti, 1985). Yet
questions that cue a search for those impor- synthetic phonics is an effective way to
tant elements. This instruction involves new teach beginning reading (National Reading
knowledge about text structure and how to Panel, 2000). Similarly, it is reasonable to
use this knowledge in a strategy. suggest that early comprehension instruc-
tion might be more effective when it is not
Different types of text are organized in based strictly on a model of proficient read-
different ways. Narrative text can be de- ing. Texts that are well structured (even if
scribed as following a single general struc- “contrived” for specific instructional pur-
tural pattern (often called story grammar; poses) may prove to be useful materials for
Mandler & Johnson, 1977), and informa- the classrooms.
tional text comes in a variety of patterns
(e.g., description, sequence, compare–con- It should be kept in mind that the specific
trast, cause–effect, and problem–solution). structures that we are talking about are not
Children develop sensitivity to narrative limited to text; they are rhetorical structures
structure early, and they use it to compre- that reflect universal cognitive processes.
hend simple stories before they enter school. The thinking of young children exhibits
But informational text, because it comprises forms of all these structures. By the time
a variety of structures and also because it children enter school, they tell stories, com-
more often deals with unfamiliar content, is pare and contrast objects, order events in a
more difficult to comprehend (Kucan & temporal sequence, and may even impute
Beck, 1997). It has been considered so chal- causality in rudimentary ways. But children
lenging for young children that until recent- have not had sufficient practice to be able to
ly, one rarely saw any informational text in use these structures easily, and sometimes
elementary school classrooms. they do not even recognize opportunities for
using them to enhance their comprehension.
There was a considerable amount of re- Helping students recognize the structure in-
search on text structure in the 1980s. But in herent in text—and match it to their own
the 1990s, under the influence of a con- cognitive structures—will help them under-
structivist philosophy (Bean, 2000; Rosen- stand and produce not only text but also
blatt, 1978), the focus of research efforts spoken discourse. And when they encounter
shifted. Attention turned away from the ef- text whose structure is complex or text that
fects of features of text to issues such as the is poorly organized, they will be able to sim-
characteristics of the reader and the social plify or reorganize it in order to better com-
nature of comprehension. Unfortunately, prehend it.
this change in research focus occurred long
before the implications of the findings on The Student with Learning Disabilities
text structure had been thoroughly investi-
gated and effective instructional programs The field of learning disabilities has been
developed. plagued since its inception by difficulties of
definition (Hammill, 1990). When I first be-
One of the hallmarks of the new con- gan to work on reading instruction for chil-
structivist approach that has strongly influ- dren with learning disabilities, I was per-
enced the selection of instructional materi- plexed by the question of what “learning
als is its emphasis on using only authentic disabilities” really meant. I realized that
texts. Arguments have been made that be- many people had the same question. My
cause only a small proportion of authentic own main interest was in reading and in in-
text actually follows any single specific structional design, and so after considerable
structure (and yet proficient readers, of reflection, I decided that I could work with
course, do understand the more complicat- children with reading difficulties whom
ed text that they usually encounter), there is someone else had labeled as having a learn-
little reason to spend much instructional ing disability. My rationale was that what I
time on text structure. However, one could was trying to do was develop materials and
argue that to base early instruction on what techniques for poor readers in general, a
proficient readers do, as many recommend, more inclusive category. If I succeeded, then
is not necessarily the best approach to take.
For example, proficient readers do not
sound out individual letters in words; they

Teaching Text Structure to Improve Reading Comprehension 295

surely those who had specific learning dis- will tell them what to do. That is the real
abilities would be helped. challenge of much comprehension instruc-
tion . . . and the reason why so much em-
I still hold that point of view, and I be- phasis is placed on general metacognitive
lieve that the evidence is bearing me out. strategies.
That is, regardless of the important contro-
versies about the nature and extent of learn- Our initial focus was on normally achiev-
ing disabilities, instructional techniques that ing children in the fourth to sixth grades. In
are found to be effective and are recom- a series of studies (William, Taylor, &
mended for students with learning disabili- Ganger, 1981), my students and I asked
ties who have reading problems are not children to read short paragraphs and to se-
qualitatively different from those recom- lect an appropriate title from an array of
mended for other poor readers (Williams, choices and write a summary sentence for
1992). For each of the studies I describe the paragraph. These are typical tasks that
here, I provide a description of the popula- one might see in a classroom. All of our
tion drawn from. In all cases the students paragraphs were written on a low readabili-
who have been designated as having learn- ty level, so that difficulties with decoding
ing disabilities have difficulty with reading; would not confound our findings.
I refer to them either as students with learn-
ing disabilities or with reading disabilities. Consider the following paragraph:

Finding the Main Idea Cowboys had to protect the herd from cattle
robbers. Cowboys had to brand cattle to show
“Comprehension” is a general term, encom- who owned them. They had to ride around the
passing a wide variety of skills and perfor- ranch to keep cattle from straying too far.
mances. The core of reading comprehension Sometimes cowboys had to separate the cattle
is the ability to get from a text its gist or its that were to be sent to market.
point. Traditionally, this ability is called
finding the main idea (Pearson & Johnston, In this paragraph, each sentence instantiates
1978). Without being able to understand a global topic (“cowboys”), and the reader
the point of a text, one cannot draw appro- can construct a proposition that at a higher
priate inferences from it; nor can one com- level subsumes the three sentences (van
pare texts without understanding the main Dijk, 1980, p. 46): “Cowboys had jobs to
points of each. This ability is fundamental do.”
to basic comprehension, to effective study-
ing, and to critical thinking. The fact that We found that there was a clear develop-
instruction in how to find main ideas has al- mental progression in ability across school
ways been one of the most common ele- grades, and that performance was better
ments of the elementary-school curriculum when readers had merely to select the main
underscores its importance. idea from an array than when they had to
formulate it as a summary sentence. Also,
“Main idea” is rather a challenge to children performed better on paragraphs
teach. Children often have difficulty in iden- with topic sentences than on those without
tifying main ideas of even rather simple (although only when the topic sentences
texts (Baumann, 1984). Available instruc- were highlighted). We replicated these find-
tional materials have been evaluated and ings with children with learning disabilities
found wanting. It is often suggested that (Taylor & Williams, 1983); we compared
teachers give children explanations that go them with normally achieving children who
like this: “To find the main idea, pick out were matched in terms of IQ and word
the most important point,” or “If you have knowledge scores. All children were reading
picked out the most important sentence, at the fourth- or fifth-grade level. Across all
you have probably found the main idea.” our experimental tasks, the children with
Such statements communicate the impor- learning disabilities did just as well as the
tance of finding main ideas, but they do not younger children without disabilities. This
tell or show how to find one. In fact, there is finding suggested to us that the two groups
nothing you can tell children directly that were not qualitatively different with respect
to the ability to generate macrostructure,
and that instruction that focuses on the de-
velopment of main idea skills should not

296 EFFECTIVE INSTRUCTION

necessarily be different for children with We went on, in this main idea work,
disabilities than for children without dis- to develop an instructional sequence
abilities—a conclusion that is consonant (Williams, Taylor, Jarin, & Milligan, 1983).
with the general point of view that instruc- We used the same sort of simple, highly
tional development for the reading disabled structured paragraphs as we had in our ex-
should focus on whatever strategies and perimental studies and designed a program
techniques are effective for the general cate- that emphasized clear definition of main
gory of slow readers. idea, clear description of the task, and an
explanation of why it was important.
There was, however, one finding that dif-
ferentiated the two groups. We included In addition, we incorporated general
what we called parenthetical information in principles of instruction into the design: (1)
some of the paragraphs, information that the use of well-structured examples of the
was either unrelated to or else only tangen- prototypical task, (2) consistent modeling of
tially related to the propositional hierarchy the strategies being taught, (3) a sequence of
of the text. We did this on the grounds that tasks, (4) a sequence of response demands
natural text does not always consist of well- that reflected a progression from easier to
structured paragraphs, and so it is impor- more difficult material, (5) gradual removal
tant for readers to be able to disregard of the teacher from the task, and (6) provi-
anomalous information when reading for sion for extensive practice and feedback. We
gist. also chose to externalize some of the steps
in the comprehension process that are, in
We asked the children to identify the in- actuality, implicit, that is, to externalize
appropriate sentence in the paragraph. For their thinking. We did this by having the
example, in the cowboy paragraph, that students highlight some of the textual cues,
sentence might read, “Cowboys often wear that is, circle the most frequent word or
leather jackets and fancy boots.” The posi- idea, to help figure out the general category
tion of the anomalous sentence was varied that encompassed each sentence.
across paragraphs, appearing as either the
second, third, or fourth (last) sentence. Chil- Another aspect of our training was our
dren without disabilities were able to identi- use of anomalous sentences and the system-
fy a sentence as anomalous the closer it was atic introduction of different types of anom-
to the end of the paragraph, but the children aly: first, sentences totally unrelated to the
with learning disabilities showed no such ef- topic of the paragraph, then sentences that
fect, suggesting that the latter were not as were tangentially related and therefore
good at building up a representation gradu- rather confusing. Children were taught to
ally as the information in each succeeding identify the deviant sentence, to cross it out,
sentence was processed. They were just as as part of this “externalization” of the
willing to accept the anomalous sentence thought process, and then to formulate a
even when it appeared late in the para- main idea on the marked-up paragraph.
graph: They did not edit.
In evaluating the program (Williams et
I do not believe that this difficulty is the al., 1983), we worked with 11-year-old chil-
same as saying that children with learning dren with learning disabilities, about two
disabilities do not monitor their comprehen- grade levels below average reading level for
sion; that is, when they are faced with the their age. Children given 10 lessons were
task of saying whether a sentence belongs in better able than children who did not re-
the paragraph, they probably do compare ceive the lessons to identify anomalous sen-
what the sentence says with their current tences and to write sentences on both mate-
representation of the paragraph, as, pre- rials that had been used in training and
sumably, good readers do. Rather, I think similar materials that had not.
that their representation of the paragraph
develops less adequately than that of the One of the major concerns in this sort of
child without disabilities; thus when the work is to determine whether our instruc-
child with disabilities does compare sen- tional efforts lead to transfer, that is, that
tence with paragraph representation, the students will learn to work with new mater-
outcome of the comparison is not likely to ial: It is their ability to comprehend in gen-
be on target. eral that we are concerned about. In this
case, we can say that there was transfer, al-

Teaching Text Structure to Improve Reading Comprehension 297

beit only to similarly structured materials. Although by the time they start school
We did not do any posttesting on other most children have a good grasp of narra-
types of material. However, although the tive structure on the level of the plot, they
texts used in this study were structurally often have difficulty identifying the theme
simple, we are convinced that work with of a story (Lehr, 1988). We found that this
materials such as these is an appropriate be- is true especially of children with learning
ginning step in the development of a sophis- disabilities. In one study, we worked with
ticated understanding of main idea and in children who had been classified as reading
using this comprehension skill in a variety disabled by the private school for children
of more complex materials. with learning disabilities that they were at-
tending (Williams, 1993). Their test scores
It turns out that starting out by trying to fell within the normal range of intelligence,
teach students to get the main idea was not and their reading levels were at least 2 years
the simplest choice we might have made. below what was expected for their age.
What we were able to do, we decided, was Their mean age was 13½ years. A group of
to provide the students with a model or students without disabilities drawn from the
template. We could not tell students directly fourth, fifth, and sixth grades of a regular
how to find the main idea, but we could private school was matched with the group
provide a clear and simple pattern that with reading disabilities on reading level
could be used later as a standard for com- (middle of the sixth grade). Their mean age
parison. What main idea comprehension was 10½ years. Another group of students
amounts to, on this level, is categorization without disabilities was drawn from the sev-
(Williams, 1984). And categorization de- enth and eighth grades, matched on age to
pends strongly on domain knowledge. The the group with reading disabilities. Their
textual cues that we identified were content reading comprehension was substantially
dependent. That is, in our sample para- higher than that of the other two groups.
graph, readers could be directed to circle the Students listened to a story (West, 1953),
most frequent words (“cowboys”) as an aid summarized it, and answered questions
to determining what the paragraph was about its theme in a structured interview
about. However, unless they knew the con- (they listened so that any difficulties with
cept of “job” and could identify the activity lower-order reading skills would not con-
listed in each sentence as a job, they would found our findings about comprehension).
not get the main idea. This is a simple illus- The students with learning disabilities were
tration of how content and structure inter- comparable to the reading-level-matched
act in comprehension. A complete instruc- younger students without disabilities on
tional model must provide students with summarizing, predicting, and answering
domain knowledge as well as explicit in- questions. A close analysis of the students’
struction about text structure. protocols suggested that there was one dif-
ference between the groups: In their sum-
Comprehension of Narrative Text maries, the students with learning disabili-
ties imported more irrelevant and
What corresponds to finding the “main implausible information into their sum-
idea” in narrative text? The most analogous maries than did the students without dis-
task is finding a theme of a story. A theme is abilities. This suggests that they did not
sometimes expressed in terms of a concept build up an accurate representation of the
such as “friendship,” that is, as a relation- story. The inaccurate representation inter-
ship among story components in a form fered with their getting the point. The fact
that is abstracted from the specific story that students with reading disabilities seem
context. Our definition adds another ele- to have a more difficult time in identifying
ment: A theme is expressed as the concept themes than do the students without disabil-
along with a commentary, either evaluative ities does not mean that no students without
or not, for example, “Some people steal”; disabilities have difficulty with this task; in-
or “People who are upset sometimes get dis- deed, our data indicated that students with-
tracted” (Williams, Brown, Silverstein, & out disabilities demonstrated a wide range
deCani, 1994). of performance.

298 EFFECTIVE INSTRUCTION

Teaching Theme Identification pointed out earlier, it is difficult to provide
any explicit strategic routine for determining
Theme is usually found to be the most diffi- a main idea. However, a theme is derived
cult story component to teach (Dimino, from the basic plot components of a story,
Gersten, Carnine, & Blake, 1990; Gurney, which can be taught, and a series of generic
Gersten, Dimino, & Carnine, 1990), even to questions that apply generally (starting with
students without disabilities (Singer & Don- plot questions and going beyond those) can
lan, 1982). We decided to develop an in- be taught as a strategy to guide theme identi-
structional program that would focus fication. Dorfman and Brewer (1994), who
specifically on this aspect of narrative un- worked with fables, proposed that in order
derstanding. We began by considering cur- to identify a theme, one must attend to two
rent reading comprehension instruction. of the basic plot elements: the central event
Most of today’s instruction is constructivist; and the outcome. Given the event, one eval-
in this view, each reader brings a unique uates the outcome in terms of one’s moral
knowledge base to the reading of a text and understanding; this evaluation is essentially
ends up with a unique interpretation of the a moral judgment. The combination of the
text. The instruction that follows from this plot components and the evaluation results
type of approach is typically organized in the theme. Our definition of theme is more
around discussion in which students con- encompassing than this, but we believe that
tribute their individual interpretations so the Dorfman and Brewer model applies even
that all can expand and refine their own in stories whose themes are not evaluative. In
meaning construction (Allington, Guice, any event, we decided to focus on evaluative
Michelson, Baker, & Li, 1996). Teachers themes in our research because of their sim-
serve as facilitators who contribute their plicity and their ubiquity in children’s litera-
own interpretations, without imposing them ture.
on the group. This constructivist paradigm
has been found to be successful for many The purpose of our instructional program
students (Allington et al., 1996). is to help students learn about the concept of
theme, identify theme in stories, and apply
However, this is relatively unstructured in- themes to real life. The instruction follows
struction. In addition, it presumes that all the tried-and-true paradigm of teacher ex-
students have stable knowledge bases and in- planation and modeling, guided practice,
terpretations to begin with, so that the class and independent practice. It focuses on
discussions can effectively modify and refine teaching plot-level components via organiz-
the interpretations and understanding of in- ing (schema) questions, as previous studies
dividual students. We decided that this con- have done. Then it teaches teaching theme
structivist approach would not fully meet the identification via additional questions. A fi-
needs of students with learning disabilities, nal set of questions helps students generalize
who have been shown to respond well to the theme to relevant life situations. We call
structured, direct instruction (Simmons, our set of questions the Theme Scheme. We
Fuchs, Fuchs, Mathes, & Hodge, 1995). use simple stories with single, clear and ac-
Thus we designed our Theme Identification cessible themes. Some of the theme concepts
program to incorporate constructivist goals we have used include perseverance, coopera-
of comprehension instruction with an in- tion, greed, and honesty. All our themes are
structional approach that is effective for stu- of the evaluative commentary type and are
dents with learning disabilities and others at expressed in a simple, common format: “We
risk for academic failure. That is, our pro- should cooperate,” “We should not be
gram emphasized the holistic and construc- greedy.”
tive nature of the comprehensive process and
the importance of integrating text meaning The original version of the program con-
with concepts and experiences that are per- sisted of a series of 12 40-minute sessions,
sonally meaningful, but it also acknowl- each organized around a single story and
edged the demonstrated value of structured, comprising five parts as outlined herein. The
direct instruction for poor readers. 12 stories were taken from four basal reader
series. In most cases, the stories originally
There is one important difference between appeared in trade books. Five of the stories
teaching main idea and teaching theme. As exemplified a single theme, “We should per-

Teaching Text Structure to Improve Reading Comprehension 299

severe.” Each of the other seven stories ex- portant story components and derive the
emplified a different theme concept, such as thematic material. Over the course the les-
cooperation, responsibility, and respect for son series, the teacher provides opportuni-
others, all expressed in the theme format (We ties to practice these questions so that stu-
should . . . ), described earlier. dents can recall them on their own.

1. Prereading discussion about lesson The first three organizing questions focus
purpose and story topic. In the first part of on the important story components from
each lesson, theme is defined as a lesson you which a theme concept will be derived:
can learn from a story, the value of under- main character, central event, and outcome.
standing themes is discussed, and back- The questions are:
ground for the specific story for that lesson is
introduced, including its relevance to per- ț Who is the main character?
sonal experiences. This instruction is heavily ț What did he or she do?
scaffolded, with teachers initially modeling ț What happened?
each step and students gradually taking on
more responsibility. In the first three lessons, These questions direct students to focus on
teachers define theme and lead the discussion the important information and enable them
on the importance of theme and the story to extract and organize important story
topic, making associations between the story components independently. Again, instruc-
and personal experiences. Starting in lesson tion is scaffolded.
4, the students offer definitions of theme and
lead the discussions themselves. As is the The final two organizing questions are
case throughout the lessons, only a general designed to encourage the students to make
outline is given to the teachers, who use their the judgments that, when combined with
own expertise in developing discussions and the theme concept, lead to theme identifica-
guiding the instruction. tion. These questions are:

2. Reading the story. Next, the teacher ț Was this good or bad?
reads the story aloud while students follow ț Why was this good or bad?
along with their texts (so that decoding dif-
ficulties do not interfere with comprehen- Although teachers model their responses to
sion). At various points during the reading, the first four questions for four lessons, the
the teacher imposes questions. These ques- final question—Why was this good or
tions are designed to encourage students to bad?—requires teacher modeling through
process the text actively (to make associa- lesson 7. Also through lesson 7, teachers
tions between their own knowledge and the model the way in which the answers to the
text information and to clarify the text in- five questions lead to a theme, and they
formation). The teacher asks the students to state the theme. After lesson 7, responsibili-
make predictions about what would happen ty for identifying and stating the theme is
next in the story and to explain major story gradually transferred to the students. Teach-
events. Student responses are discussed, and ers provide feedback to help the students in
students are encouraged to ask their own this process.
questions.
4. Identification of the theme in standard
After reading the story, the class discusses format. Students next learn to state the
the main points and reads a summary high- theme in a standard format defined as a
lighting the main events and outcome. This is “should” statement. Teachers model two
done because students with learning disabili- generic statement frames:
ties are particularly likely to have trouble
identifying the important story components ț (Main character) should have (should not
(Wong, 1984), and their story comprehen- have)
sion is often idiosyncratic (Williams, 1993). ___________________________________.

3. Discussion of important story infor- ț We should (should not)
mation using organizing (schema) ques- ___________________________________.
tions. Teacher and students discuss five
questions designed to help organize the im- The first frame puts the theme into the
should format. The second theme applies

300 EFFECTIVE INSTRUCTION

the theme to situations and people in gener- which can be considered a near transfer task
al rather than just to those in the story. Stu- (Brown & Palincscar, 1989). Finally, they
dents practice stating the theme in this for- were better at identifying the theme of a
mat. novel story whose theme had not appeared
at all during the instruction, a task we con-
5. Application of theme to real-life expe- sidered to qualify as far transfer. Both the
riences. The last two questions help extend students with mild disabilities and the stu-
the theme to specific and often personal dents without disabilities performed at simi-
real-life scenarios: lar levels.

ț To whom would this theme apply? Encouraged by these results but recogniz-
ț When would it apply? (In what situa- ing the preliminary nature of the findings
(they were based on a no-treatment con-
tion?) trol), we turned to the evaluation of our
comprehension instruction for students with
In this step, too, instruction proceeds by more severe learning disabilities. It was for
means of scaffolding. For each story, more those students, who do not respond well to
explicit forms of the questions are also in- normal classroom instruction, that our pro-
cluded in the lesson, elaborating on “who” gram was really designed.
and “in what situation,” to be used as
prompts when necessary. Second Evaluation

Evaluation of the Program Here we compared the performance of stu-
dents who were given our Theme Identifica-
First Evaluation tion program with students who were given
more traditional instruction that empha-
Our first evaluation involved fifth- and sized vocabulary acquisition and plot-level
sixth-grade students in eight New York City comprehension (the Story Comprehension
classrooms that included both normally program) (Williams et al., 1994). Twelve
achieving students and those with learning seventh- and eighth-grade special education
disabilities that were mild enough to permit classrooms in a small city near New York
mainstreaming (Williams et al., 1994). This were randomly assigned to each treatment.
evaluation allowed us to assess the effective- All students received all their instruction in
ness of the Theme Identification program their own classrooms; they were not main-
initially with children who were making sat- streamed. All had been certified by the
isfactory progress. Our assumption was that school as having learning disabilities, al-
any difficulties we found would not be at- though many of them had IQ scores below
tributable to the children’s learning prob- 85, the usual minimum for a classification
lems but, rather, would lie in the program. as having learning disabilities. As is our cus-
In this study, students were taught by their tom, we did not impose any further criteria
own teachers, in classrooms that had been on this school-identified group because we
randomly assigned either to receive the in- feel that instructional studies are most use-
structional program or to serve in a control ful if conducted in ecologically valid settings
condition receiving no special instruction. (i.e., in actual classroom situations as they
exist in schools).
This initial study provided positive evi-
dence for the effectiveness of our instruc- The students in the Theme Identification
tional program. Specifically, the students in classrooms performed significantly better
the Theme Identification classrooms under- than the students in the traditional-instruc-
stood the concepts of theme and the con- tion classrooms. First, they understood the
cept of perseverance better, both of which concept of theme and the concept of perse-
represented content that they had been ex- verance better, and they were also better at
plicitly taught. They also did better on ap- identifying the theme of a previously un-
plication of the perseverance theme (i.e., heard perseverance story. Thus, the Theme
generating a story with that theme), which Identification program aided students in
also had been explicitly taught. And, they their comprehension and promoted near
were better at identifying the theme of a transfer. But they were not superior at ap-
previously unheard perseverance story,

Teaching Text Structure to Improve Reading Comprehension 301

plying the perseverance theme. And when would engage the students, such as draw-
we asked the students to identify the theme ing, role playing, and song writing.
of a previously unheard story that had a
novel theme, one that was not represented This evaluation involved 10 classrooms
at all during the instruction, the two in- for students with learning disabilities in
structional treatments did not differ. Thus, three junior high schools in New York City.
these students with severe learning disabili- The classrooms were randomly assigned to
ties, unlike the students in the first study, the Theme Identification program or to the
did not demonstrate any transfer to stories Story Comprehension program.
with novel themes.
Our efforts proved successful. In this
However, it should be pointed out that study, the students in the Theme Identifica-
while it might seem that we achieved only a tion classes were significantly better than
modest degree of generalization, the results the others, not only on the posttest mea-
actually demonstrated a level of transfer sures on which there had been differences in
that represents substantial achievement for the previous study, but also on application
students with severe learning disabilities of the instructed themes and on identifica-
(Pressley & McCormick, 1995). We con- tion of the theme of a new story whose
cluded that the students with learning dis- theme had not been part of the instruction
abilities were able to respond positively to (“We should be honest”). Thus students
an integrated, well-structured approach to with severe learning disabilities did demon-
comprehension. strate far transfer. The effect sizes for the
comparisons that were significant ranged
Third Evaluation from 1.18 (Theme Identification of a novel
theme) to 5.93 (Theme Identification of an
We decided to work further on our program instructed theme). We believe that it was the
because it appeared as if it could be modi- combination of the sequencing of the
fied so that it would lead to even greater lessons as recommended in the literature on
transfer. To accomplish this goal, we made transfer plus the explicit and highly struc-
one major change in the program (Wilder & tured instruction, including the Theme
Williams, 2001). In its original version, al- Scheme, that were the effective elements of
most half of the instruction focused on a our instruction.
single theme and the rest included one in-
stance each of several other themes. Tradi- Fourth Evaluation
tional recommendations concerning train-
ing for transfer propose that transfer be We conducted one final study, to determine
built into the original instruction by using whether our program could be adapted suc-
multiple instances in a variety of contexts cessfully for use with elementary school stu-
(Pressley et al., 1990). Following these rec- dents in inclusion classrooms (Williams et
ommendations, we included three themes in al., 2002). In this study, 10 intact inclusion
the instruction, each one exemplified in four second- and third-grade classrooms in
stories. These themes were “We should not Harlem were assigned randomly to the
prejudge,” “We should be ourselves,” and Theme Identification Program or the Story
“We should keep trying.” We presented Comprehension program. The only substan-
these themes in a sequence that made it im- tive modification we made was in the mate-
possible to predict what the theme of the rials—we selected stories that we thought
next story would be. would be especially appealing to younger
children. The three theme concepts we
We also made some other changes in the worked with were perseverance, greed, and
program. We added three questions to the honesty. In this study the classrooms whose
Theme Scheme (making it a 12-step pro- students had been given the Theme Identifi-
gram). We added one question at the plot cation program showed superiority on mea-
level (“What was his/her problem?”, and sures of explicit teaching; they scored higher
two redundant, reinforcing questions, “The on the concept of theme and on theme con-
theme of this story is . . .” and “In what sit- cepts. They also showed evidence of near
uation will . . . (not) help?” We also includ- transfer, that is, they were better at identify-
ed additional activities that we thought ing the three instructed themes in the (nov-

302 EFFECTIVE INSTRUCTION

el) posttest stories. Effect sizes for the com- Also, there is no consensus in the field of
parisons that were significant ranged from comprehension instruction as to what con-
0.68 (Theme Concepts) to 2.71 (Concept of stitutes an acceptable level of performance,
“Theme”). However, these students showed as there may be in decoding instruction (Al
no evidence of far transfer in that they were Otaiba & Fuchs, in press). With these
not superior on questions dealing with caveats in mind, we determined criteria for
theme identification or applications when designating nonresponders that included
dealing with a story whose theme concept low performance on either explicit teaching
(cooperation) had not come up during in- or transfer. We identified 10 nonresponders
struction. In this respect they were different in grade 2 and four in grade 3. We found
from the middle school students with severe that the only student characteristics that
learning disabilities that we had studied were associated with nonresponding were
(Wilder & Williams, 2001). This discrepan- scores on reading and listening tests. There
cy in outcomes indicates the difficulty that was no significant relationship between
abstract thinking poses for younger children nonresponders and special education status
(Williams, 1998). in either grade. That is, the probability that
students who had learning disabilities or
Responsiveness to Instruction who had been referred for evaluation for
special education were nonresponders was
We posed an additional question in this last no different from the probability that stu-
study. Was the Theme Identification pro- dents without disabilities were nonrespon-
gram effective for all students? This ques- ders.
tion is especially important in school set-
tings in which there is a wide range of Students also gained from the instruction
ability levels (including children who have presented in the comparison program,
been identified as having a learning disabili- which also featured carefully selected sto-
ty or who have been referred for evaluation ries, well-organized lessons, and substantial
for special education). Full inclusion intensi- discussion. Substantial differences between
fies the demand for a diversity of effective the two instructional programs appeared on
instructional treatments. We found that the transfer measures, suggesting that to pro-
program was effective for students at high-, mote generalization, more direct and struc-
average-, and low-achievement levels, in- tured instruction is valuable. We believe
cluding those with learning disabilities or that current recommendations to focus
who had been referred for special education comprehension instruction on discussion
evaluation. The program was effective at should be supplemented with further rec-
both second and third grade. Given that on ommendations concerning choice of content
the pretest the basic theme comprehension and its organization, as well as direct in-
scores were essentially 0 at both grade lev- struction.
els, the significant grade effects on the
posttest indicate that the program was more In summary, our Theme Identification
effective for third-graders. However, there program helped students learn the funda-
was no indication that the second-graders mental aspects of theme comprehension and
had reached a performance asymptote; it is to generalize what they had learned. The
likely that they could have attained a higher program was effective for students at all
level of performance, perhaps as high as the achievement levels, and also for students
third-graders, if they had had additional in- with learning disabilities. It was effective,
structional time. with appropriate adaptations, from second
grade through junior high school. Across
There were, as there usually are, a few the studies, effect sizes for significant effects
children who were not responsive to the were almost always substantial.
program. It is instructive to examine indi-
vidual differences in responsiveness with an Informational Text
eye toward improving the program or de-
ciding when to use it. Of course, the pro- Most recently my students and I have re-
gram is short and circumscribed in nature. turned to a consideration of informational
text. Most children do not encounter much

Teaching Text Structure to Improve Reading Comprehension 303

of this type of text until the third or even the graphs) but focuses on content and not on
fourth grade, at a time when they are ex- text structure and also with a no-treatment
pected to use reading as a tool for learning control group. Preliminary data analysis in-
new content. In the last several years the dicates that the compare–contrast program
field has recognized the need for students to is successful in teaching the strategies and in
be exposed to informational text in the ear- helping students acquire proficiency in con-
ly elementary grades (Caswell & Duke, structing, orally and in writing, well-struc-
1998) and has called for research in this tured comparative statements in their sum-
area (Pearson & Duke, 2002). Fortunately, mary of text. Moreover, there is indication
instruction does not have to start from of transfer, in that the effects are seen both
scratch. As part of our research program, on text that involves animals and on text
Kristen Lauer has found in her dissertation whose content is unrelated to animals. We
study that as early as the second grade, stu- believe that our very direct and structured
dents are sensitive to informational text approach aids children in acquiring profi-
structure; structure makes an independent ciency in reading informational text in the
contribution to comprehension, along with same way as it has been shown to be suc-
content familiarity and reading ability. cessful for narrative text.

Certain types of informational text lend Summary
themselves to an instructional approach
similar to the one we took in our Theme Our evaluation studies have clearly indicat-
Identification program, in that, in contrast ed that our instructional programs are effec-
to main idea, there are specific structures tive with respect to achievement. They have
that can be taught. Our work involves the been successful in helping children with
development of instruction that focuses learning disabilities, even severe disabilities,
solely and intensively on a single structure, to acquire higher-order comprehension
compare–contrast, and that provides ele- skills. Students also can generalize what
mentary school students with direct instruc- they have learned to other similar texts. On
tion on how to understand that structure. the basis of these findings, it appears certain
As content to teach, we have chosen the that direct instruction that provides children
classification of animals into the five classes with knowledge about text structure, and
of vertebrates (which aligns with the New how to use that knowledge strategically, is
York City Core Science Curriculum Stan- an effective way to improve their compre-
dards for the elementary grades). hension. Simply providing opportunities for
exposure to particular types of text is im-
Kendra Hall, in her dissertation study, portant but far from the whole story. Ex-
has examined a nine-lesson (15-session) plicit instruction is valuable.
program in second-grade inclusion class-
rooms that include students with learning We have had positive and enthusiastic re-
disabilities and students who have been re- sponses from both teachers and students.
ferred for evaluation for special education. Informal feedback from students indicates
The instruction includes the teaching of sev- that they enjoy the lessons. Teachers’ re-
eral strategies, such as the use of clue words sponses on evaluation questionnaires indi-
(e.g., “alike” and “but”), graphic organiz- cate that they deem the programs educa-
ers, and questions that outline the text tionally beneficial and enjoyable. They like
structure (What two things is this paragraph the explicitness, repetition, and organiza-
about? How are they the same? How are tion of the programs (Williams, 2002).
they different?) Each lesson includes reading
and discussing the content of informational In conclusion, with appropriate materials
trade books, analysis of short, well-struc- and methods, students with learning disabil-
tured, compare–contrast paragraphs, work ities who have reading problems can achieve
on content vocabulary (e.g., oxygen and competence in higher-order comprehension.
warm-blooded), and writing as well as read- The materials and methods that are effective
ing practice. for these students are also effective with
other low-functioning students—indeed,
The study compares this program with they are effective with all levels of students
more traditional instruction that involves
the same materials (trade books and para-

304 EFFECTIVE INSTRUCTION

in the classrooms that we have worked in. It struction, and social interaction. Review of Edu-
is encouraging to see that students with cational Research, 67, 271–299.
learning disabilities respond well when a Lehr, S. (1988). The child’s developing sense of
well-structured and integrated approach to theme as a response to literature. Reading Re-
comprehension is used and to realize that search Quarterly, 23, 337–357.
there is no need to limit instruction for these Mandler, J. M., & Johnson, N. S. (1977). Remem-
students to low-level tasks. brance of things parsed: Story structure and re-
call. Cognitive Psychology, 9, 11–151.
References National Reading Panel. (2000). Teaching children
to read: An evidence-based assessment of the sci-
Allington, R., Guice, S., Michelson, N., Baker, K., entific research literature on reading and its impli-
& Li, S. (1996). Literature-based curricula in cations for reading instruction. Washington, DC:
high poverty schools. In M. F. Graves, P. van den National Institute of Child Health and Human
Broek, & B. M. Taylor (Eds.), The first R: Every Development and U.S. Department of Education.
child’s right to read (pp. 141–162). New York: Pearson, P. D., & Dole, J. A. (1987). Explicit com-
Teachers College Press. prehension instruction: A review of research and
a new conceptualization of instruction. The Ele-
Al Otaiba, S., & Fuchs, D. (In press). Characteris- mentary School Journal, 88, 151–165.
tics of children who are unresponsive to early lit- Pearson, P. D., & Duke, N. K. (2002). Comprehen-
eracy intervention. Remedial and Special Educa- sion instruction in the primary grades. In C. C.
tion. Block & M. Pressley (Eds.), Comprehension in-
struction: Research-based best practices (pp.
Baumann, J. F. (1984). The effectiveness of a direct 247–258). New York: Guilford Press.
instruction paradigm for teaching main idea com- Perfetti, C. A. (1985). Reading ability. New York:
prehension. Reading Research Quarterly, 20, Oxford University Press.
93–115. Perfetti, C. A., Marron, M. A., & Foltz, P. W.
(1996). Sources of comprehension failure: Theo-
Bean, T. (2000). Reading in the content areas: So- retical perspectives and case studies. In C.
cial constructivist dimensions. In M. L. Kamil, P. Cornoldi & J. Oakhill (Eds.), Reading compre-
B. Mosenthal, P. David Pearson, & R. Barr hension difficulties: Processes and interventions
(Eds.), Handbook of reading research (Vol. 3, pp. (pp. 137–165). Mahwah, NJ: Erlbaum.
629–644). Mahwah, NJ: Erlbaum. Pressley, M. (2000). What should comprehension
instruction be the instruction of? In M. L. Kamil,
Brown, A. L., & Palincsar, A. M. (1989). Guided, P. B. Mosenthal, P. David Pearson, & R. Barr
co-operative learning and individual knowledge (Eds.), Handbook of reading research (Vol. 3, pp.
acquisition. In L. B. Resnick (Ed.), Knowing and 545–562). Mahwah, NJ: Erlbaum.
learning: Essays in honor of Robert Glaser (pp. Pressley, M., & McCormick, C. (1995). Advanced
393–451). Hillsdale, NJ: Erlbaum. educational psychology. New York: Harcourt
Brace.
Caswell, L. J., & Duke, N. K. (1998). Non-narra- Pressley, M., Woloshyn, V., Lysynchuk, L. M., Mar-
tive as a catalyst for literacy development. Lan- tin, V., Wood, E., & Willoughby, T. (1990). A
guage Arts, 75, 108–117. primer of research on cognitive strategy instruc-
tion: The important issues and how to address
Dimino, J., Gersten, R., Carnine, D., & Blake, G. them. Educational Psychology Review, 2, 1–33.
(1990). Story grammar: An approach for pro- Rosenblatt, L. M. (1978). The reader, the text, and
moting at-risk secondary students’ comprehen- the poem. Carbondale: Southern Illinois Univer-
sion of literature. Elementary School Journal, 91, sity Press.
19–32. Simmons, D. C., Fuchs, L. S., Fuchs, D., Mathes, P.
G., & Hodge, J. P. (1995). Effects of explicit
Dorfman, M. H., & Brewer, W. F. (1994). Under- teaching and peer tutoring on the reading
standing the points of fables. Discourse Process- achievement of learning-disabled and low-per-
es, 17, 105–129. forming students in regular classrooms. Elemen-
tary School Journal, 95, 387–408.
Gersten, R., Fuchs, L. S., Williams, J. P., & Baker, S. Singer, H., & Donlan, D. (1982). Active comprehen-
(2001). Teaching reading comprehension strate- sion: Problem-solving schema with question gen-
gies to students with learning disabilities: A re- eration for comprehension of complex short sto-
view of research. Review of Educational Re- ries. Reading Research Quarterly, 17, 166–186.
search 71, 279–320. Swanson, H. L, & Hoskyn, M. (1998). Experimen-
tal intervention research on students with learn-
Gurney, D., Gersten, R., Dimino, J., & Carnine, D. ing disabilities: A meta-analysis of treatment out-
(1990). Story grammar: Effective literature in- comes. Review of Educational Research, 68,
struction for high school students with learning 277–321.
disabilities. Journal of Learning Disabilities 23, Taylor, M. B., & Williams, J. P. (1983). Compre-
335–348.

Hammill, D. D. (1990). On defining learning dis-
abilities: An emerging consensus. Journal of
Learning Disabilities, 23, 120–124.

Kucan, L., & Beck, I. L. (1997). Thinking aloud
and reading comprehension research: Inquiry, in-

Teaching Text Structure to Improve Reading Comprehension 305

hension of LD readers: Task and text variations. Research-based best practices (pp. 126–139).
Journal of Educational Psychology, 75, 743–751. New York: Guilford Press.
van Dijk, T. A. (1980). Macrostructures. Hillsdale, Williams, J. P., Brown, L. G., Silverman, A. K., &
NJ: Erlbaum. deCani, J. S. (1994). An instructional program
West, J. (1953). Cress Delahanty. New York: Har- for adolescents with learning disabilities in the
court Brace. comprehension of narrative themes. Learning
Wilder, A. A., & Williams, J. P. (2001). Students Disabilities Quarterly, 17, 205–221.
with severe learning disabilities can learn higher- Williams, J. P., Lauer, K. D., Hall, K. M., Lord, K.
order comprehension skills. Journal of Educa- M., Gugga, S. S., Bak, S. J., Jacobs, P. R., & de-
tional Psychology, 93, 268–278. Cani, J. S. (2002). Teaching elementary school
Williams, J. P. (1984). Categorization, macrostruc- students to identify story themes. Journal of Edu-
ture, and finding the main idea. Journal of Edu- cational Psychology, 94, 235–248.
cational Psychology 76, 874–879. Williams, J. P., Taylor, M. B., & Ganger, S. (1981).
Williams, J. P. (1992). Reading instruction and Text variations at the level of the individual sen-
learning disabled students. In M. J. Dreher & W. tence and the comprehension of simple exposito-
H. Slater (Eds.), Elementary school literacy: Crit- ry paragraphs. Journal of Educational Psycholo-
ical issues (pp. 157–181). Norwood, MA: gy, 73, 851–865.
Christopher-Gordon. Williams, J. P., Taylor, M. B., Jarin, D. C., & Milli-
Williams, J. P. (1993). Comprehension of students gan, E. S. (1983). Determining the main idea of
with and without disabilities: Identification of expository paragraphs: An instructional program
narrative themes and idiosyncratic text represen- for the learning-disabled and its evaluation (Tech.
tations. Journal of Educational Psychology, 85, Rep. #25). Research Institute for the Study of
631–641. Learning Disabilities, Teachers College, Colum-
Williams, J. P. (1998). Improving the comprehen- bia University.
sion of disabled readers. Annals of Dyslexia, 48, Wong, B. Y. L. (1984). Metacognition and learning
213–238. disabilities. In T. Waller, D. Forrest, & E. MacK-
Williams, J. P. (2002). Using the Theme Scheme to innon (Eds.), Metacognition, cognition and hu-
improve story comprehension. In C. C. Block & man performance (pp. 137–180). New York:
M. Pressley (Eds.), Comprehension instruction: Academic Press.

19

Enhancing the Mathematical
Problem Solving of Students
with Mathematics Disabilities

Lynn S. Fuchs
Douglas Fuchs

Converging evidence (Badian, 1983; Gross- one-step problems involving addition and
Tsur, Manor, & Shalev, 1996; Kosc, 1974) subtraction number facts.
reveals that 6–7% of the school-age popula-
tion suffers from mathematics disability Although straightforward to measure and
(MD). Despite its prevalence, MD has been study, these arithmetic story problems fail
the focus of less systematic study than has to represent the kind of tasks that increas-
reading disability (RD) (cf. Rasanen & Aho- ingly are incorporated in school curricula
nen, 1995). This relative neglect is unfortu- beyond the earliest grades. In fact, mathe-
nate. In school, mathematics skill is impor- matics education reform over the past 15
tant for success; after school, mathematics years has prompted schools to emphasize
competence accounts for employment, in- the development of more complex problem-
come, and work productivity even after in- solving capacity (e.g., Resnick & Resnick,
telligence and reading have been explained 1992; Rothman, 1995), with a measure-
(Rivera-Batiz, 1992). ment focus on performance assessments
that pose real-world problem-solving dilem-
Despite this relative neglect, important mas and require students to develop solu-
work on MD has been accomplished over tions involving the application of multiple
the past two decades. The literature de- skills. Therefore, generalizations from the
scribes functional arithmetic difficulties of MD literature to MD as it occurs in schools
students with MD and demonstrates how and generalizations to the kinds of mathe-
cognitive deficits are associated with the de- matical competence required in the real
velopment of arithmetic cognition (see world are tenuous. So, a focus on mathe-
Geary, 1993, for review). This literature has, matical competence beyond arithmetic and
nevertheless, focused disproportionately on arithmetic story problems is required to un-
acquisition of basic facts, with relatively few derstand, prevent, and remediate MD in its
studies on algorithmic competence. And, many forms, as it develops in schools and as
when problem solving has been studied it pertains to the real world.
(e.g., Jordan & Hanich, 2000), it has been
confined largely to linguistically presented Our research on mathematics serves to il-
lustrate an expanding focus that corresponds

306

Enhancing Mathematical Problem Solving 307

roughly to how mathematics curriculum has (1987), three variables contribute to prob-
evolved over time within schools. Our early lem-solving transfer. Students must (1) mas-
studies, illustrated by Fuchs, Fuchs, Hamlett, ter the rules for problem solution, (2) devel-
and Stecker (1990), were dedicated to math- op categories for sorting problems that
ematics operations (addition, subtraction, require similar solutions, and (3) be aware
multiplication, and division of whole num- that novel problems are related to previous-
ber, fractions, and decimals). As schools, ly solved problems. Research has substanti-
however, broadened their mathematics do- ated the importance of the first variable,
main, we adapted by focusing on mathemat- mastering rules for problem solution (e.g.,
ics concepts and applications (see, e.g., Fuchs Mawer & Sweller, 1982; Sweller & Cooper,
et al., 1997). Of course, these measurement 1985). As students master problem-solution
methods and this instructional content were rules, they allocate less working memory to
confined largely to discrete skills and artifi- the details of the solution and instead de-
cially structured word problems. Thus, be- vote cognitive resources to identify connec-
ginning in 1995, we reoriented our research tions between novel and familiar problems
program to mathematics problem solving, and to plan their work.
more broadly conceptualized and as increas-
ingly reflected in the schools. Cooper and Sweller’s (1987) second vari-
able suggests that schemas play a role in
In this chapter, we summarize our re- transfer. As defined by Gick and Holyoak
search program on mathematical problem (1983), a schema is a generalized descrip-
solving. We begin by describing our concep- tion of two or more problems, which indi-
tualization of problem solving and the cor- viduals use to sort problems into groups re-
responding measurement work we necessar- quiring similar solutions. The broader the
ily conducted to make the study of schema, the greater the probability that in-
mathematics problem solving possible. dividuals will recognize the connections be-
Next, we describe a series of studies con- tween familiar and novel problems and that
ducted with normally developing students transfer, therefore, will occur. To gain in-
to gain insight into how we might effect bet- sight into the role schemas play in transfer,
ter learning on challenging mathematical Cooper and Sweller questioned eighth-
problem-solving tasks for students with graders as they worked novel algebra prob-
MD. Then, we discuss recent studies in lems. The researchers coded responses in
which we applied those findings to students terms of whether statements reflected
with MD. In these studies, to qualify as hav- schemas (e.g., when faced with a new prob-
ing a MD, students had to score at least 90 lem, students reported thinking about how
on an intelligence test and at least 1.5 stan- an earlier problem had been solved) and
dard deviations below the mean on a math- demonstrated that schemas strongly influ-
ematics achievement test. ence performance on problems that fall
within the boundaries of those schemas.
Conceptualizing and Identifying Tasks to They also noted, however, that participants’
Study Mathematical Problem Solving schemas were disappointingly narrow. The
challenge in effecting transfer, of course, is
Mathematical Problem Solving to help learners develop broad schemas.

We conceptualized mathematical problem At the same time, it remains unclear how
solving as a transfer challenge, which re- to address Cooper and Sweller’s (1987)
quires students to apply knowledge, skills, third transfer-inducing variable: triggering
and strategies to novel problems (cf. Brans- awareness of the connections between train-
ford & Schwartz, 1999; Mayer, Quilici, & ing and transfer tasks. Prior work does,
Moreno, 1999). Effecting this form of trans- however, reveal the importance of such
fer can be especially difficult for primary- awareness. For example, using two sets of
grade children (Durnin, Perrone, & MacK- paired associates, Asch (1969) demonstrat-
ay, 1997) and for students with disabilities, ed that many students enjoyed no savings in
who have demonstrated generalization learning one pair that was identical across
problems (White, 1984). the lists; only participants who recognized
the pair as familiar realized the savings, but
As conceptualized by Cooper and Sweller once students were told that an old pair

308 EFFECTIVE INSTRUCTION

might be on the new list, all students stead, deliberately examines the task at
demonstrated the expected savings. In this hand and generates alternative solutions by
and related studies (e.g., Catrambone & considering ways in which the novel task
Holyoak, 1989; Gick & Holyoak, 1983; shares connections with familiar tasks. So,
Ross, 1989), performance increased when with high-road transfer, abstraction pro-
participants were cued to anticipate similar- vides the bridge from one context to the
ities across tasks. This finding demonstrates other; metacognition is the conscious recog-
how awareness of the potential connections nition and effortful application of that ab-
across novel and familiar tasks is a key in- straction across contexts.
gredient for transfer. Of course, to achieve
mathematical problem-solving transfer, it is Salomon and Perkins (1989) further de-
necessary to go beyond cuing by an external scribed two forms of high-road transfer.
agent. Instead, students must independently With forward-reaching high-road transfer,
activate searches for connections between the abstraction is generated in the initial
novel and familiar tasks. learning context; as learners engage in the
initial task, they consider other situations in
To broaden schemas and evoke indepen- which the abstraction might apply. By con-
dent searches for connections between trast, with backward-reaching high-road
transfer and familiar tasks without the need transfer, abstraction occurs in the transfer
for external cueing. Salomon and Perkins situation, where the learner thinks back to
(1989) provided the following theoretical previous tasks to search for relevant connec-
framework. They distinguished between tions and abstractions. As discussed by Sa-
two forms of transfer. Low-road transfer is lomon and Perkins, both forms of high-road
accomplished via extensive, varied practice; transfer provide a theoretical basis for
it occurs as a function of the automatic trig- teaching explicity for transfer. For example,
gering of well-learned, stimulus-controlled as teachers formulate the scientific factors
behavior in a new situation; using math relevant to electricity, they highlight how
facts to solve multi-step computation prob- these principles apply to other domains of
lems is an example of low-road transfer. science. This promotes forward-reaching
Mathematical problem solving, by contrast, transfer. Furthermore, when addressing sub-
is a form of high-road transfer, which in- sequent topics, teachers capitalize on the
volves the deliberate abstraction of princi- potential for backward-reaching transfer.
ples that apply across contexts or tasks. Here, teachers encourage students to search
High-road transfer requires individuals to their knowledge of scientific principles
formulate and search for these abstract con- gleaned from electricity for relevant analo-
nections between transfer and familiar gies. Salomon and Perkins asserted that
tasks. As Salomon and Perkins suggested, helping students (a) anticipate how abstrac-
the hallmark of high-road transfer is mind- tions may facilitate success with novel learn-
ful abstraction. (For mindfulness, we substi- ing tasks (as in forward-reaching tansfer)
tute the more familiar, educationally rele- and (b) conduct independent searches for
vant term “metacognition.”) relevant abstractions that apply across tasks
(as in backward-reaching transfer) repre-
To abstract a principle is to identify a sents an untapped instructional opportunity
generic quality or pattern across instances to explicitly teach students to transfer. We
of the principle. In formulating an abstrac- used this framework to study and enhance
tion, an individual deletes details across ex- mathematical problem solving.
amplars, which are irrelevant to the abstract
category (e.g., ignoring that an airplane is Tasks to Study Mathematical Problem Solving
metal and or that a bird has feathers to for-
mulate the abstraction of “flying things”). To identify tasks with which to study math-
These abstractions are represented in sym- ematical problem solving, we relied on this
bolic form and avoid contextual specificity conceptual framework. So, we developed
so they can be applied to other instances or measures of mathematical problem solving
across situations. Because abstractions, or along a continuum of contextual realism.
schemas, subsume related cases, they pro- We predicted that as problems achieved
mote transfer. With metacognition, an indi- greater verisimiltude, the challenge of iden-
vidual withholds an initial response and, in-

Enhancing Mathematical Problem Solving 309

tifying familiar problem solutions would in- Working with this input, we collectively
crease. Essentially, we hypothesized that developed one assessment at each grade. We
contextual realism is achieved as the variety piloted each test with three students who
of possible solutions needed to answer were entering and three who were exiting
problems increases, the nature of the re- the target grade. Based on the range of per-
quired operations becomes more varied, the formances and the students’ input about
amount of information described in the what they liked, disliked, and found confus-
problem situation increases, and the loca- ing, we modified these initial tests. We next
tion of that information is removed from developed a framework for creating the re-
the problem questions. Herein, we briefly maining five parallel forms at each grade
summarize our work in developing techni- level and then developed those remaining
cally viable measurement at a point toward tests. The following sequence then recurred
the difficult end of this continuum: three times. At each grade level, five experi-
The Mathematics Performance Assessment enced teachers completed each of the six as-
(Fuchs, Fuchs, Karns, et al., 2000). sessments. These adults, who were unfamil-
iar with the development process, identified
This measure, which incorporates six al- every skill they applied in answering the
ternate forms at each level (grades 2–6), questions; described inconsistencies in diffi-
requires students to apply 10 essential culty level and required skill applications
grade-level skills to a contextually realistic across parallel forms; and noted potential
situation. The assessment presents an intro- sources of confusion within the narrative
ductory narrative that includes nonessential and questions. Based on this input, we re-
details and irrelevant numbers, with multi- vised the tests. Finally, at each grade, four
ple related questions removed from the nar- students (two exiting and two entering the
rative. At each grade, we developed parallel target grade level) completed all six tests.
forms by first creating one preliminary Based on their responses and input, we
fourth-grade assessment for illustrative pur- made a final set of revisions.
poses. Next, we held a focus group meeting
at which teachers individually completed Each two- to three-page test begins with a
the illustrative assessment, learned to use multiparagraph narrative describing the
the scoring rubric, and divided into grade- problem situation. Each dilemma also pre-
level teams to score five sample protocols sents students with tabular and graphic in-
and to make suggestions for modifying the formation for potential application. The
scoring rubric and the basic structure of the problem includes questions that provided
illustrative assessment. Then, in grade-level students with opportunities to (1) apply the
teams, teachers identified 10 core skills to core set of skills, (2) discriminate relevant
incorporate into the assessment at their from irrelevant information in the narrative,
grade level. They began by reviewing the (3) generate information not contained in
statewide mathematics curriculum to select the narrative, (4) explain their mathematical
the 10 skills most essential for successful en- work, and (5) generate written communica-
try to the next grade and the 10 skills most tion related to the mathematics.
essential for successful entry into the grade
they teach. Then, in one large group, the These tests initially were scored accord-
teachers compared (a) the skills they speci- ing to the rubric shown adapted from the
fied as important for entering the next high- Kansas Quality Performance Assessment
er grade with (b) the skills identified by (Kansas State Board of Education, 1991).
teachers in that next higher grade as critical This rubric structured scoring along four
for successful entry. With this input, teach- dimensions (conceptual underpinnings,
ers returned to grade-level teams to finalize computational applications, problem-solv-
lists of core grade-level skills. Finally, in ing strategies, and communicative value).
grade-level teams, teachers identified and Each dimension was scored on a 6-point
rank-ordered 20 themes that represented holistic scale. In later work, we enhanced
real-life situations students might face now reliability by converting this scoring system
or in the next few years, were interesting, to award points on a fine-grained basis
could incorporate the 10 core skills, and ranging from 1 to 70. As shown by Fuchs
were age appropriate. and colleagues (in press-a) (n = 412), alpha
coefficient was .94; concurrent validity with

310 EFFECTIVE INSTRUCTION

the TerraNova (CTB/McGraw-Hill, 1997) ied math operations, and the inclusion of
was .67. nonessential details. Previous work demon-
strates the deleterious effects of longer text
Because Fuchs, Fuchs, Karns, and col- with more words and more verbs per sen-
leagues (2000) illustrated the deleterious ef- tence (Cohen & Stover, 1981; Cook, 1973;
fects of unfamiliarity with performance as- Helwig, Rosek-Toedesco, Tindal, Heath, &
sessments, we administer the task in the Almond, 1999; Jerman & Mirman, 1974)
following way. Prior to measurement, as well as multiple operational steps and ir-
testers deliver a 45-min lesson on how per- relevant information. The challenge, of
formance assessments are structured, strate- course, is to separate the effects of these po-
gies for approaching performance assess- tential sources of difficulty. We attempted to
ments, and scoring procedures. This lesson provide some insight on this matter by re-
first provides students an opportunity to porting separate scores for operations and
work on a performance assessment below problem solving, so that problem-solving
their grade level; then, using that assess- capacity was not confounded with opera-
ment, presents examples of student work to tional competence. In fact, the magnitude of
illustrate the topics. In all assessment situa- drop on both performance dimensions was
tions, text is read and reread aloud as re- comparable, suggesting that operational de-
quested. mands as well as informational richness
contribute to difficulty. Multidetermined
We had hypothesized that the variety of difficulty provides some support for concep-
possible problem solutions and the amount tualizing contextual realism (determined by
and location of the information needed to variety of possible problem solutions, oper-
set up the problem all contribute to contex- ational demands, and amount and location
tual realism. This, in turn, increases the dif- of information in the problem) as the source
ficulty students experience in recognizing a of challenge in mathematical problem solv-
novel problem as belonging to a familiar ing. It is also consistent with the notion that
problem type, for which they know a solu- transfer is the essential feature of mathemat-
tion method. In fact, we (Fuchs & Fuchs, ical problem solving, whereby contextual
2002) contrasted features of our perfor- realism increases the difficulty of classifying
mance assessments with Jordan and problems as belonging to categories for
Hanich’s (2000) arithmetic story problems which solutions are known. This measure-
task (i.e., linguistically presented one-step ment work provided the groundwork for
number fact problems). All indices of prob- studying methods to enhance mathematical
lem difficulty (i.e., number of words, sen- problem solving.
tences, words per sentence, verbs, numbers,
and math steps) were lower for arithmetic Enhancing Mathematical Problem
story problems than for the real-world Solving among Nondisabled
problem-solving measure. Each feature in- Students with Varying Histories
creases the difficulty of mathematics prob- of Mathematics Achievement
lem-solving tasks. In addition, we compared
student performance as a function of task Using this conceptual framework and rely-
(Fuchs & Fuchs, 2002). Among students ing on this measurement task, we conducted
with MD without comorbid RD, accuracy a series of three studies to identify methods
fell from 75% for arithmetic story problems for enhancing mathematical problem solv-
to 12% for real-world problem solving; for ing among students without disabilities with
students with comorbid MD and RD, per- varying mathematics achievement histories.
centages were 55 and 5. Thus, performance The first study described instruction and
differentials were dramatic when comparing student learning in mathematical problem
arithmetic story problems to our perfor- solving when teacher planning was in-
mance assessments. formed by ongoing performance assessment
feedback. The second study examined the
It is also interesting to consider the source effects of explicitly teaching students to
of the dramatic performance drop for stu- transfer, and the third study extended those
dents with MD between arithmetic and real-
world problem solving, where increasing
complexity was derived from greater text-
based demands, the need to apply more var-

Enhancing Mathematical Problem Solving 311

findings by investigating the added contri- One might expect knowledge about perfor-
bution of self regulation. In this chapter, we mance assessment to increase as teachers,
focus on the first two studies and refer over the course of several months, scored
briefly to the third. performance assessments, provided students
feedback on those problem-solving assess-
Teacher Planning and Student Learning with ments, and consulted with fellow teachers
Ongoing Performance Assessment Feedback to design instruction to enhance student
performance.
We (Fuchs, Fuchs, Karns, Hamlett, &
Karzaroff, 1999) were interested in explor- More noteworthy is the finding that as a
ing how teachers designed instruction to en- function of participating in the experimen-
hance their students’ mathematical prob- tal condition, teachers’ curricular focus ex-
lem-solving performance. Because teachers panded to incorporate a greater emphasis
were relatively unfamiliar with a problem- on mathematical problem solving. Com-
solving curriculum and with the perfor- pared to the contrast group, experimental
mance assessments that measure perfor- teachers decreased emphasis on math facts
mance on that curriculum, we supported and computation and increased focus on
teacher efforts—but in a relatively unobtru- problem solving, with ESs ranging from
sive way. We familiarized them with the 0.62 for math facts to 1.51 for problem
concept of performance assessment; we solving. Thus, in response to a relatively
helped them administer, score, interpret, modest intervention, teachers’ thinking
and provide student feedback on perfor- about mathematics curriculum changed.
mance assessments every 3 weeks; and we This echoes previous work on the effects of
released them to consult with fellow teach- statewide accountability programs incorpo-
ers every 3 weeks about how to improve rating performance assessment (e.g., Koretz,
their students’ problem-solving perfor- Barron, Mitchell, & Stecher, 1996; Koretz,
mance. The design of this study was experi- Mitchell, Barron, & Keith, 1996).
mental. We identified 16 teachers at grades
2–4, then, stratifying by grade level, we ran- Importantly, however, our study extended
domly assigned teachers to “experimental” previous work by showing how teachers’
and contrast groups. At the beginning of the actual instructional plans reflect changing
school year, we identified students in each curricular emphasis. As experimental teach-
class as above, at, or below grade level in ers, over three planning cycles, brain-
math and examined effects on student learn- stormed collaboratively with colleagues to
ing as a function of these achievement desig- identify methods for increasing student per-
nations and as a function of teachers’ study formance, they each committed themselves
condition. We used teacher as the unit of to many (i.e., 9–17) instructional ideas,
analysis. which represented three important instruc-
tional strategies. First, approximately one of
Results indicated that the modest level of every four ideas that teachers incorporated
support we provided did increase teachers’ into their plans was designed to expand
understanding of what mathematics prob- mathematical problem-solving perfor-
lem-solving performance assessment is and mance. And, importantly, some of these ac-
how feedback on students’ performance as- tivities, such as peer-mediated learning (e.g.,
sessment might be used to improve prob- King, 1991) and problem demonstrations
lem-solving performance. When asked to (e.g., Cooper & Sweller, 1987), resemble re-
develop a performance assessment, experi- search-based practices with demonstrated
mental teachers constructed problems that efficacy for promoting mathematical prob-
incorporated more features associated with lem solving. Second, as represented by their
performance assessment. When asked how use of test-like practice, teachers incorporat-
performance assessment might be used to ed extended mathematical problem-solving
formulate instructional decisions, experi- activities. This is noteworthy because ex-
mental teachers cited more strategies. Re- tended problem-solving activities offer stu-
sults were statistically significant and large, dents opportunity to discover the relations
with effect sizes (ESs) exceeding 1 standard among knowledge elements and problem
deviation. Of course, this is not surprising. features, which are important to the devel-
opment of problem solving, and because

312 EFFECTIVE INSTRUCTION

prior work (Stigler & Hiebert, 1997) sug- (as evidenced by their superior performance
gests that teachers typically avoid extended on the two similarly formatted and scored
mathematics activities, allocating as much measures) might apply to the novel mea-
as 96% of students’ time to practice on rou- sure. Essentially, superficial differences in
tine problems. The third strategy represent- the novel measure may have interfered in
ed in the teachers’ instructional plans was prompting metacognitive awareness of rela-
helping students demonstrate and commu- tions across problem situations (Brown,
nicate about the mathematical competence Campione, Webber, & McGilly, 1992). This
they already possessed: One of every four argument also seems plausible in light of ex-
instructional ideas was designed to improve perimental teachers’ failure to incorporate
students’ labels and explanations. Of instructional activities designed to help stu-
course, some may construe two of these in- dents analyze and articulate similarities and
structional procedure, clarifying methods differences among problem contexts. Such
for showing/explaining work and providing pattern-finding activities, in which students
test-like practice, as “teaching to the test” analyze contrasting cases, have been shown
and view these activities as cause for con- to enhance problem solving (Schwartz &
cern. After all, these two foci collectively Bransford, 1998). Thus, we hypothesize
represented half the teachers’ instructional that metacognitive awareness of the rela-
plans—or double the teachers’ use of more tions across problem-solving situations may
guided methods for extending students’ have mediated at-grade-level students’ prob-
mathematical problem solving (e.g., peer- lem-solving capacity. This hypothesis pro-
mediated learning or problem demonstra- vided the basis for future research described
tions). Moreover, as shown in previous later.
work on the effects of statewide account-
ability programs, teachers can structure Meanwhile, results for below-grade-level
these activities in ways that constitute students were disappointing. We found one
“coaching” (Firestone, Mayrowetz, & Fair- statistically significant effect on one prob-
man, 1998), which can lead to rapid, initial lem-solving measure for the communicative
gains that do not represent true learning value scoring dimension—the dimension
(Hambleton et al., 1995). most sensitive to the experimental teachers’
guidance about how to label and explain
It is not surprising, therefore, that effects work. Moreover, below-grade-level experi-
on student learning were not entirely sup- mental students actually performed some-
portive of the teachers’ methods and dif- what lower than contrast counterparts on
fered as a function of students’ achievement the most novel transfer problem, which dif-
histories. Superior problem solving was fered in format and scoring (ES = –0.28).
demonstrated by above-grade-level students These findings echo previous work showing
(ESs = 0.93–1.47). At-grade-level students the challenges associated with effecting
also demonstrated impressive learning—at mathematical problem solving among low-
least on the two measures most similar to achieving students. For example, Cooper
the performance assessments administered and Sweller (1987) found that students with
over the course of the study (ESs = previously low achievement required longer
0.76–1.15). Yet, on the most novel measure periods and more worked examples to ac-
(a commercial performance assessment) quire mathematical problem solving. Mayer
where transfer demands were greatest, ef- (1998) demonstrated that when teachers
fects were not statistically significant, with a used an approach consistent with the Na-
correspondingly modest ES of 0.30. For the tional Council of Teachers of Mathematics’s
two measures in which we observed effects, (NCTM’s) standards, students with higher
format and scoring were identical to the incoming achievement levels benefited
classroom performance assessments, where- more. In a similar way, Woodward and Bax-
as the format and scoring of the novel mea- ter (1997) showed that students with learn-
sure differed dramatically. This provides the ing disabilities or those who scored below
basis for supposing that students may have the 34th percentile on a standardized
suffered from inadequate awareness of how achievement test profited less from a prob-
the knowledge and strategies they had in lem-solving mathematics curriculum than
fact garnered from the classroom activities did average achievers. Clearly, for below-

Enhancing Mathematical Problem Solving 313

grade-level students, results were discourag- word vocabulary, additional question
ing, and they augured poorly for students posed, and placement in a larger problem-
with MD. solving context. In the Appendix, we illus-
trate how each type of superficial feature
Explicitly Teaching Students to Transfer changes a sample problem. As shown, the
manipulation of superficial problem fea-
One major conclusion from Fuchs and col- tures affects neither the structure of the
leagues (1999) was that teachers required problem nor the required solution. We pro-
greater structure and guidance to address vided students with practice in sorting novel
mathematical problem solving effectively. problems according to which superficial
Although teachers’ instructional plans re- problem feature had changed and in solving
flected impressive attempts to incorporate those novel problems. In this way, we pro-
innovative activities, some promising ap- moted the development of broader schemas,
proaches for guiding students in the devel- which permitted students to recognize that a
opment of problem-solving capacity were broader range of problems belonged to a
noticeably absent. And, for low-achieving unifying problem structure and thereby re-
and for some average-achieving students, quired the known solution. In terms of
more explicit forms of instruction may be metacognition, we explicitly taught students
required to develop problem-solving schema the concept of transfer. Moreover, we cau-
and metacognitive awareness. tioned students that, when faced with novel
problems, they should search for connec-
We therefore designed a treatment to tions to familiar problem structures and for
teach explicitly for transfer (Fuchs et al., in the superficial problem features they knew
press-a). Relying on the conceptual work of could change a problem without altering the
Cooper and Sweller (1987) and Salomon required solution. By broadening schemas
and Perkins (1989), the treatment attempt- and triggering awareness of the connections
ed to increase awareness of the connections between familiar and novel problems, this
between novel and familiar problems by (1) explicit transfer treatment addressed Coop-
broadening the categories by which students er and Sweller’s second and third transfer-
group problems requiring the same solution inducing varibles. The combination of prob-
methods (i.e., promoting a higher level of lem solution instruction with transfer
abstraction) and (2) explicitly prompting instruction addressed all three transfer-
students to search novel problems for these inducing variables.
broad categories (i.e., increasing metacogni-
tion). We combined this explicit transfer The study included four conditions, all of
treatment with explicit instruction designed which received the school district’s mathe-
to teach solution methods for different types matics program from which the four prob-
of problems. We contrasted the effectiveness lem structures had been selected and re-
of a combined treatment (transfer + solution ceived comparable amounts of math
instruction) to solution instruction alone instruction time. Teacher-designed (i.e., con-
and to teacher-designed. trol) instruction incorporated the district’s
curriculum, which addressed the four prob-
Solution instruction addressed Cooper lem structures. The other three conditions
and Sweller’s (1987) first variable, teaching were experimental, designed to incorporate
rules for problem solution. Our explicit explicit instruction (Carnine, 1997), heavy
transfer treatment, by contrast, was de- use of worked examples (Cooper & Sweller,
signed to effect abstraction and metacogni- 1987), and peer-mediated practice (Fuchs et
tion in the service of mathematical problem al., 1997). In one experimental condition,
solving. With respect to abstraction, our ex- students received only solution instruction.
plicit transfer treatment attempted to The other two experimental groups com-
broaden students’ schemas for sorting prob- bined solution with transfer treatment. Be-
lems that require similar solutions. Toward cause the solution treatment required 20
that end, we taught four types of superficial sessions and the transfer treatment entailed
problem features, which change problems another 10 sessions, combining treatments
without altering the problems’ structure or posed a dilemma about how to control the
solution (Ross, 1989). Our four types of su- amount of experimenter-designed instruc-
perficial problem features were format, key-

314 EFFECTIVE INSTRUCTION

tional time. So, we combined treatments in increasing metacognition). Support for this
two ways. In one combined condition, the transfer treatment was found on the near-
number of experimental sessions were equal transfer measure, in the dramatically superi-
for the solution and for the combined con- or growth of the combined treatment (trans-
dition; to accomplish this, we cut the num- fer + the full set of solution lessons) versus
ber of solution sessions in half. The other the solution condition alone without trans-
combined treatment permitted the full set fer instruction (ES = 1.45). Both conditions
problem solution sessions. Intervention last- had received all the solution lessons; the key
ed 16 weeks. difference was the provision of the transfer
treatment. Although the simple addition of
Not surprisingly, we found evidence to experimenter-controlled instructional time
support the utility of the problem solution in the full solution + transfer condition
instruction. On the acquisition test, which might explaing this effect, this explanation
required students to solve the four problem is unpersuasive in light of improvement on
structures when fresh cover stories consti- the acquisition measure, where the contrast
tuted the only source of novelty, the two between these two conditions was not sta-
groups that received more of the solution tistically significant (ES = 0.19).
treatment grew significantly and substan-
tially more than the combined group that Of course, the most convincing measure of
received less of the solution treatment (ES = learning in this study was the far-transfer
1.49–2.08). This provides evidence for the test—our real-world problem-solving mea-
importance of Cooper and Sweller’s (1987) sure, which posed questions with the greatest
first transfer-inducing, as operationalized by degree of novelty: an unfamiliar cover story,
our solution treatment. Of course, the ex- simultaneous manipulation of all four super-
tent of transfer required on the acquisition ficial problem features taught in the transfer
task was minimal because problems varied treatment, and inclusion of irrelevant infor-
from the content of the solution treatment mation as well as additional problem struc-
only in terms of cover stories. Thus, the ef- tures and content taught in the district’s cur-
fectiveness of the solution treatment is illus- riculum. Also, we minimized extraneous
trated better with findings the near transfer cuing by formatting the far transfer measure
measure, which not only varied cover sto- to resemble commercial achievement tests
ries but also manipulated, for each problem, and using unfamiliar testers. Results on this
one additional superficial problem feature far-transfer measure further substantiated
addressed in the transfer treatment. Results the additive effect of the transfer treatment:
indicated that although this near transfer Both groups that received the transfer treat-
measure was aligned more closely with the ment impressively outgrew the control group
transfer than the solution treatment, all (ES = 1.01–1.16). Moreover, the solution
three experimental conditions, including the condition, which did not receive the transfer
experimental group that did not receive the treatment, did not outgrow the control
transfer treatment, again improved signifi- group (ES = 0.39).
cantly and substantially more than the con-
trol group (ESs = 2.11–2.25). Moreover, the On the improvement score for each mea-
combined condition with the full set of solu- sure, we also examined interactions be-
tion sessions persuasively outgrew the com- tween study condition and the mathematics
bined condition with only half the solution grade-level status with which students be-
sessions (ES = 1.29). gan the study. We found no significant inter-
action. Therefore, effects were not mediated
In addition to lending support for the ef- by students’ prior achievement histories.
fectiveness of the solution treatment, the This is notable in light of previous work in-
near-transfer results also provided evidence dicating that transfer is more difficult to ef-
for the utility of the explicit transfer treat- fect among low-achieving students (Cooper
ment, which was designed to broaden the & Sweller, 1987; Fuchs et al., 1999; Mayer,
categories by which students group prob- 1998; Woodward & Baxter, 1997). Effects
lems requiring the same solution methods may have accrued for below-grade students
(i.e., promoting a higher level of abstrac- because our treatments were more explicit
tion) and to prompt students to search nov- and longer in duration than most previous
el problems for these broad categories (i.e., experimental interventions.

Enhancing Mathematical Problem Solving 315

Consequently, results strengthened previ- group of students with MD. This unaccept-
ous work (e.g., Cooper & Sweller, 1987) able rate of nonresponsiveness reveals the
showing the importance of instruction de- need to develop a strong foundation in
signed to teach rules for problem solution. problem solutions before instruction de-
At the same time, the study’s major contri- signed to promote transfer can contribute to
bution was the demonstration that explicit- learning.
ly teaching for transfer, by broadening the
categories by which students group prob- For the other treatments, which incorpo-
lems requiring the same solution methods rated the full set of lessons designed to teach
(i.e., promoting a higher level of abstrac- solution methods, some findings were en-
tion) and prompting students to search nov- couraging. Specifically, on the acquisition
el problems for these broad categories (i.e., measure, effects, although smaller in magni-
increasing metacognition), facilitates mathe- tude than for students without disabilities
matical problem solving. In subsequent (where ESs ranged between 1.49 and 2.08),
work (Fuchs et al., in press-b), we also doc- were in fact respectably strong (ESs =
umented how self-regulation enhances the .66–1.78). Thus, it appeared that the explic-
effectiveness of the combined solution/ it instruction, with strong reliance on
transfer treatment. worked examples and peer mediation, was
effective in helping students with MD mas-
Applying Findings to Students with MD ter the rules for problem solution, when
novel word problems (i.e., new cover stories
Our work also provides the basis for exam- and quantities) were presented in the exact
ining related effects on students with MD. same format in which they had been taught.
In this chapter, we focus on two databases.
First, our study examining effects of explic- Unfortunately, as problems became in-
itly teaching for transfer on nondisabled creasingly novel from those practiced during
students (as just described; Fuchs et al., in instruction, the discrepancy of effects be-
press-a) also incorporated a small sample of tween students with and without MD grew.
students with MD. This database permits an For near transfer, ESs for students without
estimate of the efficacy of the treatments disabilities who received transfer instruction
when that instruction is delivered in whole- ranged between 2.11 and 2.25; for students
class arrangement. Second, we recently con- with MD, the effect, while moderate, was
ducted an experiment (Fuchs, Fuchs, Ham- dramatically lower (0.45 in the transfer con-
lett, & Appleton, 2002) examining the dition with the full set of solution lessons).
effects of the combined solution and trans- And, on the far-transfer measure where the
fer treatments specifically on students with problem situation approximated real-world
MD when that instruction is provided in problem solving, there was essentially no ef-
small groups. Below, we summarize these fect for students with MD (ES = 0.07) where-
findings. as students without disabilities clearly trans-
fered their knowledge to the novel
Effects of Explicitly Teaching for Transfer in problem-solving situation (ES = 1.01–1.16).
Whole-Class Arrangment From this database, we concluded that we
needed to examine the efficacy of our math-
As just noted, our experiment examining ematical problem-solving treatment using a
the effects of solution and transfer instruc- service delivery mechanism that permitted
tion (Fuchs et al., in press-a) on students greater intensity.
without disabilities also included a small
group of students with MD. Among these Effects of Explicit Instruction
children, effects were least supportive for on Solutions and Transfer in
the combined condition that incorporated Small-Group Arrangment
only the partial set of solution lessons. In
that condition, 60% to 80% of students Consequently, in the following year, we con-
with MD (depending on outcome measure) ducted an experiment to examine the effec-
failed to progress more than the control tiveness of problem-solving tutoring, con-
ducted in small groups of two to four, with
fourth-grade students with MD. The

316 EFFECTIVE INSTRUCTION

lessons, which were identical to those used plications, and arithmetic story problems
in the larger whole-class study, integrated (Jordan & Hanich, 2000).
explicit instruction on problem solution
rules and transfer. To create a stringest test Results showed that problem-solving tu-
of efficacy, we assessed the contribution of toring, which provided students with explic-
this treatment to computer-assisted practice it instruction on rules for problem solution
on real-world problem-solving tasks, where and explicit instruction on transfer in small
students actually had intensive, guided, di- groups, did promote mathematical prob-
rect practice on the alternate forms of the lem-solving growth among students with
study’s far-transfer task. Students were ran- MD. Importantly, this growth was manifest-
domly assigned to problem-solving tutoring ed on the full range of measures, although
(or not) and to computer-assisted practice with smaller effects on the real-world prob-
(or not), creating four conditions: problem- lem-solving, far-transfer measure.
solving tutoring, computer-assisted practice,
problem-solving tutoring plus computer-as- On the acquisition and near-transfer mea-
sisted practice, and control, all of which re- sures, tutoring produced statistically signifi-
ceived the same mathematics curriculum cant improvement compared to a control
from which the four problem structures condition, in which students had received a
were selected and the same base unit on 3-week instructional unit on word prob-
problem solving. As with Fuchs and col- lems. On the acquisition measure, effect
leagues (in press-a), tutoring incorporated sizes comparing the tutoring condition with
explicit instruction (Carnine, 1997), heavy this control group exceeded 2 standard de-
use of worked examples (Cooper & Sweller, viations and clearly were in the same range
1987), and peer-mediated practice (Fuchs et as those documented in the earlier study for
al., 1997). Computer-assisted practice in- students without disabilities (2.11–2.25).
cluded incorporated guided feedback with On the near-transfer measure, effects were
motivational scoring. reliable and large (0.88)—nearly double
those revealed for students with MD when
Six special education teachers agreed to lessons were delivered in whole-class
have their fourth-grade students with MD arrangement (0.45).
participate. Teachers nominated 62 students
who met two criteria: (1) According to cu- In addition, although results for the real-
mulative records, their standard scores on world problem-solving, far-transfer mea-
an individually administered intelligence sure were not statistically significant, the
test were 90 or above, and (2) their special corresponding effect size exceeded a half
education teachers reported that they had standard deviation. This figure approximat-
MD. To these 62 students, we administered ed the effect size (0.63) associated with the
the Test of Computational Fluency (Fuchs, computer-assisted practice condition, even
Fuchs, Eaton, Hamlett, & Karns, 2000), to though the computer-assisted group spent
identify students (n = 40) who scored at all its experimental time practicing prob-
least 1.5 standard deviations below a re- lems analogous to the real-world problem-
gional normative sample. solving task. Moreover, the effect size of
0.61 for students with MD when lessons
Stratifying so that each condition was were delivered in small groups was substan-
represented approximately equally for each tially larger than the figure of 0.07 for stu-
teacher, we randomly assigned students so dents with MD when lessons were delivered
that 10 students were in each of four condi- in whole-class arrangement. Across the
tions: problem-solving tutoring with com- three tasks, therefore, results support the
puter-assisted practice, computer-assisted value of small-group instruction (Elbaum,
practice, problem-solving tutoring, and con- Vaughn, Hughes, & Moody, 2000), where
trol. Inferential statistics indicated group opportunities to respond, to seek clarifica-
comparability on students’ sex, free/reduced tion, and to obtain guided feedback are
lunch status, race, and problematic class- substantially greater than in whole-class in-
room behavior. We also documented com- struction. Of course, this effect size of 0.61
parable mathematics pretreatment perfor- for students with MD, when lessons were
mance across the treatment groups on tests provided in small groups, still pales in com-
of computational fluency, mathematics ap- parison to effect sizes exceeding 1 standard
deviation for students without disabilities

Enhancing Mathematical Problem Solving 317

when lessons were delivered in large assisted practice was conducted on tasks di-
groups. rectly paralleling the real-world problem-
solving measure. More notable was a simi-
Thus, this study provided the basis for lar ES of 0.60 on the acquisition measure
some optimism and some caution. On the along with a small ES of 0.31 on near trans-
one hand, results documented the efficacy fer, suggesting that some “downward”
of explicit instruction on problem solutions transfer occurred. Together, these ESs pro-
and transfer delivered within the context of vide the basis for additional research on
small groups and paralleled findings for stu- computer-assisted practice designed to guide
dents without disabilities, at 3 points on the students toward enhanced problem solving.
achievement continuum, when the problem- Future research should incorporate greater
solving program was delivered in large-class power, not only with larger samples but also
format (Fuchs et al., in press-a). As docu- with software that enhances the instruction-
mented by Fuchs and colleagues (in press-a), al value of the practice provided. We cur-
the effectiveness of the problem-solving pro- rently are undertaking this effort because of
gram resides in both components: Instruc- the strong appeal of improved mathematical
tion on rules for problem solution explains problem solving without the need for ex-
growth on the acquisition measure; the ex- pensive adult guidance.
plicit transfer component explains growth
on the near- and far-transfer measures. Of It is, nevertheless, interesting to note that
course, tutoring across both the problem so- with the explicit, small-group, peer-mediat-
lution rules and the transfer components ed instruction on rules for problem solution
was explicit (Carnine, 1997), with heavy and transfer in place, our computer-assisted
use of worked examples (Cooper & Sweller, practice condition provided no added value.
1987), peer mediation (Fuchs et al., 1997), On acquisition and near transfer, the com-
and small-group arrangements (Elbaum et parison between the tutoring treatments
al., 2000)—instructional features with with and without computer-assisted practice
proven efficacy for promoting reading. The revealed small effect sizes favoring the tu-
study extends previous work by document- toring without computer-assisted practice.
ing effects on mathematical problem solv- On far transfer, which paralleled the very
ing, a curricular area that has received rela- tasks students practiced via computer, the
tively little attention, especially for students ES favoring tutoring plus computer-assisted
with MD, and where previous work indi- practice was a disappointingly low 0.14.
cates that outcomes are difficult to effect From a research design perspective, this
among low-achieving students (Cooper & lends credence to the value of the tutoring
Sweller, 1987; Fuchs et al., 1999; Mayer, treatment because the combined treatment,
1998; Woodward & Baxter, 1997). Clearly, which failed to effect better growth, incor-
findings provide the basis for an important porated twice the amount of problem-solv-
message to practitioners: Teachers can im- ing worktime. Substantively, results are bol-
prove the mathematical problem-solving stered by the fact that, on acquisition and
performance of students with MD. On the near transfer, direct contrasts between the
other hand, effects on far transfer, although computer and the tutoring treatments reli-
respectable, were lower than for students ably favored tutoring: ESs comparing tutor-
without disabilities. Thus, clearly, addition- ing plus computer-assisted practice versus
al work is warranted to identify how to en- computer-assisted practice alone were 1.27
hance outcomes for students with MD. and 0.93; for tutoring without computer-as-
sisted versus computer-assisted practice
It is also interesting to consider the effects alone, 1.50 and 1.02. As revealed with the
of computer-assisted practice, as opera- software used in this study, computer-guid-
tionalized in this study. Our software pro- ed practice failed to provide a meaningful
vided intensive, instructional feedback with substitute for or addition to carefully for-
motivational scoring. Results on real-world mulated adult tutoring. Perhaps software
problem solving, although not achieving designed to provide better elaborated in-
statistical significance, did produce scores struction would effect better outcomes—or
that exceeded the control group by a no- might be used in conjunction with problem-
table 0.63 standard deviation. This might solving tutoring to reduce the amount of
have been expected given that computer-

318 EFFECTIVE INSTRUCTION

teacher time (an expensive resource) re- level of abstraction) and (2) explicitly
quired to enhance mathematical problem prompting students to search novel prob-
solving. lems for these broad categories (i.e., increas-
ing metacognition). Our experiments illus-
Conclusions trate the contribution of such explicit
instruction on transfer for students without
On the basis of this program of research, we disabilities; effect sizes were substantial on
draw some tentative conclusions about how the range of transfer measures examined.
to enhance mathematical problem solving Moreover, as shown by Fuchs and col-
among students with MD. We first offer leagues (in press-a), this approach also pro-
some specific recommendations based on motes important improvement for students
our work; then, we provide a general com- with MD. Of course, among students with
ment about the state of knowledge about MD, results are more impressive and reli-
“effective instruction” within special educa- able for near- than for far-transfer problem-
tion. We conclude by offering some direc- solving measures. Additional work is re-
tions for future study within the area of quired to identify strategies for increasing
mathematical problem solving for students the magnitude and range of problem-solv-
with MD. ing effects.

Specific Recommendations A General Observation about the Nature of
“Effective Instruction”
Our program of research reveals that to
promote mathematical problem solving for With respect to the state of knowledge
students with MD, a strong foundation in about “effective instruction,” we offer the
the rules for problem solution is necessary. following observation. Within special edu-
This means that children must master solu- cation, research on instructional practices is
tion methods on problems with low transfer strong on process variables that provide the
demands (i.e., identically worded problems foundation for instruction. These variables
that only vary cover stories and quantities). (e.g., Englert, 1984; Gersten, Woodward, &
This is necessary before we should expect to Darch, 1986; Haynes & Jenkins, 1986) in-
see mathematical problem solving, which clude, but are not limited to, modeling,
necessarily requires transfer to problems quick pace, frequent responding and high
presented with superficial differences that proportion of engaged time, overrehearsal,
make recognition of known problem solu- and guided feedback. Our work on mathe-
tions difficult. The need for mastery of matical problem solving has incorporated
problem solutions has been substantiated in these principles by, for example, relying on
earlier research (e.g., Mawer & Sweller, modeling via worked examples; encourag-
1982; Sweller & Cooper, 1985), showing ing quick pace via scripted lessons; requir-
that as students master problem-solution ing frequent response via choral responding;
rules, they allocate less working memory to providing a high proportion of engaged
the details of the solution and instead de- time, overrehearsal, and guided feedback
vote cognitive resources to identify connec- via small-group teacher-led instruction and
tions between novel and familiar problems peer-mediated practice; and encouraging
and to plan their work (Kotovsky et al., task-focused goals via accountability of in-
1985). dividual performance and self-regulation
procedures that include goal setting, self-
A second conclusion about enhancing scoring, self-charting of progress, and re-
mathematical problem solving, based on porting transfer opportunities and events.
our research, is more novel. It involves the
need for explicit instruction on transfer, de- As important as these variables are, how-
signed to increase awareness of the connec- ever, it is important for the field to acknowl-
tions between novel and familiar problems edge that they simply provide a foundation
by (1) broadening the categories by which over which substantive explanations must
students group problems requiring the same be superimposed. And special education re-
solution methods (i.e., promoting a higher search has allocated relatively little atten-
tion to principles for designing those expla-

Enhancing Mathematical Problem Solving 319

nations, beyond the notion of explicitness the domain of mathematical problem solv-
(e.g., Carnine, 1987). In our research on ing. In terms of number fact retrival and
mathematical problem solving, we have at- arithmetic story problems, Jordan and
tempted to optimize the quality of explana- Montani (1997) examined performance on
tions by incorporating concrete algorithms spoken number facts, spoken one- and two-
that make visible the conceptual underpin- step arithmetic story problems, and nonver-
nings behind operations (e.g., drawing items bal number facts as a function of the chil-
into “bags” to concretize step-up functions dren’s language and visuospatial profiles.
before relying on repeated addition or divi- The researchers showed that these tasks
sion to derive answers) and using everyday were differentially sensitive to variation in
examples that are accessible to a wide range cognitive ability. The language-impaired
of children’s experiences to illustrate princi- group (receptive vocabulary and grammatic
ples (e.g., exemplifying transfer with babies closure < 30th percentile) performed signifi-
learning to drink from a toddler cup, then a cantly lower than nonimpaired peers on sto-
glass, then a soda pop bottle). The chal- ry problems, but not on either number fact
lenges associated with constructing effective task. By contrast, the visuospatial-impaired
explanations cannot, however, be overesti- group performed comparably to the nonim-
mated. Given the constraints imposed on paired group across tasks. Unfortunately,
teachers’ time to plan, the need to provide consensus on this performance pattern is
clearly articulated explanations for teachers lacking in the literature. For example,
to borrow is great. We believe that the field Rourke (1993) identified two similar groups
needs to dedicate effort not only to studying of children with MD: those with verbal
principles of strong explanations but also to deficits (who had MD with reading and
build those explanations within specific aca- spelling deficits) and those with tactual-
demic domains. perceptual, visuoperceptual, motor, and rea-
soning difficulties (whose functional diffi-
Some Future Directions for Research culties were limited to math). In contrast to
Jordan et al.’s findings, however, Rourke’s
Extension to the literature on enhancing visuospatial-deficit group manifested a
mathematical problem solving is required; more pervasive pattern of mathematics dif-
we identify four areas for additional work. ficulties, which involved a larger number
First, as already stated, work is needed to and wider range of procedural errors than
identify principles of effective explanations did the verbal-deficit group. Based on the
and to build effective explanations that pro- literature, a wide set of cognitive correlates
mote mathematical problem solving. Sec- provide the basis for productive study. From
ond, continued research is required to arithmetic research, these include articulato-
strenghthen performance on far-transfer, ry speed, short-term memory, speed of re-
real-world problem-solving tasks. In our trieving information from long-term memo-
studies, we have documented effects of six- ry, and general speed of processing From
tenths of a standard deviation among stu- arithmetic problem solving as well as our
dents with MD; methods are required to own work on more extended problem solv-
boost these effect sizes so that they better ing, relevant cognitive correlates are
approximate effects of 1 or more standard metacognition, language-related variables
deviation observed for students without dis- associated with reading and listening com-
abilities. Intensifying instruction by, for ex- prehension difficulties, motivation and per-
ample relying on one-to-one tutoring or severance, visuospatial ability, generaliza-
configuring computer-delivered practice in tion deficits/inference-making difficulties,
effective ways represents one potential strat- and semantics.
egy for accomplishing this goal. Better de-
veloped explanations represent another pos- Finally (and in a related way), the effects
sible method. of mathematical problem-solving treatment
for subgroups of students with MD, who do
A third direction for future study address- and do not present comorbid RD, is re-
es the fact that relatively little is known quired. MD research typically fails to ac-
about the cognitive correlates associated count for reading competence in the study
with responsiveness to instruction within of mathematical problem solving. This is

320 EFFECTIVE INSTRUCTION

unfortunate because research suggests that tude, achievement, and instruction (pp. 37–75).
unique patterns of functional competence Boston: Kluwer Academic.
(e.g., Hanich, Jordan, Kaplan, & Dick, Carnine, D. (1997). Instructional design in mathe-
2001; Jordan & Hanich, 2000; Jordan & matics for students with learning disabilities.
Mondani, 1997) as well as unique patterns Journal of Learning Disabilities, 30, 130–141.
of verbal strength and visual-perceptual or- Catrambone, R., & Holyoak, K. J. (1989). Over-
ganizational weakness (e.g., Fletcher, 1985; coming contextual limitations on problem-solv-
Fuchs & Fuchs, 2002; McLean & Hitch, ing transfer. Journal of Experimental Psychology:
1999; Rourke & Finlayson, 1978; Siegel & Learning, Memory, and Cognition, 15,
Linder, 1994; Siegel & Ryan, 1989) underlie 1127–1156.
MD and MD + RD. In light of the semantic Cohen, S. A., & Stover, G. (1981). Effects of teach-
challenges associated with mathematics ing sixth-grade students to modify format vari-
problem solving, regardless of whether ables of math word problems. Reading Research
problems are accessed through reading or Quarterly, 16, 175–199.
listening, the verbal performance deficits of Cook, B. (1973, February). An analysis of arith-
comorbid students along with their more metic, linguistic , and algebraic structural vari-
pervasive disruptions of language (Rourke, ables that contribute to problem solving difficulty
1993) may render mathematical problem in algebra word problems. Paper presented at the
solving a more difficult outcome to effect annual meeting of the American Educational Re-
for students with comorbidity. Future re- search Association, New Orleans. (ERIC Docu-
search should identify effective practices for ment Reproduction Service No. ED 076 433)
effecting this outcome for students with se- Cooper, G., & Sweller, J. (1987). Effects of schema
rious problems in reading and math. acquisition and rule automation on mathematical
problem-solving transfer. Journal of Educational
Acknowledgments Psychology, 79, 347–362.
CTB/McGraw-Hill. (1997). TerraNova technical
The research described in this chapter was support- manual. Monterey, CA: Author.
ed by Grants No. H324V980001 and No. Durnin, J. H., Perrone, A. E., & MacKay, L. (1997).
H327A000035 from the U.S. Department of Educa- Teaching problem solving in elementary school
tion, Office of Special Education Programs, and mathematics. Journal of Structural Learning and
Core Grant NHD15052 from the National Institute Intelligent Systems, 13, 53–69.
of Child Health and Human Development to Van- Elbaum, B., Vaughn, S., Hughes, M. T., & Moody,
derbilt University. Statemens do not reflect the posi- S. W. (2000). How effective are one-to-one tutor-
tion or policy of these agencies, and no official en- ing programs in reading for elementary students
dorsement by them should be inferred. at risk for reading failure? A meta-analysis of the
intervention research. Journal of Educational
References Psychology, 92, 605–619.
Englert, C. S. (1984). Effective direct instruction
Asch, S. E. (1969). A reformulation of the problem practices in special education settings. Remedial
of associations. American Psychologist, 24, and Special Education, 5, 38–47.
92–102. Fletcher, J. M. (1985). Memory for verbal and non-
verbal stimuli in learning disabilities subgroups:
Badian, N. A. (1983). Dyscalculia and nonverbal Analysis by selective reminding. Journal of Ex-
disorders of learning. In H. R. Myklebust (Ed.), perimental Child Psychology, 40, 244–259.
Progress (pp. 235–264). New York: Grune & Firestone, W. A., Mayrowetz, D., & Fairman, J.
Stratton. (1998). Performance-based assessment and in-
structional change: The effects of testing in
Bransford, J. D., & Schwartz, D. L. (1999). Re- Maine and Maryland. Educational Evaluation
thinking transfer: A simple proposal with multi- and Policy Analysis, 20, 95–113.
ple implications. In A. Iran-Nejad & P. D. Pear- Fuchs, L. S., & Fuchs, D. (2002). Mathematical
son (Eds.), Review of research in education (pp. problem-solving profiles of students with mathe-
61–100). Washington, DC: American Education- matics disabilities with and without comorbid
al Research Association. reading disabilities. Journal of Learning Dis-
abilites, 35, 563–573.
Brown, A. L., Campione, J. C., Webber, L. S., & Fuchs, L. S., Fuchs, D., Eaton, S., Hamlett, C. L., &
McGilly, K. (1992). Interactive learning environ- Karns, K. (2000). Supplementing teacher judg-
ments: A new look at assessment and instruction. ments of mathematics test accommodations with
In B. R. Gifford & M. C. O’Connor (Eds.), objective data sources. School Psychology Re-
Changing assessments: Alternative view of apti- view, 20, 65–85.
Fuchs, L. S., Fuchs, D., Hamlett, C. L., & Appleton,
A. K. (2002). Explicitly teaching for transfer: Ef-
fects on the mathematical problem-solving per-
formance of students with nathematics disabili-

Enhancing Mathematical Problem Solving 321

ties. Learning Disabilities Research and Practice, ple-choice tests for sixth-grade students. Journal
17, 90–106. of Educational Research, 91, 113–125.
Fuchs, L. S., Fuchs, D., Hamlett, C. L., Phillips, Jerman, M. E., & Mirman, S. (1974). Linguistic
N. B., Karns, K., & Dutka, S. (1997). Enhancing and compuational variables in problem solving in
students’ helping behavior during peer-mediated elementary mathematics. Educational Studies in
instruction with conceptual mathematical ex- Mathematics, 5, 317–362.
planations. Elementary School Journal, 97, Jordan, N. C., & Hanich, L. B. (2000). Mathemati-
223–250. cal thinking in second-grade children with differ-
Fuchs, L. S., Fuchs, D., Karns, K., Hamlett, C. L., ent forms of LD. Journal of Learning Disabilites,
Dutka, S., & Karzaroff, M. (2000). The impor- 33, 567–578.
tance of providing background information on Jordan, N. C., & Montani, T. O. (1997). Cognitive
the structure and scoring of performance assess- arithmetic and problem solving: A comparison of
ments. Applied Measurement in Education, 11, children with specific and general mathematics
1–34. difficulties. Journal of Learning Disabilities, 30,
Fuchs, L. S., Fuchs, D., Karns, K., Hamlett, C. L., 624–634.
& Karzaroff, M. (1999). Mathematics perfor- Kansas State Board of Education. (1991). Kansas
mance assessment in the classroom: Effects on Quality Performance Accreditation. Topeka, KS:
teacher planning and student problem solving. Author.
American Educational Research Journal, 36, Koretz, D., Barron, S., Mitchell, K., & Stecher, B.
609–646. (1996). Perceived effects of the Kentucky Instruc-
Fuchs, L. S., Fuchs, D., Prentice, K., Burch, M., tional Results Information System (KIRIS). Na-
Hamlett, C. L., Owen, R., Hosp, M., & Jancek, tional Center for Research on Evaluation, Stan-
D. (in press-a). Explicitly teacher for transfer: Ef- dards, and Student Testing (CRESST). Santa
fects on third-grade students’ mathematical prob- Monica, CA: Rand.
lem solving. Journal of Educational Psychology. Koretz, D., Mitchell, K., Barron, S., & Keith, S.
Fuchs, L. S., Fuchs, D., Prentice, K., Burch, M., (1996). Final Report: Perceived effects of the
Owen, R., Hamlett, C. L., & Schroeter, K. (in Maryland School Performance Assessment Pro-
press-b). Enhancing third-grade students’ mathe- gram. National Center for Research on Evalua-
matical problem solving with self-regulated tion, Standards, and Student Testing (CRESST).
learning strategies. Journal of Educational Psy- Los Angeles: University of California.
chology Kosc, L. (1974). Developmental dyscalculalia. Jour-
Geary, D. C. (1993). Mathematical disabilities: nal of Learning Disabilities, 7, 164–177.
Cognitive, neuropsychological, and genetic com- Mawrer, R., & Sweller, J. (1985). What do students
ponents. Psychological Bulletin, 114, 345–362. learn while solving mathematics problems? Jour-
Gersten, R., Woodward, J., & Darch, C. (1986). nal of Educational Psychology, 77, 272–284.
Direct instruction: A research-based approach to Mayer, D. P. (1998). Do new teaching standards un-
curriculum design and teaching. Exceptional dermine performance on an old test? Educational
Children, 53, 17–31. Evaluation and Policy Analysis, 15, 1–16.
Gick, M. L., & Holyoak, K. J. (1983). Analogical Mayer, R. E., Quilici, J. L., & Moreno, R. (1999).
problem solving. Cognitive Psychologist, 12, What is learned in an after-school computer
306–355. club? Journal of Educational Computing Re-
Gross-Tsur, V., Manor, O., & Shalev, R. S. (1996). search , 20, 223–235.
Developmental dyscalculalia: Prevalence and de- McLean, J. F., & Hitch, G. J. (1999). Working
mographic features. Developmental Medicine memory impairments in children with specific
and Child Neurology, 38, 25–33. arithmetic learning difficulties. Journal of Experi-
Hambleton, R. K., Jaeger, R. M., Koretz, D., Linn, mental Child Psychology, 74, 240–260.
R. L., Millman, J., & Phillips, S. (1995). Review Rasanen, P., & Ahonen, T. (1995). Arithmetic dis-
of the measurement quality of the Kentucky In- abilities with and without reading difficulties: A
structional Results Information System, comparison of arithmetic errors. Developmental
1991–1994. Frankfort: Office of Education Ac- Neuropsychology, 11, 275–295.
countability, Kentucky General Assembly. Resnick, L. B., & Resnick, D. P. (1992). Assessing
Hanich, L. B., Jordan, N. C., Kaplan, D., & Dick, J. the thinking curriculum: New tools for educa-
(2001). Performance across different areas of tional reform. In B. R Gifford & M. C. O’Con-
mathematical cognition in children with learning nor (Eds.), Changing assessments: Alternative
disabilities. Journal of Educational Psychology, views of aptitude, achievement, and instruction
93, 615–626. (pp. 37–75). Boston: Kluwer Academic.
Haynes, M., & Jenkins, J. R. (1986). Reading in- Rivera-Batiz, F. L. (1992). Quantitative literacy and
struction in special education resource rooms. the likelihood of employment among young
American Educational Research Journal, 23, adults in the United States. Journal of Human
161–190. Resources, 27, 313–328.
Helwig,R., Rosek-Toedesco, M. A., Tindal, G., Ross, B. H. (1989). Distinguishing types of superfi-
Heath, B., & Almond, P. J. (1999). Reading as an cial similarities: Different effects on the access
access to mathematics problem solving on multi- and use of earlier problems. Journal of Experi-

322 EFFECTIVE INSTRUCTION

mental Psychology: Learning, Memory, and Cog- metic disabilities. Developmental Psychology, 20,
nition, 15, 456–468. 200–207.
Rothman, R. (1995). Measuring up: Standards, as- Siegel, L. S., & Ryan, E. B. (1989). The develop-
sessments, and school reform. San Francisco: ment of working memory in normally achieving
Jossey-Bass. and subtypes of learning disabled children. Child
Rourke, B. P. (1993). Arithmetic disabilities specific Development, 60, 973–980.
and otherwise: A neuropsychological perspec- Stigler, J. W., & Hiebert, J. (1997). Understanding
tive. Journal of Learning Disabilities, 26, and improving classroom mathematics instruc-
214–226. tion. Phi Delta Kappan, 79(1), 14–21.
Rourke, B. P., & Finlayson, M. A. I. (1978). Neu- Sweller, J., & Cooper, G. A. (1985). The use of
ropsychological significance of variations in pat- worked examples as a substitute for problem
terns of academic performance: Verbal and visu- solving in algebra. Cognition and Instruction, 2,
al-spatial abilities. Journal of Abnormal Child 59–89.
Psychology, 6, 121–133. White, O. R. (1984). Descriptive analysis of extant
Salomon, G., & Perkins, D. N. (1989). Rocky roads research literature concerning skill generalization
to transfer: Rethinking mechanisms of a neglect- and the handicapped. In M. Boer (Ed.), Investi-
ed phenomenon. Educational Psychologist, 24, gating the problem of skill generalization: Litera-
113–142. ture review (pp. 1–19). Seattle: University of
Schwartz, D. L., & Bransford, J. D. (1998). A time Washington, Washington Research Organization.
for telling. Cognition and Instruction, 16, Woodward, J., & Baxter, J. (1997). The effects of
475–522. an innovative approach to mathematics on acad-
Siegel, L. S., & Linder, B. (1984). Short-term memo- emically low-achieving students in inclusive set-
ry process in children with reading and arith- tings. Exceptional Children, 63, 373–388.

20

Students with Learning Disabilities
and the Process of Writing:

A Meta-Analysis of SRSD Studies

Steve Graham
Karen R. Harris

A few years ago, we overheard two children Planning, in turn, is composed of three in-
swapping jokes. Most of their jokes were gredients—setting goals, generating ideas,
pretty corny and not especially funny (at and organizing ideas into a writing plan—
least to an adult), but one caught our atten- whereas reviewing includes reading and
tion because it was about writing. editing text. Skilled writers deftly orches-
trate and monitor the use of these processes
TEACHER: Why is it, Johnny, that everyone else in order to accomplish their goals and com-
had a five-page report and your paper is only plete writing tasks.
one page long?
In contrast, most students with learning
JOHNNY: I was writing about condensed milk. disabilities (LD) use a “condensed” or sim-
plified version of this model. They mostly
The joke yielded no more than a tiny grin rely on the generating ideas component of
from the listening child. For the two eaves- the Hayes and Flower (1980) model, which
dropping adults, however, the joke provided involves converting writing tasks into tasks
a useful metaphor for describing how most of simply telling what one knows about a
students with learning disabilities write. Not topic, or writing as remembering (Scar-
only is what they write condensed, but so is damalia & Bereiter, 1986). With this ap-
how they write. proach, little attention is directed at rhetori-
cal goals, whole-text organization, the needs
Recent models of skilled writing (e.g., of the reader, or the constraints imposed by
Hayes, 1996; Hayes & Flower, 1980; Zim- the topic (Graham & Harris, 1994).
merman & Risemberg, 1997) emphasize
that composing is a complex cognitive activ- An important goal in writing instruction
ity, involving the activation of a variety of for students with LD, therefore, is to help
processes (Graham & Harris, in press). them develop a more sophisticated ap-
Hayes and Flower (1980), for example, indi- proach to composing—one that draws on
cate that writing involves three basic the same types of composing processes and
processes: planning what to say and how to strategies employed by more skilled writers.
say it, translating plans into written text, In this chapter, we examine how we have
and reviewing to improve existing text. tackled this problem head on, by explicitly

323

324 EFFECTIVE INSTRUCTION

and directly teaching planning, revising, and ic, the needs of the audience, or the organi-
other self-regulation procedures for writing zation of text. The resulting composition is
to students with LD. However, we first set generally a list of topic-related ideas rather
the stage for this examination by reviewing than a coherent discussion or examination
what is currently known about these chil- of the topic. According to McCutchen
dren’s planning and revising behavior. (1988), this approach to writing functions
much like an automated and forward-mov-
How Do Students with LD Write? ing content-generation program. Although
this approach may be useful for writing a
A One-Trick Pony: Composing as note to a friend or describing a personal ex-
Content Generation perience, it is not particularly effective for
many school-related writing tasks, such as
The idea that children create multiple drafts writing an essay, a report, or even a story. A
of their compositions, planning and revising good story, for example, includes a plot, is
them, is so common today that some of the organized in a logical manner, and is sensi-
terminology surrounding these processes tive to the needs and interests of the audi-
have worked their way into more mundane ence. This process requires more than gener-
everyday tasks. One parent disclosed (Lor- ating or retrieving ideas on the fly.
anger, 2000) that his 8-year-old daughter re-
sponded, “No Dad, it’s only a rough draft,” Although content generation dominates
when he commented, “You call that a made the planning and composing process for
bed?” most students with LD, it is a relatively un-
productive approach. One of the most strik-
Though the idea of planning, writing, and ing characteristics of these students’ writing
reworking a composition may be common- is that there is so little of it. Their papers are
place for some children, students with LD inordinately short, containing little elabora-
often minimize their use of these processes. tion or detail (Graham, Harris, MacArthur,
In contrast to skilled writers who typically & Schwartz, 1991).
devote considerable effort to planning and
thinking about their compositions before One reason why these children’s writing
they write an initial draft (Graham & Har- is so impoverished is that they have difficul-
ris, 2000), students with LD generally do ty sustaining the writing effort. In one study
little or no planning before starting to write. involving fourth- and sixth-grade students
This is the case even when they are asked to with LD (Graham, 1990), the average
plan in advance. For example, when we amount of time that students spent writing
prompted fifth- and sixth-grade students an opinion essay was 6 minutes. They only
with LD to plan before writing, they aver- composed for 1 minute when they were
aged less than 1 minute of advanced plan- asked to dictate an essay. Moreover, their
ning time (MacArthur & Graham, 1987), essays typically began with a positive or
whether they were writing papers by hand, negative statement indicating either agree-
typing their compositions on a word proces- ment or disagreement with the essay topic
sor, or dictating them to an adult. (e.g., Should boys and girls play sports to-
gether?), followed by one or two briefly
The approach to writing that students stated reasons, abruptly ending without a
with LD most often employ involves little in resolution or a concluding statement. Other
the way of planning, monitoring, evaluat- researchers (Thomas et al., 1987) have not-
ing, and so forth (Graham, 1990; Thomas, ed that students with LD experience prob-
Englert, & Gregg, 1987). Instead, it relies lems sustaining their thinking about writing
heavily on a single composing process: the topics, as they have difficulty producing
generation of writing content (Scardamalia multiple statements about familiar subjects.
& Bereiter, 1986). With this approach, stu-
dents compose by creating or drawing from A second reason why students with LD
memory a relevant idea, writing it down, produce so little content is that they fail to
and using each preceding phrase or sentence gain access to the knowledge they possess.
to stimulate the next idea. Little effort is We found that the output of fourth- and
made to evaluate or rework these ideas or to sixth-grade students with LD could be dou-
consider the constraints imposed by the top- bled or even tripled by repeatedly prompt-
ing them to write more about an assigned

A Meta-Analysis of SRSD Studies 325

topic (Graham, 1990). This was true for Students with LD employ a “thesaurus” ap-
both writing by hand and composing via proach, focusing most of their efforts on
dictation. making word substitutions, correcting me-
chanical errors, and producing a neater
A third reason for their meager output is product (MacArthur & Graham, 1987:
that difficulties with the mechanics of writ- MacArthur, Graham, & Schwartz, 1991).
ing interfere the process of generating con- In contrast to skilled writers who modify
tent. The writing of students with LD is re- their text extensively, less than 20% of the
plete with mechanical miscues, including revisions made by students with LD appre-
malformed letters, misspelled words, and ciably change what they write. The majority
errors in punctuation and capitalization of their revisions involve minor changes in
(Graham et al., 1991). Having to attend to the surface-level features of text
mechanical concerns, such as how to spell a (MacArthur & Graham, 1987: MacArthur
word or write a letter, can presumably inter- et al., 1991). More than 70% of these
fere with the execution of other writing changes are attempts to correct capitaliza-
processes (Graham, Harris, & Fink, 2000, tion, punctuation, spelling, format, and oth-
in press). For example, having to figure out er mechanical errors. The changes made are
how to spell a word while composing may generally ineffective, as the only thing that
lead a child to forget writing plans or ideas typically improves across drafts is the legi-
being held in working memory. Likewise, bility of handwriting (MacArthur et al.,
students may lose ideas or plans because 1991).
their handwriting is not fast enough to keep
up with their thoughts. SELF-REGULATION

The theoretical effect of mechanical diffi- One reason students with LD revise in such
culties on writing output are supported by a limited manner is that they have difficulty
three sources of information: managing the processes involved in revising
(Graham & Harris, 2000). In two separate
1. Handwriting fluency and spelling ac- experiments (De La Paz, Swanson, & Gra-
count for 66% and 41% of the variabili- ham, 1998; Graham, 1997), we found that
ty in writing output of primary and inter- students’ revising improved when they re-
mediate grade students, respectively ceived procedural support designed to en-
(Graham, Berninger, Abbott, Abbott, & sure that the separate elements of the revis-
Whitaker, 1997). ing process were coordinated and occurred
in a regular way. In the first study (Graham,
2. Removal of mechanical demands through 1997), fifth- and sixth-grade students with
dictation usually results in a correspond- LD executed a revising routine where they
ing increase in written output (De La Paz evaluated each sentence written, explained
& Graham, 1995). For instance, the their evaluation, selected a tactic for revis-
length of stories produced by fifth- and ing the sentence (if one was needed), and ex-
sixth-grade students with LD tripled ecuted the revision.
when they were asked to dictate rather
than write or type their compositions Revising one sentence at a time, the child
(MacArthur & Graham, 1987). first selected one of seven possible evalua-
tions (each was written on a separate index
3. Providing extra handwriting instruction card). Three of the evaluations were written
for poor writers has a positive impact on from the writer’s perspective (i.e., “this
their writing output (Berninger et al., doesn’t sound right,” “this is not what I
1997, 1998). wanted to say,” and “this is not useful to
my paper”) and three from the reader’s per-
I Don’t Do Substance, But I Do Correct Errors spective (i.e., “people may not understand
this part,” “people won’t be interested in
The following revision exemplifies the gen- this part,” and “people won’t buy this
eral approach of students with LD. part”). The seventh evaluative statement
was “This is good.” Once they selected an
Please Correct: “The bull and the cow is in the evaluation, they then explained how it ap-
field.”

Subsequent Revision: “The cow and the bull is
in the field.”

326 EFFECTIVE INSTRUCTION

plied. Next, they selected a tactic for revis- 60% of the revisions made by students in
ing the sentence, drawing on five possible the first study (Graham, 1997) involved the
directives (again each was written on a sep- mechanics of writing or minor word
arate index card). The directives were changes.
“Leave it the same,” “say more,” “leave
this part out,” “change the wording,” and It is possible that inexperienced writers
“cross out and say it a different way.” After may not be able to turn off or turn down
selecting a directive, they executed the in- their attention to issues of form easily or at
tended change. This routine reduced the ex- will (Scardamalia & Bereiter, 1986). This
ecutive burden involved in revising by sig- proposition was tested in the second study
naling movement from one element of (De La Paz et al., 1998) by telling students
revising to the next and limiting the number not to worry about errors that involved
of evaluative and tactical decisions made by spelling, punctuation, and capitalization, as
students. these would be corrected by the examiner.
When given such instructions, students were
The participating students indicated that able to adjust how much attention they paid
the procedure made the process of revising to mechanics; just 19% of their changes in-
easier by helping them carry out part or all volved usage and form.
of the revising process. There was also an
improvement in the quality of their revi- The most obvious reason why these stu-
sions and the type of text that they changed. dents focus so much of their attention on
In comparison to their typical approach to form, unless otherwise directed, is that their
revising, the procedure resulted in more re- papers contain an inordinate number of me-
visions involving larger units of text, such as chanical errors. For instance, MacArthur
phrases or t-units (i.e., independent plus any and Graham (1987) reported that fifth- and
accompanying dependent clauses). sixth-grade students with LD misspelled
one-eighth of the words in their papers and
Even more impressive results were ob- one-third of their sentences lacked initial
tained in the second study (De La Paz et al., capitalization or final punctuation. But this
1998), where eighth-grade students with LD is not the only reason. Many of these stu-
used a somewhat similar routine that in- dents view revising as proofreading. They
volved making two passes through their are more likely than their normally achiev-
composition. The first pass focused on more ing counterparts to emphasize mechanical
global concerns (e.g., “two few ideas” or issues when asked to talk about writing and
“part of the essay is not in the right order”), revising (Graham, Schwartz, & MacArthur,
whereas the second pass concentrated on lo- 1993; Wong, Wong, & Blenkinsop, 1989).
cal concerns primarily situated at the sen- For example, when asked how they would
tence-level (e.g., “this one doesn’t sound revise a paper to make it better, 61% of
right” or “this is an incomplete idea”). their responses focused on the mechanical
Again students indicated that the executive attributes of text, such as “make it neater”
routine made revising easier. It also prompt- or “spell words correctly” (Graham et al.,
ed more, better, and larger revisions. The 1993).
changes that students made were substan-
tive enough to improve the overall quality A second limitation in the revising of stu-
of their text. dents with LD is that they are often indiffer-
ent to reader-based concerns. In the revis-
OTHER PROBLEMS ing/self-regulation study by Graham (1997)
reviewed previously, three of the evaluation
Although the studies by Graham and his cards were written from the writer’s per-
colleagues (De La Paz et al., 1998; Graham, spective (e.g., “this is not what I wanted to
1997) showed that students with LD experi- say”) and three from the reader’s perspec-
ence difficulty coordinating and managing tive (e.g., “people may not understand this
the elements of revision, other factors ham- part”). The participating students appeared
per their revising as well. First, the evalua- to have difficulty taking the perspective of
tion criteria that students used in both stud- an absent reader, as only 6% of their evalu-
ies generally focused children’s attention on ations focused on the possible reaction of an
substantive concerns. Even with this focus, audience to their text. Furthermore, when
they did select a reader-based evaluation

A Meta-Analysis of SRSD Studies 327

card, only 25% of the resulting revisions able territory. When asked to tell a friend
were rated as improving text. In contrast, what kinds of things are included in a story,
60% of revisions cued by a writer-based he indicated: “I would tell him main charac-
evaluation resulted in an improvement. ter, a subject, predicate, and main idea.”
These students’ incomplete knowledge is
A third limitation involves students com- further noticeable in their stories where they
petence with the individual elements under- often omit basic elements such as the loca-
lying the revising process. In both of the re- tion, problem, ending, or moral (Graham &
vising/self-regulation studies (De La Paz et Harris, 1989a; Sawyer, Graham, & Harris,
al., 1998; Graham, 1997), there were many 1992).
instances in which students failed to make a
constructive evaluation. For example, one Gaps in the writing knowledge of stu-
child told us that he changed “he loved dents with LD are not just limited to lore
farmwork” to “he liked farmwork” (a neu- about genres and other aspects of the writ-
tral change) because the reader probably ten product but also extend to knowledge
would not understand why the main charac- about how to write. When compared with
ter “loves farmwork.” Another child indi- their regularly achieving peers, for instance,
cated that he deleted three sentences that students with LD are less knowledgeable
were central to understanding his story (a about the processes involved in organizing
negative change), because “people won’t and categorizing writing ideas as well as
care about that part.” evaluating and revising text (Englert,
Raphael, Anderson, Gregg, & Anthony,
There were also a number of instances in 1989).
which revisions were ineffective because of
difficulties in executing the intended Unshakable Confidence: At Least on
change. For instance, one child correctly the Surface
identified that the sentence “But sometimes
you need a girl’s more than you do a boy’s” Students with LD often remind us of a child
was not a complete idea and decided that who told Art Linkletter, the host of House
she needed to add information. Instead of Party, “I am the best they is in English.”
adding the word advice after the word girl When interviewed by others, these children
(the preceding sentence made it clear that are generally overconfident about their ca-
this was what she meant), she added an ad- pabilities. When we assessed the self-effica-
ditional sentence: “And sometimes you cy of 10- to 14-year-old students with LD,
don’t have advice to give.” This did not cor- for example, they were just as positive
rect the problem but further clouded what about their writing capabilities as their
she was trying to say, as it did not support peers who were better writers (Graham et
the claim that she was trying to establish, al., 1993). Both groups of students favor-
namely, that boys and girls should be edu- ably rated their ability to write reports, sto-
cated together because they can give each ries, and book reports. They were also posi-
other advice. tive about their abilities to get and organize
ideas for writing, transcribe ideas into sen-
“A Theme Is a Thing” and Other Tales of tences, sustain their writing effort, and cor-
Incomplete Knowledge rect mistakes in their paper. Although a pos-
itive judgment about one’s capabilities may
When asked to define a theme, one young promote persistence in spite of a history of
man indicated that it is “a thing that runs poor performance (Sawyer et al., 1992),
down the side of your trousers” (Abbing- there is a downside as well. Children who
ton, 1952). For many students with LD, overestimate their capabilities may fail to al-
their knowledge about writing—its genres, locate needed resources and effort, believing
devices, and conventions—is quite limited. that this is unnecessary.
Even with a relatively familiar genre such as
story writing, they are often unfamiliar with Surprisingly little is known about these
the basic attributes or parts of a story. In a children’s attitudes toward writing. When
recent interview with a third-grade child the students in the aforementioned study
with LD, for example, he started off on the were asked to indicate their agreement or
right track but quickly veered into question- disagreement with six attitude questions