Repeated Reading and Reading Fluency in Learning Disabled Children

Author(s): Carol A. Rashotte and Joseph K. Torgesen

Source: Reading Research Quarterly, Vol. 20, No. 2 (Winter, 1985), pp. 180-188
Published by: Wiley on behalf of the International Reading Association
Stable URL: http://www.jstor.org/stable/747754
Accessed: 24-05-2015 17:10 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/747754?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Wiley and International Reading Association are collaborating with JSTOR to digitize, preserve and extend access to Reading
Research Quarterly.

http://www.jstor.org

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions
Repeatedreadingand readingfluency
in learningdisabledchildren
CAROLA. RASHOTTE
JOSEPH
K. TORGESEN
Florida State University

THISSTUDYinvestigatedwhether improvedfluency and comprehensionacross differentsto-

ries in repeatedreadingdepend on the degree of word overlapamong passages and whether
repeatedreadingis more effective than an equivalentamountof nonrepetitivereading. Non-
fluent, learning disabled students read passages presentedand timed by a computerunder
three differentconditions. In Conditions 1 and 2 (repeatedreading)the same passage was
read four times before proceedingto a new story in the next session. Stories in Condition2
containedthree times as many overlappingwords as stories in Condition 1. In Condition3
(nonrepetitivereading), each of the four passages in a session was different.Results suggest
that over short periods of time, increases in readingspeed with the repeatedreadingmethod
depend on the amount of shared words among stories, and that if stories have few shared
words, repeatedreadingis not more effective for improvingspeed thanan equivalentamount
of nonrepetitivereading.

Lecturerdpeteeetfacilite de lecturechez les enfantshandicappesqui

apprennent
CETTE tTUDE a examine si la facilit6 et la comprehensionam61ior6es'a traversdiff6rentes
histoires dans une situation de lecture r6p6t6edependentdu degr6 de chevauchementdes
mots parmiles passages et si la lecture r6p6t6eest plus efficace qu'unequantit66quivalente
de lecturenon r6p6titive.Les e61veshandicapp6squi ont des difficultes de lecture, ont lu des
passages pr6sent6set minutespar un ordinateursous trois conditionsdiff6rentes.En condi-
tion I et 2 (lecturer6p6t6e)le meme passage 6taitlu quatrefois avantde continueravec une
nouvelle histoire dans la session suivante. Les histoires en condition2 contenaienttrois fois
plus de mots en chevauchementque les histoiresen condition 1. En condition3 (lecturenon
r6p6titive),chacun des quatrepassages dans une session 6tait diff6rent. Les r6sultatssug-
gerentqu'a'traversde courtesp6riodesde temps, les augmentationsen vitesse de lectureavec
la m6thodede lecturer6p6t6edependentde la quantit6de mots partag6sparmiles histoires, et
que si les histoires ont peu de mots partag6s,la lecture r6p6t6en'est pas plus efficace pour
aml1iorerla vitesse qu'unequantit6equivalentede lecture non r6p6titive.

Lecturarepetidayfacilidad de lecturade niriosminusvilidos de

aprendizaje
ESTE ESTUDIO investig6 si la mejora de facilidad y comprensi6nen la lectura repetidade
varios cuentos, dependede la cantidadde palabrasutilizadasrepetidamenteentrepasajesy si
la lecturarepetidaes maisefectiva que una cantidadequivalentede lecturano repetida.Alum-
nos minusvilidos sin facilidad,leyeronpasajespresentadosy cronometradospor una compu-
tadora bajo tres condiciones diferentes. En Condici6n 1 y 2 (lectura repetida), el mismo
pasaje se ley6 cuatroveces antes de procedera un cuento nuevo en la sesi6n siguiente. Los
cuentos en Condici6n 2 conteniantres veces maispalabrasutilizadasrepetidamenteque los
cuentosen Condici6n 1. En Condici6n3 (lecturano repetida),cada uno de los cuatropasajes
durantela sesi6n eran diferentes. Los resultadossugieren que duranteperiodos cortos de

180

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions
 tiempo,la aceleraci6nen velocidadde lecturaconel m6todode lecturarepetidadependfade
de palabras
la cantidad encomtinentrelos cuentos,y quesi los cuentostenfanpocaspalabras
en comiin,la lecturarepetidano eramaisefectivaparamejorarla velocidadqueunacantidad
equivalentede lecturano repetida.

For many educators, the attainment of proximatesspeakingrate and comprehensionis

readingfluency has meantan emphasison word good because attentionis available to process
attackskills and the developmentof word accu- meaning.Poor readers,on the other hand, must
racy. Recent researchhas suggested that in or- expend processing effort, or attention, to de-
der to improve the reading fluency and code each word. As a result, theirreadingspeed
comprehensionof many poor readers, reading is slow and comprehension is more difficult.
instructionmust also focus on ways to increase Accordingto Samuels (1979), nonfluentbut ac-
the speed of word recognition (Lesgold & curate readershave acquiredword attackskills
Resnick, 1982). but requiremore practicebefore these skills are
Several investigators (Chomsky, 1978; performedautomatically.The repeatedreading
Lauritzen,1982; Moyer, 1979; Neill, 1980; Sa- method provides the necessary practice. Flu-
muels, 1979) have reportedthat the rathersim- ency and comprehensionincreasewith each re-
ple practice of repeatedly reading entire reading of a single passage and improvements
passages can be an effective supplementary are carriedover to new selections.
technique for increasing reading fluency and While Samuels'sexplanationoffers a rea-
comprehension. Essentially, the method in- sonable frameworkfor understandingthe effec-
volves the selection of passages of 50-200 tiveness of the repeated reading method, it
wordsat a level thatenables the readerto recog- provideslittle specificity as to how repetitionof
nize most of the words. The selection is orally an entire passage facilitatesword identification.
read by the studentthree or four times, or until In particular,it is unclearhow improvedfluency
a satisfactorylevel of fluency is reached,before obtainedthrough repeatedreading of the same
proceeding to a new passage. Word accuracy passage is transferredto a new passage.
rates and reading speed are usually reportedto Moyer (1982) concluded that increased
the studentafter each reading, and daily prac- gains in readingfluency occur because repeated
tice is recommended. reading of entire passages maximizes redun-
Recent reviews of the repeated reading dancy at all levels of the writtenlanguagestruc-
method (Kann, 1983; Moyer, 1982) have been ture. Repetition of entire passages gives the
supportiveof the use of the techniquewith some poor reader the needed extra practice in using
slow readers, althoughboth note that there has higher linguistic structure(contextualand syn-
been little experimentalevidence of its instruc- tactic cues) as well as in extractinggrapho-pho-
tional effectiveness. In particular, there have nemic word structure. These improved skills
been no direct comparisons of the repeated are applied to the new reading material and
readingmethod with other methods. There has thus, reading fluency is increased across
also been little research to indicate why the new passages.
techniqueis successful and whether its success Schreiber(1980) suggestedthatthe success
dependson certainpassage characteristics. of repeatedreadingwas not so much a matterof
The repeated reading method is based extra practice of certain skills, but rather the
largely on the teaching implicationsof the the- discovery that various morphologicaland syn-
ory of automatic information processing in tactical cues must be used in order for fluency
reading (LaBerge & Samuels, 1974). Accord- to develop. Accordingto Schreiber,this discov-
ing to automaticity theory, fluent readers de- ery comes about as a result of several repeti-
code text automatically, that is, they read tions of the same passage. Once discovered,
without expending processing effort in decod- these cues are used in new passages.
ing individual words. Their reading speed ap- At the presenttime there is little empirical

Repeatedreadingand readingfluency RASHOTTE& TORGESEN 181

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions
support for the above explanations. Unfortu- plete the same total numberof readingsas those
nately, the advancementof a rationale for the in Condition 1 or 2; however, they will read
effectivenessof repeatedreadinghas been ham- each passage only once. The passages will be
pered by a lack of documentationin existing similarto those in Condition1 (i.e., word over-
studies of the reading passage characteristics. lap and relatednessof context will be minimal).
None of the studies reportthe amountof word Students' reading speed, word accuracy, and
overlap or context relatedness among the re- comprehension scores in Condition 3 will be
peated passages. Thus, it is not clear whether contrastedwith those in Conditions1 and 2.
the improved reading fluency achieved across
passages with the repeated reading method is Method
obtainedonly when the reading materialhas a
high degree of word overlap and/or contextual
relatedness. Subjects
It is also not known whether equivalent Subjects were 12 nonfluent, learning dis-
amounts of nonrepetitivereading can increase abled (LD) studentsdrawn from three elemen-
readingfluency as effectively as repeatedread- tary schools. Ages ranged from 8.6 to 12.0
ing. It may be that when the repeatedreading years with an average age of 10.5 years. Stu-
method is introduced, students actually read dents were enrolled in Grades2 through5 and
more. Less time may be devotedto other read- had a mean IQ of 100. Readinglevel was Grade
3 or higher as measuredby the readingsubtests
ing skills such as wordanalysis and comprehen-
sion, or in their enthusiasm for the method, of the Peabody Individual Achievement Test
teacherand studentsmay simply allot more in- (Dunn & Markwardt,1970).
structionaltime to repeated reading. If this is Classificationas a nonfluentreaderwas de-
the case, then equivalentamountsof nonrepeti- terminedby the following criteria:(a) identified
tive reading may also increase the read- as a slow readerby the LD teacher;(b) obtained
an averagereadingrate of 65 words per minute
ing fluency of children who are slow but
accuratereaders. (wpm) or less on Grade 2 reading passages of
The present study is an attemptto investi- 90 words. The measure used was the Sucher-
Allred ReadingPlacementInventory(Sucher&
gate two aspects of the repeated reading
method. First, by controllingthe word and con- Allred, 1973); and, (c) achieved a word accu-
text characteristicsof the reading materialit is racy score of 90% or better on the same read-
hoped that the present investigation will indi- ing passages for which reading speeds
cate the extent to which the effectiveness of re- were determined.
peated reading is dependent on passage Subjects were grouped in six pairs with
characteristics.In Condition1, studentswill re- each pair matched as closely as possible on
peatedly read passages in which the numberof reading scores. Using pretest reading speeds,
words sharedamong stories is low and related- pairswere arrangedso thateach containeda rel-
ness of context is minimal. In Condition2, the atively high and low speed reader,and all pairs
same studentswill repeatedlyread passages in had similar mean readingspeeds. The six pairs
which the word overlap among stories is high. were then randomly assigned to Groups 1
These passages will be unrelated, although through6.
there may be some context similaritydue to the
high degree of word sameness. Reading speed, Materials
word accuracy, and comprehensionscores ob- Studentswere requiredto reada total of 44
tainedunderboth conditionswill be contrasted. different reading passages. These passages
Second, this study will compare the effec- were based on materialwith a stated Grade 2
tiveness of repeated reading and nonrepetitive readabilitylevel (Reader'sDigest Skill Builders,
readingwhen equivalentamountsof readingare Level 2, Berke, 1977; and Reader'sDigest Tri-
given. In Condition 3, the students will com- ple Takes, Level 2, Goodman, 1982), then

182 READING RESEARCHQUARTERLY * Winter 1985 XX/2

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions
modified to provideequivalentpassages of 100 Experimental Design
words. As a furthertest to establishequivalency and Procedure
of readabilityat the Grade2 level, Fry's (1977) The presentstudyconsistedof threeexperi-
readabilityformulawas computedfor each pas- mental readingconditions each covering7 days
sage. The numberof syllables containedin each for a total of 21 sessions. Conditions 1 and 2
passage was limited from 117 to 122, and the represented the two repeated reading condi-
number of sentences was restricted from tions. Condition3 was the nonrepetitivereading
10 to 12. condition. The general format of all three con-
Of the 44 passages, 37 were designed with ditions was the same. Students read approxi-
minimal word overlap among the stories and mately 15 minutesa day, 5 days a week for each
unrelatedcontext. Minimalword overlapmeans 7-day period. Every 15-minutesession included
that only about20 words were common to three four passage presentationswith four compre-
or more stories in a condition, and these words hension questions given directly after the first
were limited to easy, frequent words such as reading,and one comprehensionquestiongiven
the, he, to, etc. Of these 37, 2 were designated after each remaining reading. Students were
as pretest passages, 7 comprised the stories in also provided feedback on reading speed and
Condition 1, and 28 comprised the stories in word accuracy immediatelyafter each reading
Condition 3. The remaining 7 passages were of a selection. Reading conditions differed,
designed with a high degree of shared words however,in the following ways: In Condition 1,
among the stories. Approximately 60 words students orally read one of seven unrelated,
were common to at least three or more of the nonoverlappingstories four times each day for
seven stories. The stories were unrelated;how- a total of 28 readings. In Condition2, students
ever, because of word overlap, some context again readrepetitivelyexcept thatthe seven sto-
similarity was unavoidable. These passages ries in this conditioncontaineda high degree of
comprisedCondition2. shared words. In Condition 3, students read
Passageswere randomlyassignedto Condi- four different stories each day none of which
tions 1 and 3 as well as the pretestconditions. were repeated. Thus, over the 7-day period of
Ordering of passages within the three condi- this condition, studentsread 28 different, unre-
tions was also varied. Studentswere randomly lated, and nonoverlappingstories.
assigned one of seven different orderings of Each studentparticipatedin all threecondi-
the stories. tions. However,orderingof readingconditions
All selections were presentedon an Apple differedfor each of the six groupsof students.
II Plus computer. The computer was pro- An initial practice session was given to all
grammed in Apple SuperPILOT(1982) using studentsto familiarizethem with the computer
graphicsmode. This procedureenabled the ex- format. In this session, studentsread two read-
aminer to offer the readingselection to the stu- ing passages similar in length and difficulty to
dent in one screen presentationwith all lines of the test selections and were given four oral
the story appearing on the screen simulta- comprehensionquestions per passage. Follow-
neously. Characters in the text included both ing the practice session, a pretest session was
upper-and lowercase. held in which two pretest passages were pre-
Following the first reading in each session sented. Each passage was again read once and
in all conditions, four comprehensionquestions was followed by four oral comprehensionques-
were given orally to students. One comprehen- tions. The first session of the experimental
sion question was given after each of the re- readingconditionsbegan directlyafterthe prac-
maining readings. These questions involved tice and pretest sessions. In this session, stu-
literal comprehensionrequiringrecall of infor- dents were told the format of each session and
mation from the passage and were similar to were given a rationalefor their reading activi-
those presentedin readinginventories. ties. At the end of the last session studentswere

Repeatedreadingand readingfluency RASHOTTE& TORGESEN 183

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions
repetitive reading) they would prefer to use to the above measures, with the slope of these
practicetheir reading. lines becomingthe dependentmeasure.

Dependent Measures Reliability

Oral reading fluency involves two compo- Word accuracy for oral reading was re-
nents: reading speed and word accuracy. cordedby a second observerfor at least 2 of the
While reading speed may be emphasized over 21 initial readings per subject. Reliability was
accuracy in building fluency (Samuels, 1979), calculated by dividing the total error agree-
both componentswere measuredin the present mentsby the total numberof errorevents. Aver-
study. age error agreement based on 32 reading
Reading speed for each passage was de- passages was 90%.
fined as the numberof words read per minute. Agreement on comprehensionscores was
The microcomputerwas programmedto record checked by correlating scores obtained by the
the amountof time takenby each subjectto read examinerwith those of a second observer.Each
each passage and to compute the number of comprehensionscore was based on answers to
wordsreadper minute.The studentinitiatedthe four questionsgiven by the studentafter an ini-
time recordingby pressing the space bar to be- tial reading. The correlation obtained for 24
gin reading. When the reading was completed reading passages of which 2 were initial read-
and the space bar was again activated,the time ings by each studentwas .93.
clock recordedthe elapsed time and the words
read per minute. A printoutof these times was
Results
displayed immediately on the screen for
the student.
Wordaccuracy was defined as the number Pretest
of errors per passage and was recordedby the Means, standard deviations, and ranges
experimenteras the student read the passage. based on averagepretestscores for each student
Errorsincludedmispronunciations,nonpronun- were calculated on each measure. The mean
ciations (the subject pauses, but does not pro- speed for the total group (49.7 wpm) indicates
duce a response), omissions, insertions, line thatthe groupwas readingat a slow speed at the
skipping, and substitutions. No error correc- beginning of the study, but there was a fairly
tions were made during the readingof the first wide spread of slow speeds among the group
passage in each session. However, in the three (30.6 to 62.4 wpm). Word accuracy,however,
remainingreadings of a session, the examiner was quite high for all students(mean errors =
corrected errors that distorted the meaning of 6.45%, SD = 2.5) indicatingthat the reading
the passage. The total numberof errorsper pas- material was at an appropriate level for the
sage was given to the studentimmediatelyafter groupand that slow readingwas not simply due
each reading. to an inabilityto identifywords in the passages.
Four comprehensionquestions were given Comprehensionscores were also high (out of a
by the examiner after the first reading in each possible score of 4, the total mean score was
session to assess each student'scomprehension 3.5). This result suggests'thatat least with lit-
level. Answers were scored zero if totally incor- eral recall-typecomprehensionquestions at the
rect, 1/2 if partiallycorrect, and 1 if completely Grade2 level, the group'sslow readingdid not
correct. interfere with passage understanding. It also
Althoughreadingspeed and word accuracy allowed little room for improvement on
scores were recordedfor all 28 readingsin each this measure.
condition, the key scores used to determinebe-
tween-session changes were those obtained on Within-Session Changes in Fluency
the first reading in each session. The resulting Although changes in speed, accuracy,and
seven data points were then used to yield a line comprehension across sessions under various
of best fit for each studentin each condition on reading conditions were of primary interest in
184 READINGRESEARCH * Winter 1985
QUARTERLY XX/2

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions
this study,within-sessionchanges were also an- in each condition on the above three measures.
alyzed. Difference scores between Reading 1 ANOVA procedures with repeated measures
(R1) and Reading 4 (R4) for each studentin a (group by condition) using slopes for each stu-
session were computedfor speed and errors to dent in each condition revealed no significant
assess within-sessionchanges. A 3 (Conditions) effects on any of the measures.
x 6 (Groups) analysis of variance (ANOVA) Although groups were balancedon the ba-
with repeated measures indicated significant sis of pretest scores, score differences within
speed differencesamong the three readingcon- each grouping still remained, particularly on
ditions, F(2,12) = 45.74, p < .001. In Condi- the speed measure.Correlationsbetweenthe in-
tions 1 and 2, where the four readings were tercepts and slopes for each student showed a
consistent, negative relationship between the
repetitionsof the same story, the mean increase
in speed from R1 to R4 was 35.3 wpm and 33.0 level at which the studentbegan a condition on
wpm respectively.In both cases, using Tukey's each measure and the amount of performance
HSD procedureto test contrasts, the mean in- change that occurred across the condition.
crease in Conditions 1 and 2 was significantlyMean correlationsfor the three reading condi-
greaterthanthatin Condition3 (p < .01). The tions were as follows: speed, r = .50; errors, r
small mean increase in speed from Ri to R4 = -.80; comprehension, r = -.90. Thus, the
within Condition 3 (5.2 wpm), however, was lower the student'sinterceptfor speed, errors,
also significant, F(1,6) = 14.01, p < .01. and comprehension, the greater was the in-
This increase may have been due partly to the crease in speed and comprehension,and the less
fact that errors were correctedby the examinerthe errorreduction.Because of the high correla-
in R4 but not in R1. tion between the slopes and interceptson each
ANOVAproceduresto determine changes measure, analyses of covariance (ANCOVA)
in word accuracy within sessions under differ-with repeated measures were performed with
ing readingconditions indicatedthat there was the interceptsas the covariate.The choice of the
interceptsratherthan the pretest scores for the
also a significant difference in word error re-
duction among the 3 conditions, F(2,12) = covariatewas based not only on the strongerre-
7.35, p < .01. In Conditions1 and 2 the mean lationship between the intercepts and slopes,
decrease in errorsfrom R1 to R4 was 2.52 and but also on the fact that the initial performance
2.16 respectively.Both of the mean difference of some students varied across conditions and
differedfrom thatof their pretestperformance.
scores in Conditions 1 and 2 were significantly
greaterthan the difference score of Condition3 Adjusted mean slopes for each measurein
(p < .01). The small reductionof errorswithin the three readingconditions are shown in Table
sessions in Condition 3 (0.82) was not 1. Results of the ANCOVAon word errors and
significant. comprehensionindicatedno significanteffects.
However,readingconditionsdid have a signifi-
cant effect on improvements in speed across
Between-Session Changes in Fluency and
sessions, F(2,11) = 7.32, p < .01. An exten-
Comprehension sion of Tukey'smethodof multiplecomparisons
To determine the effects of the different adapted for analysis of covariance procedures
reading conditions on speed, word accuracy, was used to contrastthe adjustedmean slopes
and comprehension over several days, slopes among the reading conditions (Bryant &
and interceptswere calculatedfor each student Paulson, 1976). Comparisons indicated that

Table1 Adjustedmean slopes for speed, errors, and comprehensionin each readingcondition
Speed Errors Comprehension
Condition 1 0.1956 -0.1814 0.0419
Condition2 1.0522 -0.2887 0.0378
Condition3 0.3733 -0.1758 0.0244

Repeated reading and reading fluency & TORGESEN

RASHOTTE 185

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions
Condition2 producedgreatermean increasesin readingas a method for increasingfluency and
readingspeeds across sessions thanCondition1 comprehension.The first concernedthe role of
(p < .05). The differencebetween Conditions passage characteristics. Would increases in
2 and 3 was not significant, although the ob- readingspeed, word accuracy,and comprehen-
tained t of 3.89 was close to the requiredt of sion across sessions depend upon the degree of
4.02 (p < .05). Mean speed increasein Condi- word overlapamong passages read in different
tion 1 did not differ from thatof Condition3. sessions? The second question asked whether
To explore the possible relationship be- repeatedreading would be more effective than
tween errors and speed in this study, correla- an equivalentamountof nonrepetitivereading.
tions were computed between students' error On the issue of passage characteristics,the
slopes and speed slopes in all conditions. Find- present findings suggest that gains in reading
ings indicated a moderately negative relation- speed were affectedby the degree of wordcom-
ship, that is, as speed increased, errors tended monalityamong the stories. When speed scores
to decrease (Condition 1, r = -.48; Condition were adjustedso that beginning readingspeeds
2, r = -.28; Condition3, r = -.47). were constant across conditions, mean speed
Inspection of each student'sdaily progress gains in Condition 2, where stories contained
both within and between sessions over all read- many shared words, were significantlygreater
ing conditions indicatedthat there was individ- than gains in Condition1 where stories had few
ual variationin the amountof speed gain during overlappingwords.
a session and between sessions. To assess The amount of word commonality among
whether there might be some relationshipbe- stories, however,had less effect on errorreduc-
tween the degree of speed gain within a daily tion or comprehension gains. There were no
session and the amount of speed increase over significant differences among the readingcon-
several sessions, correlations were computed ditions on either measure even when scores
between speed slopes and speed difference were adjusted.
scores (R4 - Ri). Correlationsin each reading The lack of significantfindingson the com-
condition rangedfrom low to moderate(Condi- prehension measure probably reflects the fact
tion 1, r = .44; Condition2, r = .17; Condi- thattherewas little room for improvement.The
tion 3, r = .22). LD studentsin this study showeda good level of
The relationshipbetween intelligence and understanding on the pretest stories despite
the amount of speed improvementover several poor reading fluency. This attainmentmay be
sessions was also explored. Results indicateda peculiarto the LD populationor may reflectthe
somewhat low, negative correlation between simple structure and content of Grade 2
ability and speed slopes suggestingthat, if any- level stories.
thing, those with lower ability tended to show The findings on the speed measureindicate
greater speed increases than those with higher that there was little carryoverof fluency gains
ability (Condition1, r = -.23; Condition2, r = achieved from reading the same story several
-.10; Condition3, r = -.37). times unless the new stories sharedmany of the
In response to the question about prefer- same words. Because error reduction across
ence for practicing reading by reading repeat- Condition 2 did not differ significantly from
edly or reading nonrepetitively, 75 % of the Condition 1, increasedspeed in Condition2 ap-
students were in favor of using the repeated pearsto be largely the resultof fasteridentifica-
method. They felt that repeatedreadingmade it tion of words sharedby the stories. This finding
easier to get a high speed score. fits with Samuels's(LaBerge& Samuels, 1974)
view that word repetitionpromotesfasterword
Discussion processing or word automaticity.However,ac-
cording to Samuels (1979), use of the repeated
readingmethodshouldalso makeword analysis
The present research addressedtwo ques- skills more automatic,resultingin increasedef-
tions in evaluatingthe effectiveness of repeated ficiency in the processing of both old and new

186 READINGRESEARCH * Winter 1985

QUARTERLY XX/2

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions
words. Findingsfrom the presentstudy indicate provedfrom an averageof 54.7 to 90.0 wpm in
that, if repeatedreadingdoes have an effect on ConditionI and 57.1 to 90.1 wpm in Condition
word analysis skills, these effects are much 2) so that this procedural discrepancy should
weakeror take longer to develop thaneffects on not have influencedresults.
speed of recognitionof whole words. In the present research, new stories in the
The resultsof comparisonsbetween the re- repeatedreading conditions were always intro-
peated readingconditionsand the nonrepetitive duced in a differentsession on a new day rather
reading condition also suggest that repeated thanin the same session or duringthe same day.
readingtechniquesdo not readilyaffect a broad It is not clear when Samuels introduced new
range of reading skills. Adjusted mean speed stories to his students. However,if the new sto-
gains in Condition 1 did not differ significantly ries were given in the same session as the old
from those in the nonrepetitivereading condi- stories, there may have been some buildup of
tion, and differencesbetween Conditions2 and speed similarto that which occurredwithin ses-
3 only approachedsignificance. Whatseems es- sions in Condition3. Because fluency gains are
tablishedby these results is that when the same only useful if they persist from day to day, it
amountof practice and the same feedback for- was felt that the presentprocedurewas best for
mat of the repeated reading method are used, assessing fluency gains.
repeatedreading is not necessarily more effec- The other procedural difference in this
tive as a tool for increasingreading speed than study was the use of computers to present the
nonrepetitivereading.Moyer (1982) has argued readingtext ratherthanthe printedpage. While
that the merit of repeatedreadingis that it pro- it is difficult to assess the impactof the different
vides practicein integratingall levels of written media, informaltesting with a few LD students
language structure, not just at the word level. who were not included in the study did not re-
The minimal gains made by many students in veal speed differencesfor passages presentedin
Condition 1 suggest that this integration may the differentmodes. Also, if anything,the com-
not be easily attained. putersseemed to improvethe techniqueby giv-
Several reasons may be given for the lim- ing immediate,accuratefeedback.
ited performanceof studentsin this study under Samuelstalkedaboutthe possible speed/er-
the repeatedreadingconditions. The most obvi- ror tradeoffthat could hamperfluency gains in
ous is that the 7-day period for each condition some children. Because accuracy rather than
was too short to obtain meaningfulgains, par- speed is usually emphasizedin readinginstruc-
ticularly in Condition 1 where stories had few tion, this could become a problem even when
shated words. This charge may be true. How- readingrepeatedly.The reverse (speed over ac-
ever, Samuels's (1979) case presentationindi- curacy)can also occur in repeatedreading.Nei-
cated substantialimprovement(38 wpm) by the ther tradeoff appearedto be a problem in the
fifth repeatedreadingsession (fifth new story), present study. Instructionsand feedback were
giving the impressionthat the repeatedreading geared to encouragespeed acquisition. A mod-
methodis effective within very few sessions. erate negative correlation (-.41) between the
A second possible reason for the present speed and error slopeg for all conditions indi-
results may be proceduraldifferences. Samuels cated that as speeds increased, errors tended
(1979) had studentsreadthe same story repeat- to decrease.
edly until they reacheda preset speed criterion The last factorthatmay have influencedthe
before being introducedto a new story. In this present findings is individual differences.
study, students read each story a fixed number While Samuels (1979) reported that the re-
of times (four). The presentprocedurewas sim- peated reading method can be used with re-
ilar to that describedby Moyer (1982) and was tarded children, children with learning
necessary to keep the amount of practice con- problems, and normalchildren, other research-
stant among conditions. The data indicate that ers (Kann, 1983; Moyer, 1982) have suggested
students were able to make substantial speed that some children respond better to the tech-
gains within the four repetitions (scores im- nique than others. The present study used chil-

Repeated reading and reading fluency & TORGESEN

RASHOTTE 187

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions
dren who were identified as learning disabled. CHOMSKY, C. (1978). When you still can't read in third
Inspection of individual graphs for these stu- grade: After decoding, what? In S. J. Samuels (Ed.),
What research has to say about reading instruction
dents indicatedthat among this groupthere was
(pp. 13-30). Newark, DE: InternationalReadingAsso-
considerable variability in speed performance ciation.
in each condition. Some children increased DUNN, L.M., & MARKWARDT, EC. (1970). PeabodyIndivid-
their readingfluency while others made little or ual Achievement Test. Circle Pines, MN: American
no improvement. GuidanceService Inc.
FRY,E.B. (1977). Fry'sreadabilitygraph:Clarification,va-
It was noted that there were individualdif- lidity and extension to level 17. Journal of Reading,
ferences in fluency gains within daily sessions 21(3), 242-255.
that may have influenced performancediffer- GOODMAN,B. (1982). Reader's digest triple takes, Level
ences across sessions among students. Correla- 2. Pleasantville, NY: Reader's Digest Educational
Division.
tions betweenthese speed differencescores (R4
KANN, R. (1983). The method of repeated readings: Ex-
- R 1) and speed slopes indicatedthatthe degree panding the neurological impress method for use with
of speed gain within a daily session was posi- disabled readers. Journal of LearningDisabilities, 16,
tively related to the amount of speed increase 90-92.
over severalsessions. However,the relationship LABERGE, D., & SAMUELS, S.J. (1974). Towarda theory of
automaticinformationprocessing in reading. Cognitive
was not strongenough to makethis factora reli- Psychology, 6, 293-323.
able predictorof readerswho might profitfrom LAURITZEN, C. (1982). A modificationof repeatedreadings
the repeated reading method. Level of intelli- for group instruction. The Reading Teacher, 35 (4),
456-458.
gence also was not a reliable predictorof flu-
LESGOLD, A.M., & RESNICK, L.B. (1982) How readingdiffi-
ency gains with repeatedreading. culties develop: Perspectivesfrom a longitudinalstudy.
It shouldbe notedthatstudentsliked the re- In. J.P. Das, R.FEMulcahy,& A.E. Wall(Eds.), Theory
peatedreadingmethodand the generalfeedback and research in learning disabilities (pp. 155-187).
format regardlessof the degree of improvement New York:PlenumPress.
Even at the end of the study after a month of MOYER, S.B. (1979). Rehabilitationof alexia: A case study.
Cortex, 15, 139-144.
daily reading practice, half of the students re- MOYER, S.B.(1982). Repeatedreading.Journal of Learning
quested to continue the story reading sessions. Disabilities, 45, 619-623.
Thus, one point of usefulness for this technique NEILL,K. (1980). Turn kids on with repeated reading.
may be that it encourages students to read TeachingExceptionalChildren,12, 63-64.
The method of repeated readings.
SAMUELS, S.J. (1979).
more, or at least, to have a more positive atti-
TheReading Teacher,4, 403-408.
tude towardreading, because with each repeti- P.A. (1980). On the acquisition of reading flu-
SCHREIBER,
tion of the same story the student usually ency. Journal of ReadingBehavior, 12(3), 177-186.
achieves speed improvement. SUCHER, F., & ALLRED, R.A. (1973). The Sucher-Allred
ReadingPlacementInventory.OklahomaCity, OK: The
Economy Company.
REFERENCES
APPLE SUPER-PILOT LANGUAGE REFERENCE MANUAL.
(1982). Cupertino,CA: Apple Computer,Inc. Footnote
BERKE, S. (Ed.) (1977). Reader'sdigest reading skill build- This paperis based on a dissertationsubmittedby the first
ers, level 2. Pleasantville, NY: Reader'sDigest Educa- authorto the Departmentof Psychology, FloridaState Uni-
tional Division. versity, in partial fulfillment of the requirementsfor the
BRYANT, J.L., & PAULSON, A.S. (1976). An extensionof Tu- doctoraldegree. The authorswould like to thankthe princi-
key's method of multiple comparisons to experimental pals and teachers of Kate Sullivan, Astoria Park, and W.T.
designs with random concomitant variables. Biome- Moore schools in the Leon CountySchool District for their
trika, 63(3), 631-638. cooperationand assistancewith this study.

188 READINGRESEARCH * Winter 1985

QUARTERLY XX/2

This content downloaded from 103.5.181.57 on Sun, 24 May 2015 17:10:35 UTC
All use subject to JSTOR Terms and Conditions