Proofreading for Students With Learning Disabilities

L:arning Disabilities Research & practice. j2(l), 16-2gCopyright O 1997. Lawrence Erlbaum Associates, Inc.

Students with learning disabilities (LD) produce more spell-ing errors in their written work than do their nondisabledpeers, and they detect and correct fewer ofthese errors (Bland-ford, 1990/1991; Deshler, Ferrell, & Kass, 1978;McNaughton. Hughes, & Clark, in press). These incorrectspellings restrict the ability of writers with LD ro communi-cate effectively in writ ing (Arms, 1 984; Hinds, 1985); writtenoutput that is deticient in spelling and other writing conven-tions is difficult to read and has a strong negative influenceon overall judgments of quality (Grobe, l98l ; Moseley, 1989;Tindal & Hasbrouck, 199 I ). For many students with LD thesedifficulties persist over time despite extensive remedial inter-vention. The production of spelling errors is the most fre-quently reported problem of adults with LD (Dunham, 1987),and it otten precludes full participation in academic andvocational sertings (Hoffman et al., 1987).

Proofreading provides a final opportunity for the author todetect and correct spelling errors in a written text. In currentapproaches to writing instruction, etTective proofreadingskills take on special importance. Students are encouraged toignore misspellings during the initial composing stages of rhewriting process and to make spelling corrections during thetinal editing stage (Hine, Goldman, & Cosdan, 1990;Reynolds, Hil l, Swassing, & Ward, 1988).

There is evidence, however, that effective proofreading ischallenging for students with LD. Although students with LDfrequently focus their editing and revising efforts on thedetection and correction of mechanical errors (rather thansubstantive changes in meaning), their correction attemptshave little impact on the number of errors in their written workiGraham & MacArthur, 1988; MacArthur, Graham, &Schwartz, I 99 I ). Deshler et al. ( 1 978) found that grade schoolstudents with LD detect significantly fewer spelling mistakes

Requests for reprints should be sent to David McNaughton, 227 CEDARBuilding. Depanment of Educational and School Psychology and SpecialEduc;rtion, The Pennsylvania State University, University Park, PA 16802.E-nrail : dbm2 @email.psu.edu

Proofreading for Students With Learning Disabilities:Integrating Computer and Strategy Use

David McNaughton, Charles Hughes, and Nicole OfieshDepartment of Educational and School Psychology and Special Education

The Pennsylvania State University

We investigated the impact of integrated proofreading strategy training, combining rhe use ofa computer-based spelling checker and student strategies, on the proofreading performance ofstudents with learning disabilities (LD). Three high school srudents with LD were raughr roappiy a 5-step proofreading strategy with controlled materials and in a variety of generalizationactivities. A multiple-probe, across-subjects design was used with 3 phases: baseline, inrerven-tion, and maintenance. Results provided evidence of increases in strategy use and percentageof spelling errors corrected on both controlled proofreading materials and on compositionswritten by the students. Postintervention, all3 students produced written texts with final speilingerror rates that fell within the performance range of nondisabled peers.

than do nondisabled students, even when given explicit in-structions to search for errors. Recent research suggests thatthese difficulties persist into aduithood (McNaughton et al.,in press).

To date, nine studies have reported proofreading interven-tions for students with LD: Six studies focused on interven-tions that teach students proofieading strategies (e.g., care-fully checking each word in a sentence fbr spelling errors),and three studies investigated the use ofassistive technology(e.g., a computer-based spell ing checker).

LEARNING STRATEGYINTERVENTIONS

Learning strategy interventions proceed fiom the assumptionthat students with LD are deficient in the use of effective andefficient strategies fbr the completion of the task of prooireading.

Interventions Using a GeneralRevising Strategy

Strategy instruction in a broad set of revision behaviors(including evaluation criteria for written compositions. spe-cific revision strategies, and an overall strategy for regulatingthe revision process) was examined in a series of studies byGraham and MacArthur (1988); MacArthur, Schwartz, andGraham (1991); and Stoddard and MacArthur (1993). Resultsfrom these studies indicate that the revising strategies weresuccessfully learned and were used to improve some aspectsof writing quality by students with LD in Grades 4, 5, and 6(Graham & MacArthw, 1988; MacArthur, Schwartz, & Gra-ham, l99l) as well as by students in Grades 7 and 8 (Stoddard& MacArthur, 1993). Students were also successful in usingthe strategies independently (Graham & MacArthur, 1988)and in working in pairs with peers (MacArthur, Schwartz, &Graham, 1991; Stoddard & MacArthur, 1993).

The students with LD in these studies leamed to make use ofthe strategies to make substantive revisions affecting content,organization, and style. For both age groups, and in both theindependent and peer conditions, postinstruction compositionswere judged to be higher in quality than were preinstuctioncompositions. In all three of the studies, however, long-termsuccess at independently detecting and correcting spellingerrors was improved only slightly by the instructional inter-vention. It is unclear whether the strategy instruction enabledstudents with LD and severe spelling difficulties to inde-pendently detect and correct spelling errors at levels compara-ble to that of their nondisabled peers. The impact of theremaining spelling errors on perceptions of quality also isunclear because the instructions for scoring quality in two ofthe studies (MacArthur, Schwartz, & Graham, 1991; Stoddard& MacArthur, 1993) contained specific directions to ignoremechanical errors unless thev interfered with comprehension.

Interventions Using the ErrorMonitoring Strategy

The Error Monitoring strategy is a multistep proofreadingprocedure designed to assist students in correcting a varietyof mechanical errors in their written work, including errors ofcapitalization, punctuation, spelling, and handwriting legibil-ity. Schumaker et al. (1981) and Shannon and Polloway(1993) produced evidence of improved performance follow-ing instruction, although studies reporting successful imple-mentation of the Error Monitoring strategy are difficult tointerpret with respect to proofreading fbr spelling errors. Theerror score calculated as the dependent variable in the studiesby Schumaker et al. and Shannon and Polloway collapsedspelling, punctuation, capitalization, handwriting appear-ance, and spacing errors into a single score. When a changein error scores is reported, therefore, it is not possible for thereader to determine if this change represents an improvementin handwriting, spelling, or any of the other variables. Inaddition, the Error Monitoring scoring system specificallyexcludes contextual spelling errors (i.e., correctly spelledwords used incorrectly in a given context, e.9., there for their)from the error count. Because these errors may make up asmuch as 31Vo of the spelling errors produced by students withLD (McNaughton et al., in press), the Error Monitoring errordata may underrepresent the actual spelling errors in a text.The Error Monitoring strategy also contains provisions forstudents to seek assistance from peers or from an instructorin correcting spelling errors. In reviewing the final errorscores reported, it is not possible for the reader to determinehow many of the errors were corrected by students inde-pendently and how many were corrected through the externalassistance of peers or instructors.

Interventions That GenerateAlternative Spellings

In the Blandford (1990/1991) study, students were directedto detecterrors independently and to generate three alternativespellings for each word that they believed to be misspelled.Aithough all students demonstrated increased use of the

LEARNING DISABILITIES RESEARCH 17

proofreading strategy, proofreading performance on the stu-dents' own compositions was variable. Only one of the threestudents clearly demonstrated improved performance withrespect to error detection. The percentage of detected errorscorrected improved for all three sublects. However, so fewspelling errors were actually detected and corrected that therewas no perceptible change in spelling quality for the students'written work as rated by the classroom teacher.

ASSISTIVE TECHNOLOGYINTERVENTIONS

Concerns about the poor independent error detection andconection performance of students with LD prompted re-search into the use of assistive technology to assist in theproofreading process. McNaughton et ai. (in press) investi-gated the use of a variety of assistive spelling technologies bycollege students with LD and reported that the use of aword-processor-based spelling checker assisted students inboth detecting and correcting a higher proportion of spellingerrors than did a no-technology condition. The final spellingerror rate of the technology-assisted students, however, wasstill significantly higher than that observed for nondisabledcollege students.

Dalton, Winbury, and Morocco (1990) provided trainingin the use ofa word-processor-based spelling checker to two4th-grade students with LD. Although the use of the spellingchecker assisted both students in substantially improving theproportion of errors they were able to detect and correctwithout the assistance of the technology, both students con-tinued to require external assistance from a peer or the teacherto correct approximately 407o of the "unique" errors (i.e., eachmisspelled vocabulary item was counted as a single uniqueenor, regardless of how often it occurred) detected by thespelling checker.

More recently, MacArthur, Graham, Haynes, and DeLaPaz(1996) investigated the performance of students with LD inGrades 5 through 8 with a variety of word-processor-basedspelling checkers. As in the McNaughton et al. (in press) andDalton et al. (1990) studies, the use of the spelling checkersincreased the percentage of spelling errors corrected. When acorrect suggestion for a misspelling was provided by thespelling checker, this option usually was selected by thestudents. The spelling checker failed to identify one third ofthe spelling errors because errors were other words correctlyspelled; these misspellings were neither identified nor cor-rected by students.

AII three assistive technology intervention studies indi-cated that assistive technology alone, although of benefit, wbsinsufficient to remediate the gap between the proofreadingperformance of individuals with LD and nondisabled indi-viduals.

SEARCH AND EFROR DETECTIONPERFORMANCE RESEARCH

Research from the field of human perception on the searchand enor detection performance of nondisabled individuaismay provide important guidance to the development of effec-

18 McNAUGHToNETAL.: PRoOFREADING

tive proofreading strategies. Error detection is optimized insituations with the following conditions: First, the error ortarget is well known to the subject, and there is a low level oftarget uncertainty (Craig, 1984); second, the individual makesefficient use of attentional resources and devotes sufficientattention to the task to detect known errors (Fisk, Ackerman,& Schneider, 1987); third, the subject has a good under-standing of acceptable levels of performance and is motivatedto reach these levels (Craig, 1984); fourth, the subject workswith a display in which clutter has been reduced as much aspossible; and finally, the subject makes use of a systematicscan path in the search for errors.

These factors associated with improved error detection canbe applied to proofreading instruction in the following man-ner: First, the error or target should be well known to thesubjecl in the case of proofreading, performance should beenhanced if the subject knows whether a word was spelledcorrectly or incorrectly, or if only a limited number of differ-ent types of errors are to be considered at any one time.Spelling checker technology should assist in target certaintyby assisting in the detection and correction ofnoncontextualerrors (i.e., words that are spelled incorrectly regardless ofthecontext, e.g., kat for cat).

Second, routines should be developed that reduce the at-tentional demands on the individual; the individual should betrained to make full use of available resources in error detec-tion and correction activities. In the case ofproofreading, theuse ofthe spelling checker to detect (and assist in correcting)the noncontextual errors in a text should "free up" the proof-readerto concentrate attentional resources on other proofread-ing activities (e.g., detecting contextual errors). Furthermore,training and repeated practice in specific proofreading stepsshould promote efficient allocation of attentional resources.For example, students should be instructed to read for specificspelling enors (e.g., subject-verb agreement) and not forcomprehension.

Third, the subject should have a clear understanding ofacceptable levels oferror detection and should be motivatedto reach these levels. With respect to proofreading, the subjectshould have a clear understanding of an acceptable spellingerror rate for a given text and should have a good under-standing ofhow a low spelling errorrate enhances the reader'sperception of writing quality.

Fourth, the clutter on the display should be reduced toenhance search activities. In the case of proofreading, thismay involve nanowing the search area by reading sentenceby sentence. Finally, a well-defined search path should beused by the subject to search for errors; proofreading instruc-tions should provide suggestions for the use of an orderlysearch path.

These modifications serve to facilitate the search processand to reduce attentional demands on the observer, leading tofaster and more effective error detection. It was hypothesizedthat the incorporation of these factors into proofreading in-structional activities would enhance the error detection com-ponent of proofreading performance. This study, therefore,examined the use of an integrated proofreading srategy ap-proach incorporating elements of past learning strategy andassistive technology interventions as well as activities identi-fied as optimizing search and error detection performancegenerally. The study addressed the following issues: (a) the

participants' ability to master the application of the proofread-ing strategy, (b) the effects of strategy instruction on theparticipants' correction of spelling enors with controlledproofreading activities, and (c) the effects ofsrategy instruc-tion on the participants' correction of spelling errors in gen-eralization activities.

METHOD

Participants

Three students participated in the study. The selection ofparticipants was based on six criteria: Each participant was(a) between the ages of 15 and 18, (b) identified as having alearning disability as defined by the Commonwealth of Penn-sylvania for the purposes of providing special educationservices, (c) experienced in the use of Microsoft Word wordprocessing software and able to demonstrate basic compe-tency in spelling checker use, (d) identified as having afunctional disability in spelling as indicated by a spelling enorrate more than 2 standard deviations below the spelling per-formance of their peers even when a spelling checker wasused, (e) assessed as having a spelling score that was 2standard deviations below that of their academic peers on astandardized norm-referenced test ofspelling, and (f) refenedby the learning disabilities resource room teacher for instruc-tion in proofreading skills. Additional demographic informa-tion is provided in Table 1.

TABLE 1Characteristics of the ParticiDants

Participants

Characteristics Stuart Wanda

DemographicGenderAgeGrade

AchievementSpellingi

Predictable wordsUnpredictable wordsCombined

Composition error ratebWithout spelling checkerWith spelling checker

ReadingcWriringd

Dictation subtestGrade point average

English gradeAptitudet

VerbalPerformanceFull

Malel8tz

6J

o6

/ )

l3.5Vo4.99o

96't6

o43. lcc-

7780t )

9.87o6.67o

846l

6l1.8c-

Female Female16 16l0 l0

'7278'74

15.57o| | .49o't264J6

3. le

c

_es

8l

aAIl scores reponed as standard scores as measured by performance onthe Test of Writien Spelling-Revised (Larsen & Hammill, 1986). bAs

measured by nran performance on word-processed writing composition.cAll scores reported as standard scores as rneasured by performance on theWoodcock-Johnson Tesl of Achievement (Woodcock & Johnson, 1989).dAll scores reported as standard scores as measured by performance on theWoodcock-Jotrnson Test of Achievement (Woodcock & Johnson, 1989).elncludes grades for courses with reduced academic expectations. fAil

scores reponed as standard scores as measured by the Wechslcr IntelligenceScale foiChildren-Revised (Wechsler, 1974;. e5u5,.., score nor available.

95t0296

9l9290

The participants were enrolled in a high school in a small,northeastern college town. All three students attended a learn-ing support program for at least some of their academic day.Stuart was completing coursework to prepare for attendanceat a junior college; the school programs for Tammi and Wandafocused on vocational skills. The need to improve proofread-ing and editing performance was identified in the IndividualEducation Program objectives for all three students.

Materials

Both controlled and student-generated materials were used inthe instructional activities. Controlled materials were used toallow the students to focus their full attention on the strategyto be used and to build confidence and fluency with stand-ardized materials. Student-generated materials (developedduring Generalization Activities I and 2) were used to prac-tice the strategy with the types of materials (i.e., studentessays) used in the criterion environment (Lenz, Schumaker,Deshler, & Beals, 1993) in order to promote transfer andgeneralization of the integrated proofreading strategy to awide variery of writing activities.

Controlled proofreading materials- A set of 24 con-trolled proofreading materials was created by DavidMcNaughton. These materials were used both for instruc-tional and probe activities as well as for providing correcrivefeedback to participants. Each set of proofreading materialsincluded a passage containing spelling errors representativeof those made by individuals with LD (Hughes, Clark, &McNaughton, 1993) and a list of words identified as problem-atic spelling items for high school and college studenrs(Furness & Boyd, 1958).

Each proofreading passage was approximately 200 wordsin length and was modified so that certain spelling featuressimulated typical errors made by individuals with LD (asidentified by Hughes et al., 1993). Twenty errors were inrer-spersed throughout the passage; these 20 errors were furtherdivided among three error types. There were 8 noncontextualerrors for which the spelling checker option window producedthe correct suggestion; these words are referred to as noncon-textual errors, correct suggestion provided. Four noncontex-tual errors were included that did not result in the productionof the correct response by the spelling checker; these wordsare referred to as noncontextual errors, no correct suggestionprovided. Eight contextual errors were also inserted into thepassage. The proportion of error types was based on pastresearch with students with LD (Hughes et al., 1993). Pas-sages were prepared as word processing documents for useby the students in the proofreading acrivities, and they wereread aloud to students prior to proofreading to ensure com-prehension of the material and familiarity with the text be-cause lack of familiariry with a text may lead to reducedproofreading performance (Levy & Begin, 1984). In order toassess student performance on "challenging" vocabulary, 10spelling items randomly selected from the Furness and Boyd(1958) list of common misspellings were also dictated to theparticipants (who transcribed the words using a word proc-essing package) for use in the probe proofreading activities.

LEARNINC DISABILITIES RESEARCH 19

Generalization Activity 1. During Generalization Ac-tivity 1, students wrote a composition on an assigned topic.The topics for the student compositions were developed inconsultation with the classroom teacher and were designed toappeal to the interests of this age group (e.g., "If I were incharge of this high school, I would ...") without requiring theuse of specialized topic-specific vocabulary that might havebeen unfamiliar to the students. Topics were randomly se-lected and assigned to students with the provision that eachtopic was used only once for each student.

Generalization Activity 2. During Generalizarion Ac-tivity 2, students transcribed onto a computer a passage dic-tated by the instructor. The passages were taken from writingsamples produced by same-age peers who were judged to beaverage or above average in language arts, as reflected by agrade of C or better. The dictation passage was used toaugment the information provided by the student composi-tion. Current research provides evidence that students withLD use a limited word set (Gajar, 1989). Use of a restrictedvocabulary appears to represent a conscious effort by studentswith LD to stick to words that are easier to spell (Cowen,1988). When dictating stories, for example, students with LDhave been observed to use a wider variety ofvocabulary thanthey use in their written work (MacArthur & Graham, 1987).The use of a limited vocabulary in an attempt to avoid spellingerrors is ultimately self-defeating, however, because thenumber of different words used in a composition has beenidentified as a strong predictor of holistic ratings in postsec-ondary settings (Gajar, 1989). The written work of nondis-abled peers was used for this activity to evaluate the impactof the proofreading strategy on a wide range of age-appropri-ate vocabulary items.

Experimental Design

A multiple-probe across-subjects design (Tawney & Gast,1984) was used to evaluate the effectiveness ofthe proofread-ing instruction. The study included three phases: baseline,intervention, and maintenance. Data were collected on strat-egy use, spelling error correction rates, and final error ratesduring probes in all three phases. In addition, generalizationdata on student-generated compositions (Generalization Ac-tivity 1) and student-transcribed compositions (Generaliza-tion Activity 2) were collected during baseline and mainte-nance.

Independent Variable

The independent variable was the insfiuction in the five-stepproofreading sfrategy InSPECT. The following procedureswere used in developing the strategy: (a) identification ofspelling errors common in the writing of students with LD,(b) identification of important proofreading skills specified inprevious proofreading research with nondisabled students andstudents with LD, and (c) examination of the factors thatoptimize search and error detection performance as definedin the human perception literature.

20 McNAUGHToNETAL.:PRooFREADINc

The five steps of the InSPECT strategy (after students were"In their document") were (a) Start the spelling checker, (b)Pick conect alternatives, (c) Eliminate unrecognizable words,(d) Conect additional errors, and (e) Type in your corrections.Figure 1 contains the steps and substeps for the strategy. Thefirst step, Start the spelling checker, directed the student tobegin the spelling checker routine at the beginning of thedocument. Classroom teachers reported that starting the spell-ing checker routine from a later point in the text (and therebyfailing to review at least some of the errors) is a commonstudent mistake (8. Wilson, personal communication, De-cember 22, 1993). Use of the spelling checker technologyprovided students with a means to detect noncontextual spell-ing errors and with access to a large store ofcorrect spellings,thus serving as a prosthesis to compensate for their owndifficulties in detecting and correcting spelling errors. Use ofthe spelling checker technology also served to maximize theattentional resources available by freeing students to focustheir attention on the selection of correct spelling alternativesfor noncontextual errors and the detection and correction ofcontextual errors. The detection of noncontextual errors.which account for the largest proportion of student spellingerors, was completed automatically by the spelling checker.By specifying that the spelling checker should be started atthe beginning of the text, the strategy emphasized the needfor a systematic scan path; according to error detection theory,

In your document,

1. Start the spelling checker

a) place the cursor at the start of the text

b) under the "Tools" heading, select "Speiling"

2. Pick correct alternatives

a) read adjacent text to determine target vocabulary item

b) wait for all alternabives to appear

c) review all alternatives

d) look at beginning and end of aiternatives

e) double-click on target

3. Eliminate unrecognizable words

a) correct typos

b) mite out parts of the word that are known

c) spell word phonetically

d) use + for unknown sections of the word

e) use a s5monym

4. Correct additional errors

a) print a copy oftext

b) read from end, look for "wrong words"

- look at beginning and end ofword

c) read from beginning, look for "subject-verb" agleement

- slow down and focus on each sentence

5. Type in your corrections

a) type.in your corrections

b) run spelling checker

c) print final copy

FIGURE 1 Proofreading strategy steps and substeps.

reviewing material in an organized manner is superlor to arandom start-and-stop procedure.

The second step of the sftategy, Pick conect alternatives,assisted the student in reviewing the alternatives presented bythe computer. Students often impulsive ly select the first wordsuggested by the spelling checker and fail to "page down"through all the alternatives presented (Dalton et al., 1990).Error detection theory suggests that observers typically makerandom impulsive responses in situations of low target cer-tainty (i.e., limited knowledge about target characteristics)because they question their own ability to perform abovechance levels. Studens with LD have been observed to makehasty, impuisive responses in situations of high responseuncertainty (Gerber & Hall, 1987). By providing studentswith a sffucture for reviewing siowly and carefully the alter-natives presented by the spelling checker, we hypothesizedthat improved performance would be observed.

Students with LD may also experience difficulty in select-ing the correct spelling when a number of derivational formsthat differ only in their suffix (e.g., compensate, compensa-tion, compensatory) areprovided (Adams, 1990). The optionspresented by the spelling checker are, in some respects, a"cluttered display" (Monk, 1984, p. 297) in which target andnontarget items compete for attention. It was hypothesizedthat specific instructions to scan the options using an organ-ized scan path would be ofassistance.

The third step of the strategy, Eliminate unrecognizablewords, provided the student with sFategies to deal with enorsdetected by the spelling checker for which the speilingchecker did not offer the colrect alternative. Students weredirected to first make changes to address typographical errors.If necessary, students then created closer approximations ofthe target item (so as to activate recognition by the spellingchecker) through the use of existing spelling knowledge (e.g.,writing down the parts of the word they knew, spelling theword phonetically). The literature suggesa that students withLD often fail to make efficient use of their existing spellingknowledge in approaching new spelling problems (Gerber,1984). When prompted to give their full attention to a spellingproblem (i.e., to allocate additional attentional resources tothe problem) and make full use of their existing spellingknowledge, students with LD have demonstrated the abilityto make spelling approximations significantly better thanthose initially produced (Bailet, 1990; Gerber, 1984). Stu-dents were also taught to make use of a wild card feature (inwhich an asterisk could be used to replace unknown letters)to obtain correct spellings for a word when only some of theletters were known.

The fourth step, Correct additional elrors, was importantfor addressing contextual errors that were not detected by thespelling checker. Vacc (1987) reported that the students withmild academic handicaps in his study did not routinely proof-read materials before completing writing assignments. Theuse ofaspelling checkermay exacerbate this problem becauseit is commonly believed by students that the spelling checkercan be relied upon to detect all errors (Dalton et al., 1990).However, current spelling checker technology fails to identifycontextual errors in the text.

In order to detect contextual elrors, a broad set of cues(including surrounding context) must be considered by thestudent. Research provides evidence, however, that errors are

more difficult to detect in situations in which the contextualsupport for word recognition is strong (Holbrook, 1978).High-level semantic and syntactic knowledge of a text mayreduce the need to pay careful attention to the visual featuresof the individual words (Daneman & Stainton, 1991). Themore a reader knows about a topic, the less visual informationis necessary to recognize the word (Adams, 1990).

This step ofthe strategy incorporated the benefits discussedin error detection theory of using an organized scan path(Monk, 1984) and of increasing target certainty by reducingthe variety of errors searched for in a single sweep. Str.rdentswere direited to read sentence by sentence from the end ofthe passage with the primary goal of detecting "wrong word"errors (e.g., their for there) and then to read from the begin-ning to the end in order to detect subject-verb agreementerrors (e.g., 1 w ere going). We hypothesized that the students,by reading sentence by sentence backward from the end of atext, could balance two competing needs most effectiveiy: (a)the acquisition of enough contextual information to supportthe detection of contextual errors and (b) the limitation ofcontextual information so as not to reduce the detection ofcontextual errors.

In the fourth step, students were directed to proofread thetext twice because of experimental evidence that indicated thediminishing returns obtained by additional readings are rarelyjustified by the additional time invested (Reifer, 1991). Stu-dents were directed to read from a paper copy to incorporatethe research evidence that subjects proofread more effectivelyand efficiently from a paper copy than from a computerdisplay monitor (Gould & Grischkowasky, 1984).

In the final step of the strategy, Type in your corrections,participants were directed to type in the corrections made ontheir paper copy, review the text with the spelling checker (tocatch any errors made during the transcription process), andprint a finai copy.

PROCEDURES

A multiple-probe, across-subjects experimental design withthree phases (i.e., baseline, intervention, and maintenance)was used. The students participated in all three phases; how-ever, Tammi was unable to complete all of the testing activi-ties identified for the maintenance phase because the schoolyear ended during this phase. Sessions were scheduied to beheld three times per week, with the maintenance data collected1 .2. and 4 weeks after the final intervention session. The totaltime for each participant involved approximately 4 months ofdata collection.

Baseline Phase

The baseline condition documented the student's spellingperformance prior to intervention while using the computer-based spelling checker. During the baseline condition, partici-pants completed a series of proofreading probe activities (i.e.,detecting and correcting errors in texts in which the investi-gator had inserted errors and on dictated spelling vocabulary).Students also participated in two sets of generalization activi-ties and in norm-referenced testins activities.

LEARNING DISABILITIES RESEARCH 21

All proofreading activities were carried out using aMacin-tosh computer and Microsoft Word 5.1 word processingsoftware with accompanying spelling checker.

lntervention Phase

Instruction in the integrated proofreading strategy was pro-vided by David McNaughton. This instruction was introducedto the first participant once a stable baseline on the relevantdependent variable (i.e., strategy use) was achieved. Theremaining pafticipants remained in the baseline conditionuntil the first participant's data indicated an improving, up-ward trend for integrated strategy use with the probe materi-als. Instruction in the integrated proofreading strategy wasthen introduced with the second participant. The interventionwas introduced to the remaining participant after the secondstudent demonstrated an increase in integrated proofreadingsrategy use. Intervention was continued until the studentsperformed with 807o success for integrated strategy use forthree successive probes.

A six-step insftuctional sequence was developed for train-ing use of the integrated proofreading strategy, following theguidelines for strategy instruction suggested by the Universityof Kansas Institute for Research in Learning Disabilities (Ellis&Lenz, 1987;Lenz et al., 1993). The six instructional stepsare described.

Pretest and make commitments. The investigatorand the student examined the student's baseline proofreadingperformance. The investigator and student then discussed thegoal oftraining (i.e., to detect and colrect a greater proportionof spelling errors), why this goal was important, and how areduction in the number of spelling errors would improvecommunication between the writer and the reader. Each stu-dent was then asked to make a written commitment to learninga strategy for proofreading. In addition, the investigator em-phasized his commitment to providing the necessary instruc-tion to achieve student success,

Describe strategy. The investigator described whyand how each step ofthe snategy should be used in proofread-ing written work. Information on the anticipated benefits ofusing the strategy and the costs involved (anticipated instruc-tionai time) was provided. The five-step InSPECT strategywas introduced to the students using a small chart.

Model the strategy. During the modeling step, a chartoutlining the five steps of the sEategy was placed in front ofthe student, and the investigator modeled the use of thestrategy while revising one ofthe student's baseline essays.During this modeling procedure, the investigator demon-sfated while "thinking out loud" so that the student couldwitness the cognitive processes (as reported by the investiga-tor) and the overt behaviors necessary for successful strategyuse. The investigator modeled the entire strategy three times'During the second and third model, the investigator askedquestions designed to actively involve the students (e.g.,

identifying the next strategy step to be used and describingthe activities involved in the relevant strategy step).

22 McNAUcHToN ET AL.: PROoFREADING

Verbal practice. Students memorized the strategysteps through the instructor's use ofrapid-fire verbal rehearsalactivities. During the initial stages of training, students hadaccess to the chart listing the steps in the strategy. After tworehearsals with the strategy list in view, the list was removedand students were called upon to list the steps in order bothfrom the beginning and from investigator-selected points(e.g., "What step comes after Eliminate unrecognizable spell-ings?"). In order to progrcss to the next stage of taining,students had to state each strategy step in order withoutassistance, and they had to be able to describe in their ownwords the activities required for the completion of each step.

Practice with controlled and student-producedmaterials. Participants practiced the use of the strategywith both controlled and student-produced materials; theyalso received corrective feedback on their errors in followingthe strategy. The controlled materials were developed by theinvestigator to provide the students with practice on thedetection and correction of spelling errors. The controlledmaterials included (a) proofreading passages containing con-textual and noncontextual errors representative of spellingerrors created by students with LD and (b) proofreadingspelling attempts for challenging vocabulary (Furness &Boyd, 1958) generated by the students themselves.

The writing samples generated during the generalizationactivities were reviewed by the investigator and the parrici-pant. At these times, the investigator provided positive andcorrective feedback. During the corrective feedback activi-ties, the investigator used both the completed proofreadingprobe materials and the writing samples from the generaliza-tion activities to (a) identify specific errors, (b) review strat-egy steps associated with each error, (c) model the correct useofthe step, (d) have the student practice the correct procedure,(e) have the student paraphrase a description of the correctbehaviors, (0 have the student provide suggestions for futurepractice, and (g) communicate positive expectations for futureperformance.

Provide generalization training. The investigator re-viewed student progress with the students individually andcongratulated them on their performance. The investigatorthen obtained a commitment to generalization, and the inves-tigator and participant discussed how the strategy might beused in classroom writins activities.

Maintenance Phase

Proofreading probes with the controlled materials were usedto evaluate maintenance in the use of the proofreading strat-egy and its impact on spelling performance. Probes werecollected 1, 2, and 4 weeks followine the end of intervention.

Generalization

Subjects' error correction performance on prepared passagesmay differ from their performance with texts that they wrote

themselves (Kelly & Kerst, 1989). Therefore, participantswere administered two generalization testing activities duringthe baseline and maintenance phases of this investigation.During generalization activities, participants were asked to(a) write an extended composition on an assigned topic and(b) transcribe a passage dictated by the instructor. Althoughthe investigator was present when generalization testing tookplace, no references were made by the investigator to the useof the proofreading strategy.

MEASURES

In the probe activities, the main dependent variable was thepercentage of strategy use. Strategy use was selected as themain dependent variable in order to highlight the relationbetween sftategy instruction and strategy use. Informationwas also collected on the percentage of errors corrected andthe percentage of spelling errors in the final proofread text(i.e., the final error rate). Of additional interest was the differ-ential impact of the intervention on (a) noncontextual andcontextual errors and (b) noncontextual errors for which acorrect suggestion was provided and those for which one wasnot provided. For Generalization Activities I and 2, informa-tion was collected on the percentage of spelling errors cor-rected in the proofread text and the final error rate. To evaluatethe social validity of the proofreading instruction, data werecollected on the perceived usefulness of the instruction asreported by the participating students. Each of the measuresis discussed in further detail.

Strategy Use

Data on strategy use were collected during all probe activitiesin all three phases to determine the individual strategy behav-iors used by the participants. Strategy use is reported as apercentage of the use of the five steps in the strategy. Thepercentage ofstrategy use was calculated by scoring students'performance on strategy steps and substeps (see Figure 1).Points were awarded both for strategy steps that occurred onlyonce (e.g., Start the spelling checker) and for strategy use ona previously identified set oferrors.

Percentage of Errors Corrected

Writing samples were reviewed for spelling errors, based onthe spellings obtained from Webster's Ninth New CollegiateDictionary (Mish, 1990). A spelling error was defined as aword that was spelled incorrectly (e.g., katch for catch),inappropriate in context (e.9., hatch for catch, or their forthere in the sentence, "I am going their"), or omitted (Ander-son, 1987; Gregg, Hoy, & Sabol, 1988; Leuenberger & Mor-ris, 1990). All writing samples (including the dictation portionof the probe activity) were reviewed for spelling errors bothbefore and after proofreading by the participants. The percent-age of errors that were corrected was calculated as the numberof errors before proofreading minus the number of enors afterproofreading divided by the number of errors before proof-reading multiplied by 1A0Vo.

Final Spelling Error Rate

Spelling elrors were calculated both as frequencies and per-

r"ntag"s of total words written for both the pre- and post-

proofreading versions. The percentage of spelling errors in

itre final text (i.e., the final error rate) was calculated by

dividing the number of speiling errors by the total number of

words in the probe or composition and then multiplying theresult by 1007o. Correctly spelled words that were changed

by the participants to incorrect spellings contributed to the

minor variation observed between final spelling elrorrate andpercentage of errors corrected.

RESULTS

The results of the study are organized in three sections toaddress the major questions addressed in this study: (a) theparticipants' ability to master the application of the proofread-ing strategy, (b) the effects of strategy instruction on theparticipants' correction of spelling errors with controlledproofreading activities, and (c) the effects of strategy instruc-tion on the participants' correction of spelling errors in gen-eralization activities. Information also is provided on theusefulness of the strategy from the participants' point of view'

Mastery of the Strategy

All participants increased their use of strategy componentsfollowing intervention and achieved mastery in the use of theproofreading strategy (i.e., achieved a strategy use score of80Vo or greater for three successive sessions). Figure 2 displaysthe strategy use for the 3 participants in the baseline withspelling checker, instruction, and maintenance conditions. Stu-art's percentage of strategy use increased ftom397o in baselineto 887o in the maintenance phase, a mean difference of 497o.For Tammi, strategy use incre asedftom 42Vo (baseiine) to 89%(maintenance), a mean difference of 47Vo' it should be notedthat the maintenance data for Tammi are based on a singlefollow-up session because the end oftheschool yearprecludedadditional data collection. Wanda's baseline data were themost variable of the 3 participants because she intermittentlymade appropriate use ofthe spelling checker during this phase.Because of the variability in Wanda's baseline data, it wasnecessary to collect additional data to demonstrate acceptablestability in ievel and trend before introducing the intervention.During a follow-up interview following completion of thebaseline phase, Wanda reported that there was no pattern toher use of the spelling checker, she just used it "when she feltlike it." In the baseline phase, Wanda had an average strategyscorc of 267o; this improved to 80Vo in the maintenance phase,a mean difference of 54Vo.

Each ofthe participants required four practice sessions toreach the mastery level (i.e., strategy use score of 80Vo orgreater for three sessions in a row). During maintenance, theaverage strategy use scores were at 80Vo or higher for all 3participants, although Wanda's strategy score dipped to 61Voduring the final data collection session, 36 days after her lastinstructional session.

LEARNING DISABILITIES RESEARCH

BASELINE INTERVENTION MAINTENANCE

Stua(

+ PerccntofsFllingErors

ConEd

_-_+- Pffccnl of Shrcgy Usc

FIGU RE 2 Percentage of strategy use and errors corrected during control-

led proofreading activities.

Proof reading Performance onControlled Proofreading Materials

Errors corrected. Figure 2 provides an illustration of

the participants' elror correction performance on controlledproofreading materials during probes for the three experimen-ial conditions. For all 3 participants the strategy, interventionwas associated with an increase in the percentage of errors

corrected. The increase in percentage of spelling elrors cor-

rected from baseline to maintenance ranged ftsm2l.5Vo (for

Stuart) to 46.57o (for Wanda).

Spelling error rate. Figure 3 provides an illustration of

the participants' final spelling error rate on controlled proof-

r"uding miterials during probes for the experimental condi-

tions. The difference in postproofreading spelling errors from

baseline to maintenance ranged from -2.95 (for Stuart) to -5'4(for Wanda).

Correction of errors by error type. Table 2 provides

information on the impact of the intervention on the detection

and correction of three spelling error types (viz', noncontex-

tual errors with correct suggestion provided, noncontextual

enors with no correction suggested, and contextual errors)

23

100

80

60

40

U

ql 100

€80

*4* FinalhtuE

.-.4-- kcdof Sh@8, Ue

24 McNAUCHTON ET AL.: PROOFREADINC

BASELINE INTERVENTION MAINTENANCE participant, percentage of spelling elrors corrected (i.e., thenumber of preproofreading errors minus the postproofreadingerrors, divided by the preproofreading enors x 1007o) wasused as the menic for the comparison of baseline and main-tenance phase performances. For each participant, a greaterpercentage oferrors was corrected following the intervention.For Stuart, spelling error correction scores of 65Vo and6l%ofor the two baseline compositions improved dunng the main-tenance phase to 72Vo and 85Vo. For Tammi, error colrectionscores of 33 7o and 22Vo for the baseline compositions rose to73Vo on the single maintenance probe available for this par-ticipant. The error correction scores for Wanda improvedfrom sc ores af 33Vo and 32Vo for the baseli ne composi ti ons to'7}Vo and'l3%o for the maintenance compositions.

Dictation. During dictation activities, students tran-scribed a passage dictated by the instructor. Table 3 providesinformation on the effect of the intervention on proofreadingperformance on generalization dictation activities. As in theanalysis ofthe compositions produced during the generaliza-tion activities, percentage of spelling errors corrected wasused as the metric to compare baseline and maintenanceperformance. Proofreading activities carried out after thestrategy training intervention eliminated a greater percentageoferrors than did those carried out before the strategy training.For Stuart, error reduction performance improved from 65Vo(baseline) to 83Vo (maintenance). No information is availablefor Tammi because she did not participate in maintenancephase dictation activities. Wanda's error reduction rate im-proved from a baseline score of 24Vo to a maintenance phasescore of 72Vo.

Participants' Perception ofStrategy Efficacy

All 3 participants reported that the strategy training hadimproved their proofreading performance and that the use ofthe strategy was an efficient use of their time. Self-perceptionof the ability to detect and correct spelling errors followingthe strategy training ranged from "good" (Tammi and Wanda)to "very good" (Stuart). For all 3 participants, this represenrsan improvement over preintervention ratings of correctionability (when using a spelling checker) of "adequate" (Tammiand Wanda) and "good" (Stuart). All the participants reportedthat it would be a good idea for other students to learn thestrategy, and they had no suggestions for changing the strat-egy or the way in which it was taught.

Interrater Reliability

During the baseline, intervention, and maintenance phases ofthe study, 20Vo of the participants' probes were randomlyselected for rescoring. Each of these 14 probe samples (andthe videotape of the participants' probe performance) wasscored independently by two judges for the relevant depend-ent measures (i.e., strategy use, correction ofspelling errors,and final error rate). Interrater agreement, as calculated usingCohen's Kappa (Suen & Ary, 1989), was determined for the

t7

t0

E

a

. l o. -aa

8^'**

a"'a'

. -Qt f

80

60

20

0

g

t2

L8s= ,.E&?

0

80

50

!0

l0

0

a--___-.:d-o

rraune g ,***"r" tr*,"rr l]l-o 0"" "*o, *o o"n"r.;;"-,proofreading activities.

across the experimental conditions. For all 3 participana,performance improved from baseline to maintenance for allthree types of errors.

Although noncontextual errors were corrected more fre-quently than were contextual errors in both the baseline andmaintenance phases, the largest increase in percentage oferrors corrected was observed for the contextual error cate-gory. In the baseline with spelling checker phase, noncontex-tual enors (scores of 88%o,'l8Vo, and 60Vo for Stuart, Tammi,and Wanda, respectively) were corrected more frequentlythan were contextual errors (scores of l4Vo, l}Vo, and 4Vo forStuart, Tammi, and Wanda, respectively). This pattern ofiesults was observed for all 3 participants. In the maintenancephase, although noncontextual errors again were correctedmore frequently (scores of 98Vo,92Vo, and 98Vo for Stuart,Tammi, and Wanda, respectively) than were contextual errors(scores of 637o,63Vo,and507o for Stuart, Tammi, andWanda,respectively), the correction ofcontextual errors showed sub-stantial improvement. This pattern of results also was ob-served for all 3 participants.

Proofreading Performance onGeneralization Activities

Composition The effect of strategy training on proof-reading performance in the generalization composition activi-ties is summarized in Table 3. Because of the variationobserved in error rates on the different compositions for each

TABLE 2Mean Frequency and Proporlion of Corrections for the Three Errot

LEARNING DISABILITIES RESEARCH

Types Across the Experimental Conditions

25

Particirynt and.

Condition

Noncontextual Error,Correct Suggestion

Provided (8 Errors Inserted)

Noncontextual Error,No Correct Suggestion

Provided (4 Errors Inserted)Contestual Error

(8 Errors Inserted)

StuartBaseline

Inrcrventist

Maintenance

TammiBaseline

Intervention

Maintenancea

WandaBaseline

Intervention

Maintenance

t -s

( .e l )7.7(.%)8.0

( 1.00)

7.7(.%)t .J

i / A?\

8.0( 1.00)

4.9(.61)I -O

(.e5)

(.96)

3.3, R?\

5-9

(.e8)-1- /

L;t(.43)3.0(.7s)3.0(.7s)

2.3(.s8)3.8/ os\4.0

(r.00)

t - l

( .14)A\(.56)

(.63)

.8(. t0)

(.46)5.0(.63)

U.J

(.04)2.9(.36)4.0(.50)

ly'ore. Parenthetical values are proponions of corrections.aAIl maintenance scorcs for Tammi are based on a sinele follow-uo session.

TABLE 3Spelling Error Rate and Percentage of Spelling Errors Corrected on Generalization Activities

Panicipnt

Baseline Mainlenance

Composition I Composition 2 Dictation I Composition I Comoosition 2 Dictation I

StuartPrcproofreading error ratePostproofreading error ratePercenage of errors correcred

TammiPreproot'reading error filtePostproofreading eror ratePercentage of errors corrected

WandaPreproofreadin g error raaePostproofreading error ratePercentage of errors corrected

13.84.8

6SVo

I 1.6't.7

33Vo

r2.08.0

33Vo

13.25.1

619o

t9.515. I170/^

/ . )) - t

JZ"/O

13.83.6

I ZYO

9.32.5

73Vo

l4)

7OVo

14.82.2

859o

-a

-a

_a

9.62.6

739o

t4.42.5

83%

a

20.77.2

65Vo

2s.8l l A

t89o:"_a

12.8J-O

'12Vo

17.8I J-)

244o

'End of school year precluded data collection.

participants' use of each of the five steps of the strategy:Scores for each of the five steps ranged from .79 to 1 .00, witha mean score of .92. A Cohen's Kappa of .95 was obtainedfor interrater agreement for participants' corrections ofproof-reading errors in probe activities. Interrater agreement on thenumber of final errors in a text was calculated as .93 usinsCohen's Kappa.

DISCUSSION

This study was designed to investigate the effects of proof-reading strategy training on the spelling performance of highschool students with LD. Past proofreading research withstudents with LD investigated the impact of learning strategyinterventions (B landford, I 990/1 99 1 ; Schumaker et al., 1 98 1 ;Shannon & Polloway, 1993) and assistive technology (Dalton

et al., 1990; McNaughton et al., in press). However, it isunclear from past learning strategy intervention researchwhether the intervention closed the performance gap betweenstudents with LD and nondisabled students. It is also unclearwhether the training resulted in generalized use ofthe proof-reading sEategy. In addition, strategies that incorporate theuse ofpeers and teachers to assist in correcting spelling enorsmay produce enhanced rates of error colrection; however,students preparing for higher education or vocational settingsmust be prepared to work independently (McNaughton et al.,in press). Although assistive technology has assisted studentsin detecting and correcting errors in their written work (Daltonet al., 1990), the isolated use of technology has failed to closethe gap between the spelling performance of students with LDand the performance of nondisabled students (McNaughton

et al., in press). In this study, therefore, we examined the

reading strategy training (Blandford, 1990/1991). This studydemonstrated that high school students can be taught to usean integrated approach to proofreading that enables them todetect and correct a large percentage of their spelling errorswithout seeking peer or teacher assistance.

As in past research in the areas of proofreading(McNaughton et al., in press) and mathematics (Horton,Lovitt, & White, 1992), this study demonstrates that accessto technology in and of itself will not enable students with LDto perform at levels comparable to their nondisabled peers.The final error rate for the baseline with spelling checkercondition was far removed from that observed with nondis-abled individuals. Specific training in the effective use of thetechnology, in combination with additional strategies to com-pensate for current inadequacies of the technology, was nec-essary to bring performance within the realm of normal limits.

Limitations and FutureResearch Directions

Although this study makes a contribution to the currentknowledge concerning effective instructional practices forstudents with LD, there are a number of limitations to thestudy that require consideration. First, the participants werenot randomly selected from the population of students withLD; therefore, the results should be interpreted with care untilthe limits of their generalizability can be determined.

Second, the study does not examine the efficacy of theproofreading intervention in writing in specific content areasthat may require the use of specialized vocabulary (e.g.,scientific terminology). There may be content-specific vo-cabulary that students wish to use in their writing that isunknown to the computer spelling checker; additional re-search is necessary to examine the generalized use of thisstrategy across a wide variety of writing tasks.

Third, this investigation examines the use of a strategypackage that combines a number of specific proofreadingbehaviors. This study does not include a component analysisof the steps and substeps in order to determine which contrib-uted to the final outcome. Future research should include anexamination of the individual components in this proofread-ing strategy in order to evaluate their contribution to theoutcome.

Fourth, this study investigates only one aspect of theediting and revising process: the detection and correction ofspelling errors. Students with LD may also need assistancein learning strategies to make substantive revisions in mean-ing as they edit their written work (Graham & MacArthur,1988; MacArthur, Graham, & Schwartz, 1991; Stoddard &MacArthur, 1993). Directions to combine the search forerrors while making revisions in meaning have typically notresulted in substantive improvements in spelling perform-ance (Graham & MacArthur, 1988). It may be more appro-priate to approach revision as a multistep process that in-volves a number of passes though a document for specific,different purposes (MacArthur, Schwartz, & Graham, l99l;Stoddard & MacArthur, 1993).

With respect to additional directions for future research, itshould be noted that the selection of the behaviors chosen fortraining in the integrated proofreading strategy was based on

LEARNING DISABILITIES RESEARCH 27

a review of the literature on proofreading for nondisabledindividuals, on proofreading for individuals with LD, and onsearch and error detection performance. At present, there area large number of articles that describe what are believed tobe "best practices" in proofreading, but only limited empiricalvalidation of these techniques is available. An important lineof future research is the identification of strategies, behaviors,and resources that distinguish good and poor proofreaders(Gerber & Hall, 1989).

Future research should also examine the long-term impactof improved proofreading performance on holistic measuresof writing performance. It has been hypothesized that in orderto develop higher level writing strategies, students with LDmust be freed from the struggle to master basic spelling andwriting conventions (Rapp, 1988). Future research shouldexamine the impact of improved proofreading performanceon other text and writing features.

REFERENCES

Adams, M. J. (1990). Beginning to read: Thinking and leaming about pint.Cambridge, MA: MIT Press

Anderson, K. F. (1987). Ao analysis of the spelling errors made by ttueecollege students in essay writing. Journal of Research and Developmenti n Ed.uc at io n, 20(2), 5U58.

Arms, V. A. (1984). A dyslexic can compose on a computer. EducationalTechno logy, 24(1), 3941.

Bailet, L. L. (1990). Spelling rules usage among students with leamingdisabilities and normally achieving students. J.rurnd I of lzarning Dis-ubi litie s, 2 3, 121 -128.

Blandford, B. J. (1990/1991). Effects of a spelling proofreading strategy onthe percentage of identified misspelled words and corrected words inlearning-disabled students' compositions (Doctoral dissenation, Uni-versity of Virginia, 1990). Dissertation Abstracts lntemational, 51,/J+J.

Cowen, S. E. (1988). Coping strategies of university students with leamingdisabilities. Joumal of Leaming Disabilities, 21, 16l-164.

Craig, A. ( 1984). Human engineering: The control of vigilance. In J. S. Warm(Ed.), Sustained attention in humon perJbrmance (pp- 247-291). NewYork: Wiley.

Dalton, B., Winbury, N., & Morocco, C. C, (1990). "lf you could just learnto push a button": Two fourth gade leaming disabled students learn touse a computer spelling checker. Journal of Special Education Tech-nology, 10,177-191.

Daneman, M., & Stainton, M. (1991). Phonological recoding in silent read-ing. Journal of Experimental Psychology: Izarning, Memory, andCo gnition, I 7, 6 18-632.

Deshler, D. D., Fenell, W. R., & Kass, C. E. (1978). Error monitoring ofschoolwork by learning disabled adolescents. Journal of LearningDisabilities, 1 1, 401414.

Dunham, T. (1987).lzarning Disabled College Writers ProjecL Minneapo-lis: University of Minnesota. (ERIC Document Reproduction ServiceNo. ED 286 i88)

Ellis, E. S., & Lenz, B. K. ( 1987). A component analysis ofeffective learningsrategies for LD students. Learning Disabilities Focus,2,94-107.

Fisk, A. D., Ackerman, P. L., & Schneider, W. (1987). Automatic andcontrolled processing theory and is application to human factors prob-lems. In P. A. Hancock (Ed.), llumanfactors psychoktgy(pp.159-197).New York: Elsevier Science.

Fumess, E. L., & Boyd, G. A. (1958). 231 rcal spelling demons for highschool students. English Journal, 67, 267-270.

Gajar, A. H. (1989). A computer analysis of written language variables anda comparison of compositions written by university students with andwithout learning disabilities. Journnl of Leaming Disabililies, 22,125-130.

Gamer, R. (198'l). Metacognition and reading comprehenrion. Norwood,NJ: Ablex.

28 MCNAUGHToN ET AL.: pRooFREADING

Cerber, M. ( 1984). Orthographic problem solving ability oflearning disabledand normally achieving students. Learning Disability Quarterly, 7,157-164.

Gerber, M. M., & Hall, R. J. (1987). Information processing approaches rostudying spelling defrciencies. Joumal of lzarning Disabilities, 20,)#4.

Gerber, M. M., & Hdl, R. J. (1989). Cognitive-behavioral training in spellingfor leaming handicapped students. Izaming Disability Quanerly, 12,1 59-171.

Gould, J. D., & Grischkowasky, N. (1984). Doing the same work with hardcopy and with cathode-ray tube (CRT) computer terfiinals. HumanFactors, 26,323-337.

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

Gregg, N., Hoy, C., & Sabol, R. (1988). Spelling error patrerns of normal,leaming{isabted, and underprepared college witers. Joumal of Psy-choeducational Assessment, 6, l+-23.

Grobe, C. (1981). Syntactic maturity, mechanics, and vocabulary as pre-dictors of quality ratings. Research in the Teaching of English, 15,75-85.

Healy, A. F., & Cunningham, T. F. (1992). A developmental evaluation ofthe role of word shape in word recognition. Memory and Cognition,20,141-150.

Hinds, K. (1985, January). Dyslexia. Brown Alumni Magazine, 25-31.Hine, M. S-, Goldman, S. R., & Cosdan, M. A. (1990). Error monitoring by

leaming handicapped students engaged in collaborative microcom-puter-based witing. Journal of Special Education, 23, 407-422.

Hoffman, F. J., Sheldon, K. L., Minskoff, E. H., Snutter, S. W., Steidle, E.F., Baker, D. P., Bailey, M. B., &Echols, L. D. (1987). Needs of learningdisabled adults. Journal of karning Disabilities, 20, 43-52.

Holbrook, M. B. (1978). Effect ofsubjective verbal uncertainty on p€rceprionof typographical errors in a proofreading task. Perceptual and Motorskitk. 47.243-250.

Honon, S. V., Lovitt, T. C., & White, O. R. (1992). Teaching mathematicsto adolescents classified as educable mentally retarded: Using calcula-tors to remove the computational ows. Remedial and. Special Educa-tion, I3(3),3640.

Hughes, C., Clark, K., & McNaughton, D. (1993, February). Az analysis ofthe spelling errors produced by college .students with learning disabili-teJ. Paper presented at the meeting of the l€aming Disabilities Asso-ciation, San Francisco.

Kelly, L. P., & Kerst, S. (1989, March). The interference of meaning in errordetection during e diting. Papr presented at the meeting of the AmericanEducational Research Association, San Francisco. (ERIC DocumentReproduction Service No. ED 306 589)

Larsen, S. C., & Hammill, D. D. (1986). Test of Written Spelling-Revised.Austin. TX: PRO-ED.

Lenz, B. K., Schumaker, J. B., Deshler, D. D., & Beals, V. L. (193\. Thcword identification strategy. Lawrence: University of Kansas Press.

Leuenberger, J., & Morris, M. (1990). Analysis of spelling errors by leamingdisabled and normal college students. Izaming Disabilities Focus, 5,103- l 18.

Levy, B. A., & Begin, J. (1984). Proofreading familiar text: Allocatingresources to perceptual and conceptual processes. Menory aad. Cogni-tion. 12.621-632.

MacArthur, C. A., & Craham, S. (1987) Learning disabled students'composing under three methods of text production: Handwriting,word processing, and dictation. Journal ofSpecial Education,2l(3),

MacArthur, C. A., Graham, S., Haynes, J. 8., & DeLaPaz, S. ( 1996). Spellingcheckers and students with learning disabilities: Performance compari-sons and impact on spelling. Joumnl of Special Educatkn, 30,35-57.

MacArthur, C. A., Graham, S., &Schwartz, S. (1991). Knowledgeof revisionand revising behaviors among students with Iearning disabilities. l,e4rn-ing Disability Quarterly, 1 4, 6t-73.

MacArthur, C. A., Schwartz, S., & Graham, S. (1991). Effects of a reciprocalpeer revision strategy in special education classrooms. Learning Dis-abilities Resedrch & Practice. 6.201-210.

McNaughton, D., Hughes, C., & Clark, K. (1994). Spelling insrruction forstudents with leaming disabilities: lmplications for research and prac-tice. Learning Disability Quarterly, 17, 169-185.

McNaughton, D., Hughes, C., & Clark, K. (in press). The effecr of fiveproofreading conditions on the spelling performance ofcollege studentswith leaming disabilities. Journal of lzaming Disabilities.

Mish, F. (Ed.). (1990). Webster's Ninth New Collegiare Dictk,nary. Spr.ng-Fteld, MA: Merriam-Webster.

Monk, T. H. (1984). Search. In J. S. Warm (Ed.), Sustained auention inhuman performance (pp.294-321). New York: Wiley.

Moseley, D. (1989). How lack of confidence in spelling affects children'swritten expression. Mucational Psychoktgy in Practice, 5,4246.

Rapp, L. C. (1988, March). Proofreading skills and writing proficiency:Enor detection, editing accuracy and linguistic competence. Paperpresented at the Teaching English as a Second Language Convention,Chicago. @RIC Document Reproduction Service No. ED 307 802)

Reifer, D. M. (1991). Behavior engineering proposals: 4. Is 'backwardsreading' an effective proofreading strategy? Perceptua! and Motorskilk. 73.767-:777 .

Reynolds, C. J., Hill, D. S., Swassing, R. H., & Ward, M. E. (1988). Theeffects of revision strategy instruction on the writing performance ofstudents with leaming disabilities. "/auntal of Leaming Disabilities,2l,540-545.

Schumaker, J. 8., Deshler, D. D., Nolan, S.. Clark. F. L., Alley, G. R., &Warner, M. M. (1981). Error monitoring: A learning strategy Jbrimproving academic performance of LD adolescents (Tech. Rep. No.32). Lawrence: Univenity of Kansas, Institute for Research in LearningDisabilities.

Shannon, T. R., & Polloway, E. A. (1993). Promoting error monitoring inmiddle school students with LD. Inlerventk)n in School and Clinic.28.t60-164.

Stewart, M. F., & Grobe, C. H. (1979). Syntactic maturity, mechanics inwriting, and teacher quality ratings. Re.redrch in the Teaching of Eng-lislt 13.2M-215.

Stoddard, 8., & MacArthur, C. A. (1993). A peer editor strategy: Guidinglearning-disabled students in response and revision. Research in theTeaching of English, 27,76-103-

Suen, H. K., & Ary, D. (1989). Analyzing quantitative behavioral obsewa-tiotr dera. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Tawney, J. W., & Gast, D. L. (1984). Single subject research in specialeducation. Columbus. OH: Menill.

Tindal, G., & Hasbrouck, J. (1991). Analyzing student writing to developinstructional strategies. Leaming D isabilities Researc h & P ractic e, 6,23'7-245.

Vacc, N. N. (1987). Word processor versus handwriting: A comparativestudy of writing samples produced by mildly mentally handicappedstudents. Exceptional Children, 54, 156-165.

Wechsler, D. (1974). Wechsler Intelligence Scalefor Children-Revised. SanAntonio, TX: Psychological Corporation.

Woodcock, R. W., & Johnson, M. B. (1989). Woodcock-JolnsonTests ofAchievement. Allen, TX: DLM.