22
Children’s reading comprehension and oral reading fluency in easy text LOUISE SPEAR-SWERLING Southern Connecticut State University, New Haven, CT, USA Abstract. This study explored third-graders’ oral reading fluency (ORF) in easy text in relation to their third- and fourth-grade reading comprehension. It also examined the children’s performance on two different measures of text exposure, a self-report ques- tionnaire and a title-recognition test. Although third-graders’ ORF related significantly to their reading comprehension, oral language comprehension accounted for most of the variance in reading comprehension, whereas single word reading speed accounted for most of the variance in ORF. Third-grade reading comprehension and ORF each predicted unique variance in children’s scores on a fourth-grade state-mandated reading comprehension assessment. Scores on the self-report questionnaire correlated signifi- cantly with third-grade ORF and fourth-grade reading; the self-report accounted for reliable variance in ORF even with all of the other reading ability variables entered first. Results are consistent with the viewpoint that text exposure affects reading fluency. They also demonstrate that ORF is a valuable predictor of middle-elementary children’s reading comprehension, even when the ORF measure employs very easy text in which children achieve near-perfect word accuracy. Key words: Comprehension, Fluency, Reading, Text exposure Oral reading fluency (ORF) in text can be viewed as an indicator of general reading competence, involving facility with and integration of a range of important sub-word, word, and comprehension-level processes (Fuchs, Fuchs, Hosp, & Jenkins, 2001; Good, Simmons, & Kame’enui, 2001; Wolf & Katzir-Cohen, 2001). Fluent readers demonstrate several key characteristics: highly accurate word identification; automatic (i.e., fast and effortless) word identification; and attention to prosodic features in oral reading such as appropriate intonation, phrasing, and stress (Kuhn & Stahl, 2003; Rasinski & Hoffman, 2003). Typically measured as words read correctly per minute during oral reading of passages, ORF in elementary children correlates substantially with conventional measures of their reading comprehension, with correlations sometimes as high as Reading and Writing (2006) 19:199–220 Ó Springer 2006 DOI 10.1007/s11145-005-4114-x

Children’s Reading Comprehension and Oral Reading Fluency in Easy Text

Embed Size (px)

Citation preview

Children’s reading comprehension and oral reading fluency

in easy text

LOUISE SPEAR-SWERLINGSouthern Connecticut State University, New Haven, CT, USA

Abstract. This study explored third-graders’ oral reading fluency (ORF) in easy text inrelation to their third- and fourth-grade reading comprehension. It also examined thechildren’s performance on two different measures of text exposure, a self-report ques-tionnaire and a title-recognition test. Although third-graders’ ORF related significantly

to their reading comprehension, oral language comprehension accounted for most of thevariance in reading comprehension, whereas single word reading speed accounted formost of the variance in ORF. Third-grade reading comprehension and ORF each

predicted unique variance in children’s scores on a fourth-grade state-mandated readingcomprehension assessment. Scores on the self-report questionnaire correlated signifi-cantly with third-grade ORF and fourth-grade reading; the self-report accounted for

reliable variance in ORF even with all of the other reading ability variables entered first.Results are consistent with the viewpoint that text exposure affects reading fluency.They also demonstrate that ORF is a valuable predictor of middle-elementary children’s

reading comprehension, even when the ORF measure employs very easy text in whichchildren achieve near-perfect word accuracy.

Key words: Comprehension, Fluency, Reading, Text exposure

Oral reading fluency (ORF) in text can be viewed as an indicator ofgeneral reading competence, involving facility with and integration of arange of important sub-word, word, and comprehension-level processes(Fuchs, Fuchs, Hosp, & Jenkins, 2001; Good, Simmons, & Kame’enui,2001; Wolf & Katzir-Cohen, 2001). Fluent readers demonstrate severalkey characteristics: highly accurate word identification; automatic (i.e.,fast and effortless) word identification; and attention to prosodic featuresin oral reading such as appropriate intonation, phrasing, and stress(Kuhn & Stahl, 2003; Rasinski & Hoffman, 2003). Typically measured aswords read correctly per minute during oral reading of passages, ORF inelementary children correlates substantially with conventional measuresof their reading comprehension, with correlations sometimes as high as

Reading and Writing (2006) 19:199–220 � Springer 2006DOI 10.1007/s11145-005-4114-x

.80 or .90 (e.g., Fuchs et al., 2001). Moreover, ORF has been found torelate more strongly to reading comprehension than do measuresinvolving timed oral reading of isolated word lists (Fuchs et al., 2001;Thurlow & van den Broek, 1997), suggesting that ORF taps abilitiesbeyond rapid, accurate identification of single words.

Reading comprehension and ORF draw upon many of the sameunderlying processes, such as oral language comprehension, word iden-tification accuracy, and naming speed (Wolf & Katzir-Cohen, 2001). Therelationship between ORF and reading comprehension may be a re-ciprocal one (Meyer & Felton, 1999), with causality running in bothdirections. ORF likely is a causal influence on reading comprehension,because inaccurate or nonautomatic word reading impairs the reader’sability to understand a text. However, comprehension abilities may alsocausally influence ORF; for example, if a reader does not understand themeaning of words in a text, it may be difficult to read quickly or withappropriate phrasing and intonation.

Among typical readers, ORF develops during the early to middleelementary grades. In kindergarten and first grade, most children acquirean understanding of the alphabetic principle, knowledge of letter-soundcorrespondence, and accurate decoding skills for many common words.In youngsters learning to read English, fluency increases rapidly fromaround the end of first grade to third grade (Chall, 1996; Kuhn & Stahl,2003), providing an important foundation for dealing with the increas-ingly challenging texts, as well as the escalating reading volume, in themiddle elementary grades and beyond.

However, the strength of the relationship between ORF and readingcomprehension appears to decline with age (Jenkins & Jewell, 1993;Saiegh-Haddad, 2003). This finding may relate to developmental changesin the relative contributions of word identification and oral languagecomprehension to both ORF and reading comprehension. Jenkins andJewell (1993) suggest that the strong relationship between ORF andreading comprehension in young elementary children may relate pri-marily to the fact that both measures are highly sensitive to individualdifferences in word identification at beginning stages of reading devel-opment. At more proficient stages of reading, most children have highlyaccurate, automatic word identification, so reading comprehension beginsto depend relatively more heavily on language comprehension (Hoover &Gough, 1990; Rupley, Willson, & Nichols, 1998), reducing the strength ofthe relationship between ORF and reading comprehension. Consistentwith this line of reasoning, Saiegh-Haddad (2003) found that ORFand reading comprehension did not correlate significantly in a group ofHebrew- and Arabic-speaking adults reading adult-level text in their

200 LOUISE SPEAR-SWERLING

native languages; however, for both groups, correlations were significantwhen participants were reading a second language, English. In a secondlanguage, readers would typically have less well-developed languagecomprehension skills than in their native language, rendering individualdifferences in word identification relatively more important to readingcomprehension, and increasing the strength of the ORF–reading com-prehension relationship.

A key issue in assessment of ORF involves the grade level of text to beused in measurement. Experimental investigations of ORF often haveemployed grade-appropriate texts. However, for the poorer readers in agrade, grade-appropriate texts may be frustrating, and they make it dif-ficult to differentiate problems in word identification accuracy from thoseinvolving only speed or ease of reading. An ORF measure involving veryeasy text would presumably result in high levels of word accuracy fornearly all children, allowing the differentiation of children based only onspeed of reading text. It is not clear to what extent this kind of ORFmeasure would duplicate previous findings involving ORF in harder(grade-appropriate) text, because the former might tap underlying com-ponent reading abilities somewhat differently than does the latter. Forinstance, Carver and David (2001) found that children’s reading rate inrelatively easy text related more strongly to word identification accuracyand naming speed than to language comprehension. This result suggeststhat easy text might reduce the utility of ORF as a predictor of readingcomprehension in the middle elementary grades, when individual differ-ences in reading comprehension begin to revolve more around languagecomprehension than around word accuracy.

One important influence on both reading fluency and comprehensioninvolves children’s exposure to text, for example, through independentreading for pleasure. Practice reading text is generally believed to beessential for the development of fluency (Adams, 1990; Chall, 1996;LaBerge & Samuels, 1974; Stanovich, 1986). In addition, independentreading gives children exposure to a wide range of skills essential toreading comprehension, including new vocabulary and backgroundknowledge. Even in school, there may be substantial differences in chil-dren’s amount of exposure to text, based at least in part on children’sreading abilities (Stanovich, 1986) and the socioeconomic level of theschool (Duke, 2000). Of course, children who are avid readers outside ofschool have greater exposure to text than do those who do not read forpleasure.

In elementary children, exposure to text has been assessed in a numberof ways, including activity diaries that require children to log the numberof minutes each day spent reading (Anderson, Wilson, & Fielding, 1988),

201CHILDREN’S READING COMPREHENSION AND ORF

self-report questionnaires (Baker & Wigfield, 1999; Guthrie, Wigfield,Metsala, & Cox, 1999; Wang & Guthrie, 2004), and title-recognition orauthor-recognition measures (Allen, Cipielewski, & Stanovich, 1992;Cipielewski & Stanovich, 1992; Cunningham & Stanovich, 1991). Self-report questionnaires typically involve a series of questions about howoften the child reads for pleasure and about which books have been read,whereas checklist recognition measures provide a list of children’s books(or authors) interspersed with fake items to detect guessing; the task forchildren is to check off the names of all book titles or authors that theyare certain are real. Allen et al. (1992) found that the different measures oftext exposure – activity diaries, self-report questionnaires, and checklistrecognition measures – correlated significantly with each other.

Although activity diaries might be viewed as the most valid measure oftime spent reading out of school, they are time-consuming and may beespecially difficult for children, because they require estimating amountsof time (Allen et al., 1992). Self-report questionnaires and checklist rec-ognition measures are a faster, less burdensome way of getting at indi-vidual differences in exposure to text. An advantage of checklistrecognition measures is that they avoid social desirability effects, or thetendency to inflate the amount of time spent in an activity viewed associally desirable (Stanovich & West, 1989). Nevertheless, although socialdesirability effects may be a problem for self-report questionnaires, thereis some research to validate their use (Guthrie et al., 1999). Both self-report questionnaires and checklist recognition measures have been foundto relate significantly to children’s reading comprehension (Cipielewski &Stanovich, 1992; Cunningham & Stanovich, 1991; Guthrie et al., 1999).Checklist recognition measures have also been found to account for sig-nificant variance in children’s vocabulary development, spelling, andgeneral knowledge (Cunningham & Stanovich, 1991). However, therelationship of text-exposure measures to children’s oral reading fluencyhas been less widely studied.

This study explored third-graders’ ORF in very easy (first grade) textin relation to their reading comprehension. The children were also giventests of underlying component reading abilities (e.g., word identification,vocabulary, single word reading speed) and two different text-exposuremeasures, a self-report of independent reading for pleasure and a title-recognition measure. The following year, the children’s performance on astate-mandated reading assessment for fourth graders was examined todetermine which of the third-grade variables uniquely predicted fourth-grade reading comprehension. The main questions of the study werethese: (1) Would word accuracy, single word reading speed, and orallanguage comprehension account for comparable variance in third-graders’

202 LOUISE SPEAR-SWERLING

ORF vs. reading comprehension? (2) Would the two text-exposure mea-sures relate significantly to each other, to children’s ORF, or to theirreading comprehension? (3) How useful would ORF in easy text be, rela-tive to the other third-grade measures, in predicting children’s fourth-gradereading?

Method

Participants

Sixty-one third graders (29 male, 32 female, mean age=103.4 months,SD=3.2 months) participated. The children came from two schools intwo different districts, a suburban school (n=26) and an interdistrictmagnet school (n=35). Most youngsters at the suburban school camefrom middle-socioeconomic backgrounds, whereas those at the magnetschool represented a wider range of socioeconomic levels, both low- andmiddle-SES. The reading program at the suburban school employed acombination of literature-based instruction with some integration ofphonics, whereas the magnet school used Success for All (Slavin,Madden, Dolan, & Wasik, 1996). All of the children were native Englishspeakers and none received special-education services at the time of thestudy.

Materials and procedure

In January and February of third grade, children were tested individuallyon the following tests: from the Woodcock–Johnson PsychoeducationalBattery-Revised (WJ-R, Woodcock & Johnson, 1989), Word Identifica-tion (WI), Word Attack (WA), Listening Comprehension (LC), andPassage Comprehension (PC); the Peabody Picture Vocabulary Test-III(PPVT-III, Dunn & Dunn, 1997); ORF in two primer-level (middle offirst grade) passages of 100 words each; rapid naming of digits and first-grade-level printed words (two trials each); a title-recognition test; and aself-report of independent reading. Examiners were also asked to providea qualitative rating of children’s ORF on a 5-point scale (1=very labored,inaccurate, and/or expressionless, 5=highly accurate, expressive, andeffortless). Testing took approximately 45 minutes to 1 hour and wasadministered in a single session, with order of presentation counterbal-anced across participants. The following school year, in October of fourthgrade, children took the Connecticut Mastery Test (CMT, Connecticut

203CHILDREN’S READING COMPREHENSION AND ORF

Department of Education, 2000), a statewide assessment given to allpublic school children in Connecticut in fourth, sixth, and eighth grades.Fourth-grade CMT scores in reading were available for 57 of the 61 third-graders who participated in the study.

Woodcock–Johnson-Revised (WJ-R) subtestsFour subtests of the WJ-R were administered: WI, WA, PC, and LC. WIrequires the child to read a series of real words presented out of context.WA requires the child to read nonwords such as tat, presented out ofcontext. PC uses a cloze format; the child reads a series of sentences orpassages and provides a contextually appropriate word to fit in a blank.LC has a format similar to that of PC except that the examiner reads aseparate set of sentences or passages aloud to the child. The split-half,Spearman–Brown reliabilities for these subtests are as follows: .94 (WI),.91 (WA), .88 (PC), and .81 (LC).

Peabody Picture Vocabulary Test-III (PPVT-III)The PPVT-III is a measure of receptive vocabulary in which childrenlisten to the examiner read a word, and then pick an appropriate pictureto go with the word out of a series of four options. The internal reliabilityfor this test is .94.

Rapid naming and fluency tasksThe children were administered two sets of rapid naming tasks. Rapidnaming of digits required the child to name an array of single digitnumbers (1, 2, 3, 4, 5, 6, 8, and 9) arranged in 8 rows of 6 digits each. Thetotal time the child took to name the entire array was recorded with astopwatch, and the final score was converted to digits named per secondso as to make higher scores indicative of better performance. Two trials ofthis task were done, using the same digits arranged in a different sequence;the Pearson correlation between the two trials was .88 (P<.001).

The second rapid naming task involved rapid naming of printedwords. The printed words were high-frequency words (e.g., me, play, get)selected from pre-primer and primer-level graded word lists on the JohnsOral Reading Inventory, Form A (Johns, 1994). The task was administeredand scored in a manner similar to the digit-naming task, using an array of48 different words arranged in 8 rows of 6 words each. Two trials of thistask also were done, with the correlation between the trials again .88(P<.001).

Oral reading fluency was assessed through the child’s reading of twoprimer-level narrative passages, of 100 words each, both adapted withminor modifications from the Johns Oral Reading Inventory. Children

204 LOUISE SPEAR-SWERLING

were instructed to read the passages aloud ‘‘as well as they could’’ and to‘‘sound out’’ any unknown words; however, word accuracy on the pas-sages was very high (mean>99%) and most youngsters did not struggleto read any words. As in the case of the rapid naming tasks, each child’stime was recorded with a stopwatch and scores were converted to wordsread correctly per second. Children’s performance on the two differentpassages correlated .94 (P<.001).

Measures of text exposureTwo different text-exposure measures were given. Directions from bothmeasures were read aloud to all children. The title-recognition measureemployed the usual format of a set of real book titles interspersed with aset of fake book titles (foils). There were 23 real books, representing bothfiction (e.g., The Not-So-Jolly Roger, Catwings, Tales of Uncle Remus)and non-fiction (e.g., Incredible Cross-Sections, To the Top!) appropriateto middle-elementary children, as well as 15 foils (e.g., Searching theWilds, The Rollaway Bed). Books that had been used in classroominstruction or assigned for homework were screened out of the measureby children’s third-grade teachers. The foils, general format of the test,and administration procedures were adapted from the work of Stanovichand his colleagues (Allen et al., 1992; Cunningham & Stanovich, 1991).Children were instructed to check off all the names of books that theywere confident were real books; they were cautioned not to guess.Analyses used a derived score that was calculated from the number of realbooks a child correctly checked off minus the number of false alarms tofoils. Cronbach’s alpha for the number of real books checked by childrenwas .85.

The second measure was a self-report of independent reading forpleasure. The self-report involved six questions, as shown in AppendixTable A.1. The first three questions asked children about how often theyread ‘‘just for fun’’; it was emphasized that children should not countreading done only as part of schoolwork or homework. These questionswere adapted from a similar measure used by Baker and Wigfield (1999).The first question (‘‘Have you read any books or magazines for fun,outside of school, in the past week?’’) was scored 1 if the child answered‘‘yes’’ and 0 for ‘‘no.’’ The second question asked children to name thetitles of all books or magazines read in the past week; the score was thenumber of real books or magazines named. The third question askedchildren to estimate about how often they read for fun, with four mul-tiple-choice options: ‘‘almost never’’ (scored 0); ‘‘about once a month’’(scored 1); ‘‘about once a week’’ (scored 2); or ‘‘almost every day’’(scored 3).

205CHILDREN’S READING COMPREHENSION AND ORF

The last three questions were adapted from the work of Stanovich andWest (1989) and asked children to name their favorite books, magazines,and authors. To receive credit, a child had to name a specific book,magazine, or author that appeared to be real (not, e.g., ‘‘that magazineabout cars’’ or ‘‘the author of Harry Potter’’). Children were encouragedto name as many ‘‘favorites’’ as they wished. Occasionally children’s re-sponses involved names that sounded plausible but were unfamiliar to theinvestigator; these could almost always be subsequently verified as gen-uine. Most of the time, children named very recognizable books (e.g., ThePolar Express, Ramona), magazines (e.g., Ranger Rick, Sports Illustrated),and authors (e.g., Beverly Cleary, J. K. Rowling), so no verification wasnecessary. The child’s score for each of these questions was the number ofbooks, magazines, or authors named. Cronbach’s alpha for the total scoreon the self-report measure was .59.

Connecticut Mastery Test (CMT)The CMT has two subtests for assessing reading, the Degrees of ReadingPower (DRP, Touchstone Applied Science Associates, 2001) and ReadingComprehension (RC, Connecticut State Department of Education, 2000).The DRP uses a maze format. Children are given a series of informationalpassages containing blanks and must provide a contextually appropriateword to fit in the blank, with multiple-choice options provided for eachblank. In RC, children are given a series of passages that include narrative,expository, and poetic material. After reading the passages, they answer avariety of both multiple-choice and open-ended questions (e.g., ‘‘write abrief summary of the story’’). Answers to multiple-choice questions arescored 0 (incorrect) or 1 (correct), whereas answers to open-ended ques-tions are scored 0 (unsatisfactory), 1 (marginal), or 2 (satisfactory).The reading subtests are group-administered over several sessions, andalthough there are time limits for the subtests, most children finish the testwithin the allotted time limits. Children’s reading achievement on theCMT is reported as a scaled score representing a composite of perfor-mance across the two subtests. Cronbach’s alpha for the DRP was .95 andfor RC was .87 (Connecticut State Department of Education, 2002).

Results

Descriptive statistics for third-grade measures

Table 1 lists the means and standard deviations for all of the third-grademeasures. WJ-R and PPVT-III scores are standard scores with a mean of

206 LOUISE SPEAR-SWERLING

100 and a standard deviation of 15; thus, the means shown in the table, allwell above 100, illustrate that the children were generally high-achieving.Scores for the rapid naming tasks and ORF are given as the correctnumber of items named per second, averaged across two trials. On thesetimed tasks, children read words in passages significantly faster than theynamed digits (t=4.428, P<.001) or words out of context (t=9.098,P<.001).

On the title-recognition measure, children checked off 6.7 real books,on average (SD=4.8); however, they also showed a relatively high rate offalse alarms to foils (mean=3.2, SD=3.5). On the self-report measure,most children said they had read ‘‘for fun’’ in the previous week, namingan average of 1.8 books or magazines (SD=1.6). When asked how often

Table 1. Means and standard deviations (third-grade measures).

Measure Mean SD

WJ-R LC 118.9 16.5

WJ-R WI 116.1 15.2

WJ-R WA 112.1 16.6

WJ-R PC 119.0 12.6

PPVT 110.6 13.9

ORF (passages) 2.3 .7

Examiner’s fluency rating 3.7 .9

Rapid naming-words (Word speed) 1.7 .5

Rapid naming-digits (Digit speed) 1.9 .4

Title recognition test

Derived score 3.4 2.7

Real books checked 6.7 4.8

False alarms to foils 3.2 3.5

Reading self-report

Have you read for fun in past week? .8 .4

Number of books/magazines (past week) 1.8 1.6

How often do you read for fun? 2.4 .8

Number of favorite books named 3.2 1.6

Number of favorite magazines named 1.0 .9

Number of favorite authors named 1.7 1.1

Note: WJ-R subtest scores are standard scores with a mean of 100 and a standard devi-

ation of 15; ORF and rapid naming tasks are expressed as number of words/digits read

correctly per second; Examiner’s fluency rating is a qualitative rating based on a 5-point

scale.

207CHILDREN’S READING COMPREHENSION AND ORF

they read for pleasure, most children said either ‘‘about once a week’’ or‘‘almost every day.’’ On average, children were able to name morefavorite books (mean=3.2, SD=1.6) than favorite magazines(mean=1.0, SD=.9) or favorite authors (mean=1.7, SD=1.1). More-over, although every child in the study was able to name at least onefavorite book (minimum=1, maximum=8), some children were unableto name any favorite magazines (minimum=0, maximum=3) or authors(minimum=0, maximum=4).

Relationships among component reading abilities, ORF, and readingcomprehension

Table 2 displays the Pearson correlations of all study measures with third-grade ORF, third-grade passage comprehension, and fourth-grade CMTreading achievement. As shown in the second column, the examiner’squalitative rating of fluency correlated highly (r=.86, P<.001) with ORFas measured by number of words read correctly per second. That is,children who read passages quickly and accurately were typically rated as

Table 2. Pearson correlations of study measures with third-grade ORF, third-grade PC,

and fourth-grade reading comprehension.

Measure

ORF

(Grade 3)

WJ-R PC

(Grade 3)

CMT reading

(Grade 4)

ORF – .57*** .65***

WJ-R PC – – .70***

Examiner’s fluency rating .86*** .62*** .65***

Digit speed .44*** .12 .09

Word speed .82*** .44*** .47***

WJ-R LC .37** .56*** .39**

WJ-R WI .73*** .68*** .66***

WJ-R WA .60*** .46*** .50***

PPVT .44*** .74*** .59***

Title recognition (derived score) ).11 ).09 ).06Reading self-report (total score) .37** .20 .26*

Oral language .45*** .71*** .54***

Word accuracy .70*** .60*** .62***

Note: Oral language is an average of children’s z-scores for WJ-R LC and the PPVT; Word

accuracy is an average of children’s z-scores for WJ-R WI and WA.

*P<.05; **P<.01; ***P<.001.

208 LOUISE SPEAR-SWERLING

also reading very expressively and with ease. Correlations of the com-ponent reading measures with third-grade ORF and third-grade passagecomprehension all were significant, with one exception: Digit namingspeed correlated with ORF (r=.44, P<0.01) but not with passage com-prehension (r=.12, ns). Also, ORF and passage comprehension corre-lated significantly with each other (r=.57, P<.001).

The far right-hand column of the table shows the correlations of thethird-grade measures with children’s performance on the state-mandatedassessment in fourth grade. Once again, all component reading-relatedmeasures except for digit speed correlated significantly with fourth-gradereading. Fourth-grade reading also correlated significantly with third-grade ORF (r=.65, P<.001) and third-grade passage comprehension(r=.70, P<.001).

Finally, for use in subsequent analyses, two composite scores weregenerated for each child. Oral language was an average of children’sz-scores for WJ-R Listening Comprehension and the PPVT. Wordaccuracy was an average of children’s z-scores for WJ-R WI and WA. Asshown in the bottom two rows of Table 2, both composite variablescorrelated significantly with ORF, third-grade passage comprehension,and fourth-grade reading.

Contributions of component reading abilities to ORF vs. passagecomprehension

To address the first question of the study, a series of hierarchicalregressions explored the relative contributions of oral language, wordaccuracy, and speed of single word reading to ORF as compared to PC inthird grade. The previously described composite z-scores were used fororal language and word accuracy; scores from the rapid naming-wordstask, averaged across two trials, were used for speed of single wordreading, or word speed. The component reading variables were enteredindividually rather than in blocks. Table 3 lists the results of theseregressions, with the left-hand column of the table showing the order ofvariable entry for the component variables, the middle column showingthe results of each regression for ORF, and the right-hand column theresults for passage comprehension. As shown in the middle column, thestrongest contributor to ORF was word speed, which, even with wordaccuracy and oral language accounted for, still explained approximately24% of the variance in ORF. With oral language and word speed ac-counted for, word accuracy explained a small but significant percentageof variance in ORF (3.5%). With word accuracy and word speed

209CHILDREN’S READING COMPREHENSION AND ORF

Table

3.Hierarchicalregressionanalysesusingcomponentreadingvariablesto

predictthird-gradePassagecomprehensionandOralreading

fluency.

Componentreading

variablesandorder

ofvariable

entry

Oralreadingfluency

Passagecomprehension

DR2

Unstand.

Beta

SEB

bDR2

Unstand.

Beta

SEB

b

1.Word

speed

.672***

.620

.085

.620***

.191***

.086

.106

.086

2.Orallanguage

.041**

.149

.082

.136

.368***

.604

.103

.549***

3.Word

accuracy

.035**

.271

.096

.256**

.047*

.313

.120

.295*

1.Orallanguage

.198***

.149

.082

.136

.507***

.604

.103

.549***

2.Word

speed

.515***

.620

.085

.620***

.052*

.086

.106

.086

3.Word

accuracy

.035**

.271

.096

.256**

.047*

.313

.120

.295*

1.Word

accuracy

.492***

.335

.091

.316**

.362***

.368

.099

.347***

2.Word

speed

.242***

.625

.086

.625***

––

––

3.Orallanguage

––

––

.239***

.606

.103

.551***

210 LOUISE SPEAR-SWERLING

1.Word

speed

.672***

.625

.086

.625***

.191***

.086

.106

.086

2.Word

accuracy

.062**

.335

.091

.316**

.178***

.313

.120

.295*

3.Orallanguage

––

––

.236***

.604

.103

.549***

1.Orallanguage

.198***

.149

.082

.136

.507***

.606

.103

.551***

2.Word

accuracy

.313***

.271

.096

.256**

.095***

.368

.099

.347***

3.Word

speed

.238***

.620

.085

.620***

––

––

1.Word

accuracy

.492***

.335

.091

.316**

.362***

.368

.099

.347***

2.Orallanguage

––

––

.239***

.606

.103

.551***

3.Word

speed

.242***

.625

.086

.625***

––

––

Note:Word

speedischildren’saveragescore

ontherapid

naming-w

ordstask;Orallanguageisanaverageoftheirz-scoresforWJ-R

LCandthe

PPVT;Word

accuracy

isanaverageoftheirz-scoresforWJ-R

WIandWA.

*P<

.05;**P<

.01;***P<

.001.

211CHILDREN’S READING COMPREHENSION AND ORF

accounted for, oral language did not explain any additional variance inORF. Across all six regressions for ORF, standardized beta weights (bs)were consistently largest for word speed and nonsignificant for oral lan-guage.

For passage comprehension, the pattern was reversed, as shown in thefar right-hand column of Table 3. With the other two variables con-trolled, oral language explained the largest amount of the variance inpassage comprehension, approximately 24%. With oral language andword speed controlled, word accuracy still explained a small but signifi-cant percentage of variance in PC (4.7%). However, with oral languageand word accuracy accounted for, word speed did not explain anyadditional variance in PC. Across the six regressions for PC, oral lan-guage consistently had the largest beta weights, whereas the beta weightsfor word speed were consistently nonsignificant.

These results suggest that, for the third graders in this study, ORF ineasy text tapped speed of single word reading, and to a lesser extent, wordaccuracy, more than oral language comprehension. However, PC ap-peared to tap oral language comprehension and word accuracy more thanword speed.

Text-exposure measures and their relationships to reading ability

Regarding the second question of the study, Table 2 illustrates that thetitle-recognition test did not correlate significantly with any readingmeasure. However, the total reading self-report score did correlate withORF (r=.37, P<.01) and with fourth-grade CMT score (r=.26, P<.05).A closer look at the individual items on the self-report showed that threeof the six questions correlated significantly with ORF: the third question,which asked children how often they read for pleasure (r=.30, P<.05),the favorite magazines question, and the favorite authors question (bothr=.31, P<.05). Question three and the favorite authors question alsocorrelated significantly with third-grade passage comprehension (r=.27,P<.05, and r=.28, P<.05, respectively), although the correlation be-tween total self-report score and passage comprehension did not attainsignificance. Only question one, which asked children whether they hadread for pleasure in the past week, correlated significantly with fourth-grade CMT score (r=.29, P<.05). Although the title-recognition test didnot correlate significantly with overall performance on the self-report(r=.10, ns), one self-report question did approach significance with title-recognition score: the favorite books question (r=.23, P=.08).

212 LOUISE SPEAR-SWERLING

An additional hierarchical regression examined the extent of overlapbetween the component reading measures, on the one hand, and the self-report, on the other, in accounting for children’s ORF. With wordaccuracy, oral language, word speed, and PC entered first, children’sscores on the self-report still accounted for a significant, though small,amount of variance in ORF, about 2.9% (P<.01, b=.175, P<.01). Thus,the self-report measure did not overlap entirely with the reading abilitiesassessed in the study, but rather accounted for some unique variance inchildren’s oral reading fluency.

Third-grade measures as predictors of fourth-grade reading comprehension

To address the final question of the study, a standard multiple regres-sion analysis was used to predict children’s performance on the fourth-grade state-mandated assessment (CMT), with third-grade PC, ORF,single word reading speed, oral language, word accuracy, and self-reportscore entered as predictors. Together, the six predictor variables ac-counted for approximately 60% of the variance in fourth-grade CMTscore, R2=.620, adjusted R2=.574, F(6, 50)=13.600, P<.001. However,only two of the standardized beta weights were significant, those forpassage comprehension (b=.401, P<.01) and for oral reading fluency(b=.407, P<.05).

Hierarchical regression analyses further explored the relative con-tributions of the six predictor variables to fourth-grade readingcomprehension. As shown in the left-hand column of Table 4, eachindividual third-grade variable was entered first, with the remainingfive variables entered together in a block. The order of entry was thenreversed, to determine the influence of each individual variable with theother five variables accounted for. The three right-hand columnsindicate the final unstandardized weights, standard errors, and stan-dardized betas for all six predictor variables.

When entered first in the regression analyses, all of the third-gradevariables accounted for significant variance in fourth-grade reading,ranging from a high of 49% of the variance for PC (P<.001) to a low of6.9% of the variance for the self-report measure (P<.05). However, onlytwo variables had significant final beta weights, and only two variablesaccounted for unique variance in fourth-grade reading with the othervariables controlled: PC (b=.401, P<.01, DR2=.067, P<.01) and ORF(b=.407, P<.05, DR2=.041, P<.05).

213CHILDREN’S READING COMPREHENSION AND ORF

Discussion

Answering the study questions

The ORF measure involving easy text primarily tapped single wordreading speed, not oral language comprehension, similar to the findings of

Table 4. Hierarchical regression analyses using third-grade measures to predict fourth-

grade CMT reading comprehension.

Third-grade variables and order

of variable entry

CMT reading comprehension

DR2 Unstand.

Beta

SEB b

1. PC .490*** .413 .139 .401**

2. ORF, WAcc, OL, WS, RSR .130*

1. ORF, WAcc, OL, WS, RSR .553***

2. PC .067**

1. ORF .420*** .422 .182 .407*

2. PC, WAcc, OL, WS, RSR .200**

1. PC, WAcc, OL, WS, RSR .579***

2. ORF .041*

1. WAcc .379*** .173 .137 .159

2. ORF, PC, OL, WS, RSR .242***

1. ORF, PC, OL, WS, RSR .608***

2. WAcc .012

1. OL .292*** .076 .139 .068

2. WAcc, ORF, PC, WS, RSR .328***

1. WAcc, ORF, PC, WS, RSR .618***

2. OL .002

1. WS .217*** ).121 .159 ).1122. OL, WAcc, ORF, PC, RSR .403***

1. OL, WAcc, ORF, PC, RSR .616***

2. WS .004

1. RSR .069* ).007 .026 ).0262. WS, OL, WAcc, ORF, PC .551***

1. WS, OL, WAcc, ORF, PC .620***

2. RSR .001

Note: ORF, Oral Reading Fluency; PC, Passage Comprehension; WAcc, Word Accuracy;

OL, Oral Language; WS, Word Speed; RSR, Reading Self-Report.

*P<.05; **P<.01; ***P<.001.

214 LOUISE SPEAR-SWERLING

Carver and David (2001), whereas the PC measure showed the reversepattern: It primarily tapped oral language comprehension, not wordspeed. Both ORF and PC tapped word accuracy. Nevertheless, despitethe use of easy text, there was a substantial, statistically significant rela-tionship between ORF and PC in third graders (r=.57, P<.001).

Only one of the text-exposure measures, the self-report questionnaire,related significantly to children’s reading; it correlated significantly withboth ORF and fourth-grade reading comprehension. The self-reportquestionnaire accounted for unique variance in ORF even with all of theother reading-ability variables entered first, indicating that the self-reportdid not overlap entirely with the reading abilities that were tested. Inother words, the self-report appeared to draw upon something besideschildren’s component reading abilities, perhaps their motivation to read(Baker & Wigfield, 1999; Guthrie et al., 1999) or their reading volume.

Third-graders’ ORF was very helpful in predicting their fourth-gradereading comprehension, accounting for a larger percentage of variancethan any other variable except for third-grade reading comprehension.Furthermore, even when entered after all of the other third-grade vari-ables, ORF still accounted for unique variance in fourth-grade reading.That is, ORF improved the prediction of fourth-grade reading compre-hension beyond what was accounted for by third-graders’ PC and theircomponent reading abilities of word accuracy, oral language, and wordspeed. These findings suggest that ORF, even in very easy text, tapssomething beyond important component reading processes, perhaps theability to integrate these processes in a fast, effortless manner (Fuchset al., 2001; Good et al., 2001; Wolf & Katzir-Cohen, 2001).

Another way to think about the results of the regression analyses inTable 4 is in terms of children’s growth in reading comprehension fromthird to fourth grade (see, e.g., Cipielewski & Stanovich, 1992). Thisinterpretation requires considerable caution, because the reading com-prehension measures used in third and fourth grade were different.However, it appears that individual differences in ORF were particularlyimportant in accounting for children’s growth in reading comprehen-sion, relative to the other measures that were given. This explanation isconsistent with developmental models of reading that highlight theimportance of fluency to reading comprehension in middle-elementaryyoungsters (e.g., Chall, 1996; Rupley et al., 1998).

The failure to find a relationship between the title-recognition measureand the self-report, as well as between the title-recognitionmeasure and anyreading ability, was puzzling given numerous other studies documentingthese relationships (e.g., Allen et al., 1992; Cipielewski & Stanovich, 1992;Cunningham & Stanovich, 1991). The lack of significant findings may

215CHILDREN’S READING COMPREHENSION AND ORF

relate in part to the limited number of items on the title-recognition testused in this study. In addition, despite admonitions to children not to guessat titles, the number of ‘‘false alarms’’ to fake titles was relatively high. Bycontrast, on the latter half of the self-report, where children had to generatespecific names of books, magazines, and authors, they typically namedeasily recognizable items and almost never appeared to be guessing.

To sum up, these results suggest that the use of very easy text in anORF measure, such that almost all children achieve perfect or near-perfectword accuracy, taps somewhat different underlying component abilitiesthan do standardized measures of reading comprehension in this agegroup. Nevertheless, children’s ORF correlated significantly with theirreading comprehension and was the best predictor of fourth-grade readingcomprehension with third-grade reading comprehension accounted for.Thus, the findings support the viewpoint that ORF is an important indi-cator of overall reading competence in middle-elementary children. Theresults also demonstrated a relationship between ORF and text exposure asmeasured by the self-report, even after accounting for component readingabilities and PC.

Limitations

Some limitations of the study must be noted. The sample size was smalland involved only two different schools; thus, it was not feasible toanalyze the data beyond the level of individual students (e.g., at the levelof the classroom or school). Also, most children were relatively high-achieving readers. Component reading abilities such as word accuracymight well have emerged as more critical to both ORF and readingcomprehension in a sample containing many low-achieving youngsters(see, e.g., Shankweiler et al., 1999). The study focused upon middle-ele-mentary youngsters, and the relationships obtained could be different forchildren at other stages of reading development, particularly in the case ofmore advanced readers, for whom there may be a weaker relationshipbetween ORF and reading comprehension (Jenkins & Jewell, 1993; Sai-egh-Haddad, 2003). In addition, the reading comprehension measuresused in this study did not emphasize speed; the third-grade measure is anuntimed, individually administered test, and the fourth-grade measure hasfairly liberal time limits. Word speed might be more important to readingcomprehension on a test involving stringent timing. Finally, the reliabilityof the self-report measure, at .59, was only marginally adequate, perhapsbecause of the limited number of items on the questionnaire. Increasingthe number of items – for instance, by asking questions related to different

216 LOUISE SPEAR-SWERLING

domains of reading (Guthrie et al., 1999) – might improve the reliabilityof the self-report.

Educational implications

These results indicate that text difficulty may be an important variable toconsider in investigations and educational evaluations of ORF. In partic-ular, ORF in easy, independent-level text may tap underlying processessuch as oral language comprehension and single word reading speed ratherdifferently than do conventional measures of reading comprehension.However, the findings also are consistent with the viewpoint (e.g., Fuchset al., 2001; Good et al., 2001) that ORF, indexed by the number of wordsread correctly per second in text, is a useful tool in identification of at-riskreaders, at least in this age group. TheORFmeasure used in this study tookonly a few minutes per child to administer and score. It would require onlyminimal training of teachers as comparedwithmany standardized tests, yetit was a better predictor of fourth-grade reading comprehension than anyother variable except for third-grade reading comprehension.Furthermore,children with fast, accurate reading in text were consistently rated byexaminers as having expressive and effortless oral reading, suggesting thatthe words-per-minute (or words-per-second) index is an excellent proxy forchildren’s attention to prosodic features during text reading. Future studiesshould include an examination of the underlying processes tapped by ORFmeasures, as well as the relationship between ORF and reading compre-hension, under varying conditionsof text difficulty at a rangeof grade levels.

The study provides some additional support for the use of self-reportmeasures in assessing text exposure in children. Although the self-report inthis study appeared to show some social desirability effects – because re-sponses to questions about how often children read for pleasure wereskewed positively – some children did admit to reading for pleasure rarelyor never. The ‘‘favorite authors’’ and ‘‘favorite magazines’’ questions alsoshowed promise as ways of assessing text exposure with children in this agegroup, because they related significantly to children’s ORF and readingcomprehension; moreover, children rarely guessed answers to these ques-tions, andwhen they did, guesses were easy to detect. Evenwith componentreading abilities and reading comprehension controlled, the self-reportcontinued to explain unique variance in children’s ORF. For educatorsattempting to develop children’s fluency, these findings imply that it isimportant to address not only key reading-related abilities such as wordaccuracy and oral language comprehension, but also factors that increasetext exposure, such as motivation to read and independent reading.

217CHILDREN’S READING COMPREHENSION AND ORF

Acknowledgements

This research was supported by a Connecticut State University researchgrant. I would like to extend my thanks to the children, teachers, andprincipals who participated in this project; to the staff at the Center forCommunity and School Action Research (CCSAR) at SCSU, whoprovided assistance with data analysis; and to Karen Christman andLeslie Barry for their help with data collection.

Appendix

Table A.1. Reading self-report measure.

1. Have you read any books or magazines for fun, outside of school, in the past week?

(Do not count reading done as part of your homework.)

YES NO

2. If ‘‘yes,’’ ask the child to name as many titles of books and/or magazines as possible

that he or she has read in the past week:

TITLE(S):

3. About how often do you read a book or magazine outside of school, for fun? (Again,

do not count reading done as part of your homework.)

ALMOST NEVER ABOUT ONCE A WEEK

ABOUT ONCE A MONTH ALMOST EVERY DAY

4. Out of all the books you have read yourself, which ones are your favorites? Name the

titles of your favorite books. These can include books you read a long time ago, as well

as books you have read more recently.

FAVORITE BOOK(S):

5. Do you have any favorite magazines that you read regularly? Name your favorite

magazines.

FAVORITE MAGAZINE(S):

6. Do you have any favorite authors whose books you look forward to reading? Name

your favorite authors.

FAVORITE AUTHOR(S):

218 LOUISE SPEAR-SWERLING

References

Adams, M. J. (1990). Beginning to read: Thinking and learning about print. CambridgeMA: MIT Press.

Allen, L., Cipielewski, J., & Stanovich, K. E. (1992). Multiple indicators of children’sreading habits and attitudes: Construct validity and cognitive correlates. Journal ofEducational Psychology, 84, 489–503.

Anderson, R. C., Wilson, P. T., & Fielding, L. G. (1988). Growth in reading and howchildren spend their time outside of school. Reading Research Quarterly, 23, 285–303.

Baker, L., & Wigfield, A. (1999). Dimensions of children’s motivation for reading andtheir relations to reading activity and reading achievement. Reading Research

Quarterly, 34, 452–477.Carver, R. P., & David, A. H. (2001). Investigating reading achievement using a causal

model. Scientific Studies of Reading, 5, 107–140.

Chall, J.S. (1996). Stages of reading development (2nd ed.). Fort Worth, TX: HarcourtBrace.

Cipielewski, J., & Stanovich, K. E. (1992). Predicting growth in reading ability from

children’s exposure to print. Journal of Experimental Child Psychology, 54, 74–89.Connecticut Department of Education. (2000). Connecticut mastery test-language arts

handbook. Retrieved December 31st, 2000, from http: www.state.ct.us/sde/ .

Connecticut Department of Education. (2002). Connecticut mastery test technical report.Hartford, CT: Author.

Cunningham, A. E., & Stanovich, K. E. (1991). Tracking the unique effects of printexposure in children: Associations with vocabulary, general knowledge, and

spelling. Journal of Educational Psychology, 83, 264–274.Duke, N. K. (2000). Print environments and experiences offered to first-grade students

in very low- and very high-SES school districts. Reading Research Quarterly, 35,

456–457.Dunn, L., & Dunn, M. (1997). Peabody Picture Vocabulary Test-III Circles Pines, MN:

American Guidance Service.

Fuchs, L. S., Fuchs, D., Hosp, M., & Jenkins, J. R. (2001). Oral reading fluency as anindicator of reading competence: A theoretical, empirical, and historical analysis.Scientific Studies of Reading, 5, 239–256.

Good, R. H., Simmons, D. C., & Kame’enui, E. J. (2001). The importance and decision-making utility of a continuum of fluency-based indicators of foundational readingskills for third-grade high-stakes outcomes. Scientific Studies of Reading, 5, 257–288.

Guthrie, J. T., Wigfield, A., Metsala, J. L., & Cox, K. E. (1999). Motivational and

cognitive predictors of text comprehension and reading amount. Scientific Studies ofReading, 3, 231–256.

Hoover, W. A., & Gough, P. B. (1990). The simple view of reading. Reading and

Writing: An Interdisciplinary Journal, 2, 127–160.Jenkins, J. R., & Jewell, M. (1993). Examining the validity of two measures for

formative teaching: Reading aloud and maze. Exceptional Children, 59, 321–342.

Johns, J. (1994). Johns basic reading inventory (7th ed.). Dubuque, IA: Kendall/HuntPublishing Company.

Kuhn, M. R., & Stahl, S. A. (2003). Fluency: A review of developmental and remedialpractices. Journal of Educational Psychology, 95, 3–21.

LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic informationprocessing in reading. Cognitive Psychology, 6, 293–323.

219CHILDREN’S READING COMPREHENSION AND ORF

Meyer, M. S., & Felton, R. H. (1999). Repeated reading to enhance fluency: Oldapproaches and new directions. Annals of Dyslexia, 49, 283–306.

Rasinski, T. V., & Hoffman, J. V. (2003). Theory and research into practice: Oralreading in the school literacy curriculum. Reading Research Quarterly, 38, 510–522.

Rupley, W. H., Willson, V. L., & Nichols, W. D. (1998). Exploration of the

developmental components contributing to elementary school children’s readingcomprehension. Scientific Studies of Reading, 2, 143–158.

Saiegh-Haddad, E. (2003). Bilingual oral reading fluency and reading comprehension:The case of Arabic-Hebrew (L1) – English (L2) readers. Reading and Writing: An

Interdisciplinary Journal, 16, 717–736.Shankweiler, D., Lundquist, E., Katz, L., Stuebing, K., Fletcher, J., Brady, S., Fowler,

A., Dreyer, L., Marcuione, K., Shaywitz, S., & Shaywitz, B. (1999). Comprehension

and decoding: Patterns of association in children with reading difficulties. ScientificStudies of Reading, 3, 69–94.

Slavin, R. E., Madden, N. A., Dolan, L. J., & Wasik, B. A. (1996). Every child, every

school: Success for All. Newbury Park, CA: Corwin.Stanovich, K. E. (1986). Matthew effects in reading: Some consequences of individual

differences in the acquisition of literacy. Reading Research Quarterly, 21, 36–406.

Stanovich, K. E., & West, R. F. (1989). Exposure to print and orthographic processing.Reading Research Quarterly, 24, 402–433.

Thurlow, R., & van den Broek, P. (1997). Automaticity and inference generation.Reading and Writing Quarterly, 13, 165–184.

Touchstone Applied Science Associates. (2001). The degrees of reading power program:DRP in brief. Brewster, NY: Author.

Wang, J. H., & Guthrie, J. T. (2004). Modeling the effects of intrinsic motivation,

extrinsic motivation, amount of reading, and past reading achievement on textcomprehension between U. S. and Chinese students. Reading Research Quarterly,39, 162–186.

Wolf, M., & Katzir-Cohen, T. (2001). Reading fluency and its intervention. ScientificStudies of Reading, 5, 211–238.

Woodcock, R. W., & Johnson, M. B. (1989). Woodcock-Johnson tests of achievement-

revised Boston, MA: Houghton-Mifflin.

Address for correspondence: Louise Spear-Swerling, Department of Special Educationand Reading, Southern Connecticut State University, 501 Crescent St. New Haven, CT06515, USA

Phone:+1-203-392-5949; Fax:+1-203-392-5927;E-mail: [email protected]

220 LOUISE SPEAR-SWERLING