Accuracy and Fluency in List and Context Reading of Skilled and RD ...

Learning Disabilities Research & Practice, 18(4), 237–245 C 2003, The Division for Learning Disabilities of the Council for Exceptional Children Copyright

Accuracy and Fluency in List and Context Reading of Skilled and RD Groups: Absolute and Relative Performance Levels Joseph R. Jenkins University of Washington

Lynn S. Fuchs Peabody College of Vanderbilt University

Paul van den Broek, Christine Espin, and Stanley L. Deno University of Minnesota The purpose of this study was to examine (1) the performance levels and the magnitude of performance difference between students with reading disabilities (RD) and skilled readers when reading a typical classroom text; (2) the hypothesis that students with RD have specific difficulty using context in such a way that reading fluency is affected; and (3) whether RD subtypes may be differentiated according to performance on contextual and context-free reading tasks. Two groups of fourth graders (85 skilled readers and 24 students with RD) completed a standardized test of reading comprehension, read aloud a folktale, and read aloud the folktale’s words in a randomly sequenced list. Performance was scored as correct rate and percentage correct. Based on the number of words per idea unit in the passage, we also estimated the rate at which reader groups encountered and processed text ideas. Compared to the RD group, skilled readers read three times more correct words per minute in context, and showed higher accuracy and rates on all measures. Both context and isolated word-reading rates were highly sensitive to impairment. We found no evidence for RD subtypes based on these measures. Results illustrate differences in reading levels between the two groups, the temporal advantage skilled readers have in linking text ideas, how word reading differs as a function of task format and performance dimension, and how limited word-identification skills (not comprehension) produce contextual reading difficulties for students with RD.

Much of the research in reading disabilities (RD) has focused on understanding and treating word-level reading problems. A focus on word reading seems justified for three reasons. First, individuals with specific RD experience extraordinary difficulty in acquiring word-reading proficiency (Ehri & Saltmarsh, 1995; Morrison, 1987; Stanovich, 1988). Second, children with specific RD, by most definitions, demonstrate an adequate level of verbal comprehension, which presumably can be applied to comprehending text once they have overcome word-level obstacles. Third, wordrecognition skill is necessary for text comprehension. In fact, word-reading skill correlates strongly with reading comprehension (e.g., Gough, Hoover, & Peterson, 1996). Together with listening comprehension, word-reading skill accounts for nearly all the reliable variance in reading ability (Hoover & Gough, 1990) and individual differences in word recognition explain significant variance in reading ability, even after controlling for listening comprehension (Curtis, 1980; Hoover & Gough, 1990). This is not to suggest that students with RD are inoculated from difficulties in language comprehension. Indeed, many children with RD have shown improved comprehension from direct teaching of comprehension and metaRequests for reprints should be sent to Joseph R. Jenkins, College of Education, Box 353600, University of Washington, Seattle, WA 98195-3600.

comprehension skills (Gersten, Fuchs, Williams, & Baker, 2001). Many RD researchers (e.g., Mastropieri, Leinart, & Scruggs, 1999; Meyer & Felton, 1999; Torgesen, Wagner, & Rashotte, 1997; Vellutino, Scanlon, & Sipay, 1997; Wolf, 2001) subscribe to some version of limited capacity or verbal efficiency theory (Lesgold & Perfetti, 1978; Perfetti, 1985) in which efficient word-identification processes serve as the foundation for text comprehension. According to this view, skilled readers process print with a high degree of efficiency, thereby releasing attentional resources and extending working memory capacity for comprehension work. By contrast, individuals with RD are limited by inefficient wordrecognition processes. These inefficiencies drain cognitive resources and consume working memory capacity needed for integrating and constructing meaning from text; thus, the conceptual rationale for RD researchers’ focus on word reading. SKILLED READERS VERSUS READERS WITH DISABILITIES No matter what aspect of reading is targeted (e.g., phonics, decoding, word identification, context reading, comprehension), it is a good bet that students with RD perform

238

JENKINS ET AL.: ACCURACY AND FLUENCY IN LIST AND CONTEXT READING

substantially below skilled readers, whether performance is indexed as accuracy or speed. Reading ability is usually described using standardized norm-referenced tasks (e.g., reading a list of words or nonwords that increase in difficulty, reading brief paragraphs and supplying a missing word or answering a question) that bear little resemblance to typical classroom reading tasks, where students might read aloud to a group or a peer, or might read and discuss a story (Baumann, Hoffman, Duffy-Hester, & Ro, 2000). With norm-referenced tests, performance is expressed as a percentile, standard score, or grade equivalent. Finding that fourth-grade students with RD perform 1.5 standard deviations below average or are two grade levels below the norm on the Woodcock Reading Mastery Word Test provides useful information about their reading skill. Nevertheless, such descriptions seem disconnected from the reading that classroom teachers observe (e.g., students’ accuracy or fluency on a classroom text). Indeed, many teachers and researchers would be hesitant to estimate the accuracy and fluency levels of skilled and RD readers, or the magnitude of differences between the two reader groups, on typical reading passages. Hence the main purpose of the current study: to obtain a clearer picture of the absolute and relative accuracy and fluency levels of skilled readers and those with RD when reading the type of passage encountered in classrooms. In addition, we were interested in estimating the extent to which fluency differences between skilled readers and those with RD affect the relative rates at which these students encounter (and process) idea units in text. CONTEXT-FREE AND CONTEXTUAL READING TASKS Word-reading skill can be measured in or out of context, and performance levels can be indexed as accuracy or fluency. Context-free and contextual reading tasks result in different performance levels and involve different processes. As for performance, words in context are usually read more accurately and faster than the same words out of context (Jenkins, Fuchs, van den Broek, Espin, & Deno, in press; Doehring, 1976; Nicolson, 1991; Perfetti, Finger, & Hogaboam, 1978; Stanovich, 1980). As for processes, contextual reading skill depends to a considerable degree on pure (context-free) wordrecognition skill, but is also influenced by processes that originate in context (Stanovich, 1980). Posner and Snyder (1975) described two context-based expectancy processes that facilitate word recognition in context. The two expectancy processes are independent, operate concurrently, follow different time courses, and are distinguished by the presence or absence of conscious attention in their operation. The first process consists of an automatic, fast-spreading semantic activation that operates when stimulus information (i.e., context) activates a memory location (e.g., word meaning) that automatically spreads to neighboring or related semantic memory locations, thereby privileging the retrieval of some words over others. In effect, this spreading activation lowers the threshold for perceiving the activated words, thus speeding recognition. The second expectancy process involves slow-acting, attention-demanding,

conscious use of preceding context and world knowledge for word identification. Together, these expectancy processes exert prelexical influences on contextual word recognition, resulting in superior word-reading accuracy and fluency in context. In addition, postlexical feedback from subsequently processed syntactic and semantic information sources also affect word recognition in context, helping readers to correct word-reading errors (Bowey, 1984, 1985). Thus, reading words accurately and fluently in context is not the same as reading words accurately and fluently in isolation. Contextual reading involves the orchestration of word recognition with comprehension processes that do not occur in context-free reading (i.e., receiving activation from semantic networks, enlisting conscious prediction processes to aid in reading less familiar words, reconciling postlexical feedback from syntax and sentence meaning to correct misread words). These comprehension influences on context reading are substantial; in a representative sample of Grade 4 readers, individual differences in comprehension ability were found to outweigh individual differences in pure word-recognition skill as independent predictors of contextual reading rate (Jenkins et al., in press). Further, a large body of research has established that when reading words in context, students with RD (relative to skilled readers) rely more on conscious prediction processes to compensate for weak word-identification skills. Given the contribution of comprehension processes to reading in context, along with the demonstrated comprehension difficulties that many students with RD demonstrate (Englert & Thomas, 1987), we wondered whether these students encounter particular difficulties in context reading, over and above those they experience in context-free reading tasks. If students with RD are less adept in using contextual information, or in orchestrating contextual and word-level information sources, then their contextual reading performance should be more discrepant from peers than is their word-list reading. Thus, the second question addressed in this research was whether readers with RD exhibit specific difficulties in reading in context that go beyond their difficulties in contextfree reading. SUBTYPES OF READING DISABILITIES Regardless of whether students with RD, as a group, demonstrate specific contextual reading difficulty, a subgroup of these students may encounter such reading problems. Research has identified various subtypes of RD (Berninger, Abbott, Thomson, & Raskind, 2001; Fletcher et al., 1997; Castles & Coltheart, 1993; Stanovich, Siegel, Gottardo, Chiappe, & Sidhu, 1997; Wolf & Bowers, 1999), but of interest here were RD subtypes that arise from limitations in wordidentification versus specific comprehension deficits (Cain & Oakhill, 1998; Oakhill, Cain, & Yuill, 1998). For example, Shankweiller et al. (1999) found that although decoding and comprehension skills were concordant for most students with RD, two other subtypes could be detected. Of these, the larger group demonstrated better comprehension than decoding; the smaller subtype, better decoding than comprehension. Such individual differences in context-free wordidentification skill and comprehension ability among students

LEARNING DISABILITIES RESEARCH

with RD may differentially affect performance in contextfree and contextual reading tasks. For example, some RD students may exhibit impairments in reading isolated words but not in context reading because their language skills allow them to compensate for weak word-identification skills. Alternatively, some students with RD may exhibit impaired contextual reading because their poorly developed language comprehension skills undermine automatic prediction processes and/or disrupt conscious use of contextual sources of information (e.g., using prior context to figure out unfamiliar words, exploiting feedback from semantic and syntactic text features). Either of these situations will result in RD subtypes, with impairments detected according to the nature of the reading task (i.e., reading words in or out of context). Thus, a third purpose of this research was to investigate the possibility of RD subtypes as revealed by impaired reading on different formats (context and list). SUMMARY OF STUDY’S PURPOSE This research sought answers to three questions regarding students with RD: What are the performance levels and group differences between students with RD and skilled readers when reading a typical classroom text? Do students with RD encounter problems in context reading that cannot be traced to weak word-identification skills? And, are subtypes of RD differentiated according to performance on contextual and context-free reading tasks? Results should extend our understanding of word-level reading skills of students with RD. At the same time, findings should help both reading researchers and practitioners to identify appropriate reading measurement tasks for describing student competence, for tracking the development of reading skill over time, and for identifying disability status. METHOD Participants Students were drawn from six schools in one school district in the United States From a sample of 109 fourth graders without any disability, we identified 85 skilled readers, defined as scoring at or above the 50th percentile on the reading comprehension test of the Iowa Test of Basic Skills (ITBS). We also identified a group of 24 fourth graders who met the following criteria for RD. All had been classified by their schools as having a learning disability in reading, demonstrated a discrepancy of at least 1.5 standard deviations between IQ and reading achievement (ITBS), and scored at or below the 25th percentile on the ITBS. Our criterion of a second indication of IQ-achievement discrepancy, based on ITBS and an IQ score, constitutes an additional check on the schools’ classification decision and a more stringent criterion for RD status. For skilled readers and students with RD, we show descriptive and inferential statistics in Table 1 on age, sex, race, subsidized lunch status, absenteeism, classroom behavior status, and ITBS normal curve equivalents. Groups differed on each variable except absenteeism history. Table 1 also gives full-scale WISC-R scores for the RD group.

239

Measures The measures used in this study focused on students’ contextfree and context reading performance. Contextual and Context-Free Reading Following an assessment approach known as curriculumbased measurement (Deno, 1985), we had students read aloud for one minute and counted the number of words read correctly and incorrectly. The reliability and validity of this simple measurement procedure has been well established in more than 100 studies conducted by multiple investigators (e.g., L. S. Fuchs, 1995; Jenkins & Jewell, 1993; Marston, 1989). To administer the assessment, we used the following directions: “I want you to read the words on these pages to me. Try to read every word. Do your best. When I say, ‘Begin,’ read the words out loud. You’ll have 1 minute to read as many words as you can. If you wait too long to say a word, I’ll tell you the word. Then, keep reading. You can skip words you don’t know. If you come to the end of the page, turn to the next page. At the end of 1 minute, I’ll say, ‘Stop.’ Do you have any questions?” Errors were omissions, insertions, mispronunciations, substitutions, and hesitations of more than three seconds. Self-corrections were not errors. If needed, testers reminded students to turn pages. Students read from a 421-word traditional folktale (“The Father, His Son, and Their Donkey”) used in research by Brown and Smiley (1977) and L. S. Fuchs, Fuchs, and Hamlett (1989). Jenkins, Heliotis, Haynes, Stein, and Beck (1986) rewrote the folktale to approximate a third-grade readability level (Fry, 1968) while preserving the gist of the stories. This folktale is a simple, engaging narrative that resembles the kind of “authentic literature” popular in many classrooms. The tale traces the journey and experiences of a father, son, and donkey as they encounter various groups of people, with each group expressing strongly differing opinions about who should be riding the donkey. We formatted this folktale in three ways. First, the folktale was presented in its natural form; we refer to this format as context. Second, we randomly ordered the words and presented them in a list; we refer to this format as list. Third, we randomly reordered the words and presented them in paragraphs without punctuation; we refer to this format as random. For each of these three formats, we calculated two scores: rate (words read correctly) and accuracy (words read correctly divided by total words read). Interscorer agreement, calculated on 15 percent of the protocols used in this report, exceeded 99 percent for each score within each format. (Throughout this study, interscorer agreement was calculated as number of agreements divided by agreements plus disagreements.) To examine the rate at which skilled readers and those with RD encounter idea units in text as a function of their reading rates, we estimated the number of idea units in the folktale, using procedures described by Brown and Smiley (1977) and Johnson (1970) for identifying pausal units within a passage. To accomplish this task, each of the five authors

240


TABLE 1 Demographic Information by Reading Status Reading Status Skilled (n = 85) Variable

M

(SD)

Age Male Race African American European American Asian American Subsidized lunch 2+ absences/month Class behavior Acceptable Occasional problem Frequent problem ITBS: Comprehension Full-scale IQ

9.79

(1.03)

67.56

n

RD (n = 24) (%)

M

(SD)

10.34

(1.55)

Skilled vs. RD n

(%)

F(1, 107) 11.44∗∗

χ2

43

(51)

19

(79)

6.23∗

12 66 7 7 5

(14) (78) (8) (8) (6)

11 13 0 14 3

(46) (54) (0) (58) (13)

12.32∗∗

60 24 1

(70) (28) (1)

13 7 4

(54) (29) (17)

10.55

(14.35)

14.42 89.00

(11.43) (8.14)

30.20∗∗∗ 1.21

491.81∗∗∗

Note. ITBS = Iowa Test of Basic Skills normal curve equivalents; RD = reading disabilities. ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.

independently read the passage and divided the text into individual units by placing a vertical line at a division point. An individual unit was defined as one “which contained an idea and/or represented a pausal unit” (Brown & Smiley, 1977, p. 3). Next, by dividing the number of words in the folktale (421) by the number of idea units identified by each author, we calculated a mean number of words per idea unit for the passage (6.36). To estimate the number of idea units that the fourth-grade readers encountered per minute of reading, we divided each student’s rate of correctly read words in context by 6.36 (the mean number of words per idea unit).

Procedure Each student read each format for one minute; the order in which students read the formats was counterbalanced. One of four research assistants, trained in administration procedures and experienced in administering reading tests to fourth graders, individually collected the three one-minute reading samples in one session. Prior to the individual testing, research assistants administered the ITBS in large groups, using standardized procedures. RESULTS

ITBS Students completed Form K, Level 10 of the reading comprehension test of the ITBS (Riverside, 1994). Kuder-Richardson 20 reliability was between 0.87 and 0.88 (Riverside, 1994). Scores were normal curve equivalents (NCE). Agreement on 15 percent of protocols was 99.1 percent. Comparison Group To examine the sensitivity of list and context tasks scored for accuracy and rate in detecting reading impairment, we compared the RD group to a representative sample of 113 Grade 4 students who were administered the same measures (Authors, 2000). Means of the representative sample were 9.61 (SD = 1.03) for age, and 51.73 (SD = 23.94) for ITBS NCE. Their means and standard deviations for the context and list tasks follow: for rate, context mean was 127.03 (SD = 51.62) and list mean was 82.86 (SD = 25.96); for accuracy, context mean was 0.96 (SD = 0.05) and list mean was 0.95 (SD = 0.06).

A preliminary analysis showed that presenting the words in random strings in paragraph format had an inhibitory effect on word reading (Authors, 2000). That is, words in the random (i.e., paragraph) format were read significantly more slowly than in list format. Consequently, we dropped the random format and relied in subsequent analyses on the list format to represent context-free reading. Performance Levels of Skilled Readers and Those with RD Means and standard deviations by task format, score, and reader type are shown in Table 2. Not surprisingly, skilled readers significantly outperformed those with RD on both list and context formats, whether performance was described as rate or accuracy. To obtain the mean rate at which reader groups encountered/processed idea, we divided the groups’ mean rate of correct reading by 6.36 (the mean number of words per idea unit for the passage). Because of the large difference in reading rates, the number of idea units processed per minute was

LEARNING DISABILITIES RESEARCH TABLE 2 Rate and Accuracy Scores by Reading Status and by Format Reading Status Skilled (n = 85) Variable Rate Context List Accuracy Context List

With RD (n = 24)

Skilled vs. RD

M

(SD)

M

(SD)

F (1,107)

ES a

154.88 92.82

(37.99) (18.8)

51.54 43.42

(18.80) (14.51)

165.64∗∗∗ 141.52∗∗∗

2.72 2.62

0.98 0.97

(0.02) (0.03)

0.86 0.83

(0.10) (0.09)

107.59∗∗∗ 141.34∗∗∗

6.00 4.67

a Based ∗∗∗ p

on standard deviation of skilled readers. < 0.001.

quite different for the two groups. Skilled readers encountered an average of 24.35 idea units per minute, compared to only 8.1 idea units per minute for students with reading disabilities. Figure 1 depicts the magnitude of the performance differences for the reader groups on rate and accuracy in context and for the number of idea units encountered per minute. Comparing Discrepancies Between Context and List Performance for Reader Groups We used skilled readers as the reference group and compared the ES for reader groups (i.e., RD vs. skilled) on the list and context reading tasks. To compute ES, we considered the skilled reader and RD groups as representing different populations because their performance levels and standard deviations were substantially different. Therefore, instead of using pooled standard deviations, we used the standard deviation of the “control” (i.e., the skilled reader) group. When comparing skilled and RD readers, we subtracted the means, then divided the difference by the standard deviation of the skilled readers. As shown in Table 2,

120

50

25

0

Skilled

RD

the ES for reader group in list rate (2.62) was virtually identical to that found on context rate (2.72). The similar ES indicated that students with RD were not more discrepant from skilled readers on rate measures in context than in a list. With regard to the accuracy measures, context and list were equally affected by a ceiling in the scores. However, despite the potential effects of the ceiling on the range of scores, differences between skilled and RD students still appear, and these differences are not greatly affected by the format of the measures; that is, large differences are seen between skilled and RD readers regardless of whether words are presented in context or list. Further, the effect sizes are similar for context and list. Because the accuracy measures were affected by a ceiling in the scores and the rate measures were not, we will not consider differences in effect sizes across rate and accuracy. Impairment Analysis Whereas the previous analysis focused on the average reading rates of the RD group and used skilled readers as the reference group, the next analysis examined individual performance of students with RD and used a representative sample of Grade 4 peers as the reference group. Specifically, we examined (1) the numbers of skilled and RD students identified as impaired according to their performance on the list and context tasks, and (2) whether subtypes of RD were detected on these tasks. Following Shankweiler et al. (1999), we used the following set of methods. Employing the means and standard deviations of the representative sample of 113 Grade 4 students who had read the word list and text, we computed an accuracy and rate z-score for each student with RD within each format (list and context). Students with RD with z-scores at or above –0.67 were classified as not impaired; students with RD with z-scores at or below –1.00 were classified as impaired. Students with RD with z-scores between –0.67 and –1.00 were in a buffer zone and therefore were omitted from

30 Idea Units Encountered Per Minute

75

Words Read Per Minute

160

Percent Accuracy

100

241

80

40

25

20

15

10

5

0

0 Skilled

RD

Skilled

RD

FIGURE 1 Mean percent of correctly read words, words read correctly per minute, and idea units encountered per minute in context for skilled readers and those with RD.

242


TABLE 3 Frequencies of Students with LD (n = 24) Classified as Impaired or Not Impaired as a Function of Format and Score Context: Impaired (1) Rate

List: Impaired—

No Yes

No 0 0

No Yes

No 0 4

Yes 0 21

Context: Impaired (2) Accuracy

List: Impaired—

Yes 0 11

analysis. Shankweiller et al. argued that inserting a buffer zone rather than using simple cut scores better divides impaired from nonimpaired readers, because simple cut scores assign individuals whose scores are just above or just below the cutting score into impaired and nonimpaired groups, even though their performance is quite similar. In contrast: “Placing a buffer between the normal and extreme scores ensures that any participant who counts as extreme does not just miss being normal by a hair, and vice versa. The buffer method mitigates some of the arbitrariness of a cutting score” (1999, p. 75). Using these results, we created two contingency tables (see Table 3) showing impairment status for (1) context versus list formats using rate as the score and (2) context versus list formats using accuracy as the score. As Table 3 shows, different combinations of tasks and score types were differentially sensitive to detecting impairment. When rate was used to index performance, 21 of 24 students with RD were classified as impaired on both list and context tasks, with no evidence for RD subtypes based on differential impairments. When performance was described in terms of accuracy, 11 of 24 exhibited impairments in both context and list, with four others falling into the impaired range for list, but not context accuracy. However, it is noteworthy that no student with RD qualified as “not impaired.” Whether the task was reading lists or words in context, or performance was scored for rate or accuracy, every one of these students either demonstrated an impairment or fell into the buffer zone between impaired and not impaired (i.e., the no-no cell was empty in both tables). Concerning our interest in specific contextual reading impairment, no students with RD showed an impairment for context reading that was not present in list reading.

DISCUSSION Differences Between Skilled and RD Performance Skilled readers outperformed students with RD on list and context tasks, whether scored for accuracy or rate. Differences were statistically significant with large effect sizes (ranging from 2.62 to 6.00). Performance differences between skilled readers and those with RD are not surprising, but the magnitude of the differences in list and context, whether

scored for accuracy or rate, are noteworthy. As a group, skilled readers read words three times faster in context and two times faster in lists. Performance differences between reader groups probably vary depending on the level of passage difficulty, and our results are based on but one reading passage of a particular level of difficulty. Nevertheless, this passage fell at a Grade 4 instructional level according to the reading accuracy (96 percent) observed for the representative sample (Betts, 1946). Thus, the observed fluency levels of 150 (for skilled readers) and 50 (for RD readers) words per minute may provide a reasonable estimate of performance for the two groups when reading grade-level texts. Reader group differences in absolute accuracy levels were not as pronounced as the differences in rate, but ES for reader group were larger for accuracy than for rate. The differences in ES were influenced by the fact that many skilled readers performed at a ceiling level in accuracy, resulting in restricted variance and a smaller denominator in the calculation of ES. On average, the skilled group performed at an independent level (i.e., 98 percent accuracy, Betts, 1946), whereas students with RD performed at a frustration level (86 percent accuracy), and the representative sample performed at an instructional level (96 percent accuracy). To the extent that accuracy level is a valid indicator of the degree to which a passage is appropriate for use in reading instruction, our results indicate that typically achieving readers, skilled readers, and those with RD cannot derive equitable learning opportunities using the same texts. Rather, texts that are instructionally appropriate for one reader group will be inappropriate for another group. This suggests that inclusive classrooms that employ a whole-class reading group may fail to provide equal learning opportunities for students of different reading abilities, unless teachers differentially scaffold instruction (e.g., choral reading, partner reading, tutoring) for students with limited word-identification skills. Reading Rates and Comprehension Processing The potential impact of individual differences in reading fluency on the processing of text ideas can be concretely illustrated by considering the number of idea units that skilled readers and those with RD encountered per minute (Figure 1). It is easy to imagine how such a sizable difference in the temporal contiguity of text ideas (24 vs. 8 per minute) might differentially affect comprehension of skilled and RD readers. Verbal efficiency theory postulates a direct relation between word-processing speed and comprehension, such that dysfluent reading interferes with comprehension processes. According to Perfetti, inefficient decoding can undermine comprehension in two ways. First, effortful decoding can deactivate memory for recently established contexts (i.e., “disrupt the temporary representation of text in working memory” (1985, p. 114)). Second, inefficient lexical access produces a low-quality code in memory (i.e., a code in which activation of semantic or phonological information is not immediate). Both mechanisms (disruption of memory for prior context and poor memory codes) interfere with propositional encoding. Rupley, Willson, and Nichols described this phenomenon: “Slower rates of word recognition would directly

LEARNING DISABILITIES RESEARCH

affect comprehension and inhibit chunking of information into meaningful information units. Thus, both comprehension of new information and expansion and elaboration of existing knowledge would be affected by children’s rate and accuracy in processing information” (1998, p. 155). Practically speaking, slow, effortful reading means that a task skilled readers can accomplish in a reasonable amount of time will be accomplished by students with RD in an unreasonable amount of time, unless teachers find ways to expedite RD students’ performance. Without effective scaffolding, slow, effortful reading is likely, over the long run, to undermine literacy motivation and reduce the chances that students with RD will choose reading over other competing activities. Problems Specific to Contextual Reading We found no evidence that students with RD, as a group, experience problems specific to reading in context. In one analysis we compared RD and skilled readers’ performance in context and list. The RD group was as discrepant from skilled readers in list reading as they were in contextual reading. In another analysis we employed Shankweiller et al.’s (1999) criteria for distinguishing impaired from nonimpaired readers. We asked two questions of this sensitivity analysis. Relating to our question about problems specific to context reading, we asked if more students with RD would manifest wordreading difficulty in context than in list reading. Using rate scores, we found an equal percentage (88) of students classified as impaired on list and context tasks, with the remaining students unclassified. This perfect congruence of impairment classification for list and context rate is consistent with findings from the discrepancy analysis (i.e., ES comparison in Table 2). That is, students with RD did not demonstrate a unique problem in coordinating the cognitive demands of reading in context, relative to their problem in reading single words. These findings suggests that the RD students’ slow correct rates in context can be traced to their weakness in context-free reading, rather than to a difficulty in orchestrating comprehension and word-identification skills or a failure to take advantage of contextual information sources. This interpretation is bolstered by the analysis of accuracy impairment, where we found that any student qualifying as accuracy impaired in context reading also qualified as accuracy impaired in list reading. When RD students were sorted into impaired and unimpaired categories on the basis of their word-reading accuracy, 63 percent were classified as impaired. About a third of those demonstrated impairment in list reading but not in context reading. Apparently, some students with RD derived enough facilitation from context to escape the impairment classification for reading accuracy. However, even these students’ context accuracy was not sufficient to qualify as nonimpaired (i.e., they fell into the buffer zone). No students who demonstrated impairment in reading words accurately in context missed being classified as impaired in list-reading accuracy. Again this suggests that limitations in context-free word identification is an important source of contextual reading difficulty for these students.

243

The other question addressed by the impairment analysis was whether different subtypes of RD were identified by the different combination of reading tasks (list and context). No evidence for subtypes was observed in reading rates for isolated word and context-reading tasks. Students with RD demonstrated perfect concordance for impairments in list and context-reading rates. However, when performance was indexed as accuracy, there was a hint of subtypes. We observed a trend for some RD student to manifest impairment in list accuracy but not context accuracy. However, the subgroup of students who were classified as accuracy impaired in list but not in context reading fell into the buffer zone on context accuracy, qualifying neither as impaired nor unimpaired in context accuracy. Additional research, with larger samples, should continue to explore the issue of subtypes for reading accuracy. In this research, we employed a definition of RD consistent with the research literature, that is, all 24 students demonstrated a discrepancy between IQ and reading achievement of at least 1.5 standard deviations, and their reading achievement was no higher than the 25th percentile. On some task and performance dimensions, a large proportion of students with RD did not demonstrate impairment (e.g., context accuracy, where fewer than half the RD group qualified as impaired). At the same time, no students with RD received a clean bill of health. That is, every student with RD either demonstrated an impairment or fell into the buffer zone on at least one performance dimension (rate or accuracy) of each reading task (list or context). Finally, the rate measures were more sensitive in detecting reading impairment, where almost 90 percent of students met Shankweiler et al.’s (1999) impairment criterion, and this occurred whether students were assessed in decontextualized or contextualized formats. The finding that rate served to discriminate RD better than accuracy is consistent with a recent meta-analysis (D. Fuchs, Fuchs, Mathes, & Lipsey, 2000) in which effect sizes comparing students with RD versus low performing nondisabled readers were significantly larger on timed than on untimed measurements. Study Limitations It is important to consider our results in light of four critical study limitations. First, the participants were fourth graders, and results may vary for younger or older students. Second, as is often the case in school-based samples, the RD group differed from the skilled reader group in ethnicity and SES, factors that limit the generalizability of findings to more purely defined reader groups. Third, data collection in this investigation occurred under timed conditions, a factor that can affect levels of reading accuracy (Bowey, 1984). Fourth, had students read the entire word list and entire passage, the overlap between words on the two tasks would have been 100 percent. Instead, we adopted curriculum-based measurement procedures (Deno, 1985), in which students read each task for one minute. As a result, word overlap between list and context was only partial (51 percent). Future study of context and list reading without time restrictions would provide a stronger control for partitioning the unique effects of list and context

244


reading. For these reasons, we temper the strength of our conclusions accordingly and recommend that future studies examine different instructions and procedures in comparing context and context-free reading performance. Conclusions and Implications Our results indicate that skilled readers’ fluency in reading a typical grade-level passage exceeds the fluency of students with RD by approximately a factor of three. This difference in reading fluency results in very different rates at which these groups encounter and process idea units in text and may account for comprehension differences observed between these groups. The large accuracy and fluency differences between reader groups imply that texts appropriate for instructing one group will be different from texts appropriate for other groups, an implication that bears on grouping practices employed in inclusive classrooms. Specifically, it suggests that classrooms employing whole-group reading texts are suboptimal learning environments for groups whose reading skills are as heterogeneous as those of students in our study. We also found no evidence of context-specific reading problems for students with RD, either as a group or individually. This suggests that the accuracy and fluency difficulties students with RD demonstrate in contextual reading can be traced to their weakness in context-free word recognition. We take this to mean that if students with RD are to make gains in contextual reading accuracy and fluency, their teachers must find ways to strengthen students’ context-free reading skills as well as language comprehension skills. Students with RD suffer serious reading impairment and warrant intensive intervention. Results of this study suggest that rate measures function better than accuracy in detecting RD. Regarding assessment practices, we recommend that diagnosticians and teachers incorporate the measurement of fluency (words correct per minute) in considering the severity of students’ reading difficulties and in monitoring growth in reading proficiency. As Compton and Carlisle (1994) point out, fluency is frequently ignored in the diagnosis of reading disability. ACKNOWLEDGMENTS This research was supported in part by Grant #H023F70010 from the U.S. Department of Education, Office of Special Education Programs, and by Core Grant #H15052 from the National Institute on Child Health and Human Development to Vanderbilt University. Statements do not reflect official policy of any agency. We also wish to acknowledge the Netherlands Institute for Advanced Study in the Humanities and Social Sciences for its support in the preparation of this manuscript. We thank Joanne Carlisle for her comments on an earlier draft of this article. REFERENCES Baumann, J. F., Hoffman, J. V., Duffy-Hester, A. M., & Ro, A. M. (2000). The first R yesterday and today: U.S. elementary reading instruction practices reported by teachers and administrators. Reading Research

Quarterly, 35, 338–377. Betts, E. (1946). Foundations of reading instruction. New York: American Book Co. Berninger, V. W., Abbott, R. D., Thomson, J. B., & Raskind, W. H. (2001). Language phenotype for reading and writing disability: A family approach. Scientific Studies of Reading, 5, 59–106. Bowey, J. A. (1984). The interaction of strategy and context in children’s oral reading performance. Journal of Psycholinguistic Research, 13, 99–117. Bowey, J. A. (1985). Contextual facilitation in children’s oral reading in relation to grade and decoding skill. Journal of Experimental Child Psychology, 40, 23–48. Brown, A. L., & Smiley, S. S. (1977). Rating the importance of structural units of prose passages: A problem of meta-cognitive development. Child Development, 48, 1–8. Cain, K., & Oakhill, J. (1998). Comprehension skill and inference making ability: Issues in causality. In C. Hulme & M. Joshi (Eds.), Reading and spelling development and disorders. Mawah, NJ: Erlbaum. Castles, A., & Coltheart, M. (1993). Varieties of developmental dyslexia. Cognition, 47, 149–180. Compton, D. L., & Carlisle, J. F. (1994). Speed of word recognition as a distinguishing characteristic of reading disabilities. Educational Psychology Review, 6, 115–139. Curtis, M. (1980). Development of components of reading skill. Journal of Educational Psychology, 72, 656–669. Deno, S. L. (1985). Curriculum-based measurement: The emerging alternative. Exceptional Children, 52, 219–232. Doehring, D. G. (1976). Acquisition of rapid reading responses. Monographs of the Society for Research in Child Development, 41 (2, Serial No. 165). Englert, C. S., & Thomas, C. C. (1987). Sensitivity to text structure in reading and writing: A comparison between learning disabled and non-learning disabled students. Learning Disability Quarterly, 10, 93–105. Ehri, L., & Saltmarsh, J. (1995). Beginning readers outperform older disabled readers in learning to read words by sight. Reading and Writing: An Interdisciplinary Journal, 7, 295–326. Fletcher, J., Morris, R., Lyon, G. R., Stuebing, K. K., Shaywitz, S. E., Shankweiller, D. P. et al. (1997). Subtypes of dyslexia: An old problem revisited. In B. Blachman (Ed.), Foundations of reading acquisition and dyslexia (pp. 67–92). Mahwah, NJ: Erlbaum. Fry, E. (1968). A readability formula that saves time. Journal of Reading, 11, 513–516. Fry, E. (1980). The new instant word list. Reading Teacher, 34, 284–289. Fuchs, D., Fuchs, L. S., Mathes, P., & Lipsey, M. (2000). Reading differences between low-achieving students with and without learning disabilities: A meta-analysis. In R. Gersten & S. Vaughn (Eds.), Contemporary special education research: Syntheses of the knowledge base on critical instructional issues (pp. 81–104). Hillsdale, NJ: Erlbaum. Fuchs, L. S. (1995). Curriculum-based measurement and eligibility decision making: An emphasis on treatment validity and growth. Paper presented at the National Research Council Workshop on Alternatives to IQ Testing. Washington, DC. Fuchs, L. S., Fuchs, D., & Hamlett, C. L. (1989). Monitoring reading growth using student recalls: Effects of two teacher feedback systems. Journal of Educational Research, 83, 103–111. Gersten, R., Fuchs, L. S., Williams, J. P., & Baker, S. (2001). Teaching reading comprehension strategies to students with learning disabilities: A review of research. Review of Educational-Research, 71, 279– 320. Gough, P. B., Hoover, W., & Peterson, C. L. (1996). Some observations on the simple view of reading. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties (pp. 1–13). Mahwah, NJ: Erlbaum. Hoover, W. A., & Gough, P. B. (1990). The simple view of reading. Reading and Writing: An Interdisciplinary Journal, 2, 127–160. Jenkins, J. R., Fuchs, L. S., van den Broek, P., Espin, C., & Deno, S. L. (in press). Sources of individual differences in reading comprehension and reading fluency. Journal of Educational Psychology. Jenkins, J. R., Heliotis, J., Haynes, M., Stein, M., & Beck, K. (1986). Does “passive learning” account for disabled readers’ comprehension deficits in ordinary reading situations? Learning Disability Quarterly, 9(1), 69– 76. Jenkins, J. R., & Jewell, M. (1993). Examining the validity of two measures for formative teaching: Reading aloud and maze. Exceptional Children, 59, 421–432.

LEARNING DISABILITIES RESEARCH Johnson, R. E. (1970). Recall of prose as a function of the structural importance of the linguistic units. Journal of Verbal Learning and Verbal Behavior, 9, 12–20. Lesgold, A. M., & Perfetti, C. A. (1978). Interactive processes in reading comprehension. Discourse Processes, 1, 323–336. Marston, D. (1989). A curriculum-based measurement approach to assessing academic performance: What it is and why do it. In M. R. Shinn (Ed.), Curriculum-based measurement: Assessing special children (pp. 18– 78). New York: Guilford. Morrison, F. J. (1987). The nature of reading disability: Toward an integrative framework. In S. Ceci (Ed.), Handbook of cognitive, social, and neuropsychological aspects of learning disabilities (pp. 33–62). Hillsdale, NJ: Erlbaum Associates. Mastropieri, M. A., Leinart, A., & Scruggs, T. E. (1999). Strategies to increase reading fluency, Intervention in School and Clinic, 5, 278–283, 292. Meyer, M. S. & Felton, R. H. (1999). Repeated reading to enhance fluency: Old approaches and new directions. Annals of Dyslexia, 49, 283–306. Nicholson, T. (1991). Do children read words better in context or in lists? A classic study revisited. Journal of Educational Psychology, 83, 444– 450. Oakhill, J., Cain, K., &Yuill, N. (1998). Individual differences in children’s comprehension skill: Toward an integrative model. In C. Hulme & M. Joshi (Eds.), Reading and spelling development and disorders. Mawah, NJ.: Erlbaum. Perfetti, C. A. (1985). Reading ability. New York: Oxford University Press. Perfetti, C. A., Finger, E., & Hogaboam, T. W. (1978). Sources of vocalization latency differences between skilled and less skilled young readers. Journal of Educational Psychology, 70, 730–739. Posner, M. I., & Snyder, C. R. R. (1975). Attention and cognitive control. In R. Solso (Ed.), Information processing and cognition: The Loyola symposium (pp. 55–85). Hillsdale, NJ: Erlbaum.

245

Riverside. (1994). Riverside 2000: Integrated assessment program technical summary 1. Chicago, IL: Author. Rupley, W. H., Willson, V. L., & Nichols, W. D. (1998). Exploration of the developmental components contributing to elementary school children’s reading comprehension. Scientific Studies of Reading, 2, 143– 158. Shankweiler, D., Lundquist, E., Katz, L., Stuebing, K. K., Fletcher, J. M., Brady, S. et al. (1999). Comprehension and decoding: Patterns of association in children with reading disabilities. Scientific Studies in Reading, 3, 69–94. Stanovich, K. E. (1980). Toward an interactive-compensatory model of individual differences in the development of reading fluency. Reading Research Quarterly, 16, 32–71. Stanovich, K. E. (1988). The right and wrong places to look for the cognitive locus of reading disability. Annals of Dyslexia, 38, 154–177. Stanovich, K. E., Seigel, L. S., Gottardo, A., Chiappe, P., & Sidhu, R. (1997). Subtypes of developmental dyslexia: Differences in phonological and orthographic coding. In B. Blachman (Ed.), Foundations of reading acquisition and dyslexia (pp. 67–92). Mahwah, NJ: Erlbaum. Torgesen, J. K., Wagner, R. K., & Rashotte, C. A. (1997). Prevention and remediation of severe reading disabilities: Keeping the end in mind. Scientific Studies in Reading, 1, 217–234. Vellutino, F., Scanlon, D. M., & Sipay, E. R. (1997). Toward distinguishing between cognitive and experiential deficits as primary sources of difficulty in learning to read: The importance of early intervention in diagnosing specific reading disability. In B. Blachman (Ed.), Foundations of reading acquisition and dyslexia (pp. 347–379). Mahwah, NJ: Erlbaum. Wolf, M. (2001). Dyslexia, fluency, and the brain. Timonium, MD: York Press. Wolf, M., & Bowers, P. (1999). The double-deficit hypothesis for the developmental dyslexias. Journal of Educational Psychology, 91, 415–438.

About the Authors Joseph R. Jenkins is Professor of Special Education at the University of Washington. His research focus is reading acquisition and assessment and instruction of students with learning disabilities. He received his Ph.D. from the University of Minnesota. Lynn S. Fuchs is Professor of Special Education and Co-Director of the Reading Clinic in the Kennedy Center at Vanderbilt University. Her research focuses on classroom-based assessment and instruction for students with reading and math disabilities. She earned her Ph.D. from the University of Minnesota. Paul van den Broek is Professor in the Department of Educational Psychology at the University of Minnesota, where he is Guy Bond Chair in Reading and Literacy. His research focuses on cognitive processes involved in success and failure in reading comprehension and on the development of comprehension skills. He holds doctoral degrees in developmental psychology and in experimental psychology from the University of Leiden, the Netherlands, and in educational psychology from the University of Chicago. Christene Espin is a Professor of Educational Psychology and Special Education. Her research focuses on basic reading and written expression skills of students with learning disabilities at the secondary-school level. In particular, she is seeking to design a measurement system that can be used to track the performance of students in reading and written expression over time. She received her Ph.D. from the University of Minnesota. Stanley L. Deno is Professor of Educational Psychology and Special Education. His focus over the past 30 years has been on students at risk for academic failure—especially those in learning disabilities programs. His research and development on monitoring student progress in basic skills has resulted in the development of procedures typically referred to as curriculum-based measurement (CBM). He received his Ph.D. from the University of Minnesota.