• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Layers of Reading Intervention in Kindergarten Through Third Grade: Changes in Teaching and Student Outcomes

Posted on: Friday, 30 September 2005, 06:01 CDT

By O'Connor, Rollanda E; Fulmer, Deborah; Harty, Kristin R; Bell, Kathryn M

Abstract

In this study, students and their teachers participated in a layered approach to reading intervention in kindergarten through third grade that included professional development for teachers in scientifically based reading instruction, ongoing measurement of reading progress, and additional small-group or individual instruction for students whose progress was insufficient to maintain grade-level reading achievement. Reading outcomes were compared with historical control groups of students in the same schools. The findings revealed overall improvements in reading, improved reading for students who began the study in high-risk categories, and decreases in the incidence of reading disability at the end of third grade. Implications for scaling up are discussed.

Early intervention in reading has garnered national attention through the Reading First Initiative (RFI) of the No Child Left Behind Act (NCLB; 2002), which sets goals for improving reading performance in schools where students have shown limited success in the acquisition of basic literacy by the end of third grade. These goals stem from theories in which skilled reading can be traced back in time (Ehri & McCormick, 1998; Gough & Walsh, 1991; Liberman & Shankweiler, 1985; Share & Stanovich, 1995; Spear-Swerling & Sternberg, 1994) to earlier understandings about print and its relationship to spoken words (i.e., phonemic awareness, or the ability to apprehend and manipulate the sounds in spoken words), about the development of automaticity (i.e., that early skills, such as recognizing words in print, should be speeded up or made automatic), and about the ways in which comprehension of spoken language can be applied to comprehension of written language (e.g., meanings of words, syntax, application of background knowledge). This notion of reading development opens the possibility that reading problems could be diminished if individual differences in skill development were addressed at the time they are first noticed, rather than waiting for these differences to become pronounced.

In essence, the rationale behind most early intervention efforts is to encourage at-risk children to learn the earlier understandings about sounds in words and features of print within the windows in time that these understandings develop for typically achieving children. In this view, if by the end of kindergarten all children could blend and segment spoken words (phonemic awareness) and link these speech sounds to letter sounds (the alphabetic principle), they would be prepared to learn to decode and recognize words in first grade. Likewise, if children learned to decode and recognize words in first grade, alongside maintaining adequate vocabulary growth, they would be prepared for building reading fluency and more complex comprehension in later grades.

Based on this theory of reading development and intervention, teams of reading experts around the country have developed guidelines for the content of effective reading instruction in Grades K-3, for the professional development of teachers to learn this content, and for monitoring the progress and outcomes of students over time. These guidelines appear relatively straightforward. Teachers are to implement scientifically based reading instruction and monitor the progress that students make toward reading competence; that is, instruction should address the components of reading that have been linked to reading improvement in experimental studies. These components include phonemic awareness (the ability to isolate and manipulate sounds in spoken words), phonics (the linkage of speech sounds to alphabet letters and letter combinations), fluency (the rate of reading running text), vocabulary (the meanings of words), and comprehension (understanding sentences and the overall meaning of passages), in amounts that differ depending on whether students have achieved grade- appropriate expectations in these skills.

Although support is widespread, the effects of long-term early intervention on the poorest readers are not well understood. The National Reading Panel (2000) report found few studies extending beyond one school year, despite the tenacity of many reading difficulties. Most studies of schoolwide early intervention in reading have focused on kindergarten and first grade. In these studies, researchers have addressed instruction and measurement particularly in phonemic awareness and phonics (Bus & IJzendoorn, 1999). Intervention research with children at risk for reading disability (RD) has reported positive effects for children in kindergarten (Blachman, Ball, Black, & Tangel, 1994; O'Connor, Jenkins, & Slocum, 1995; O'Connor, Notari-Syverson, & Vadasy, 1996) and first grade (Blachman, Tangel, Ball, Black, & McGraw, 1999; Clay, 1985; Foorman, Francis, Fletcher, Schatschneider, & Mehta, 1998; Vellutino et al., 1996); however, few studies have followed students into second grade or beyond, as reading becomes more complicated. Our goal for the current study was to address reading outcomes across more years of reading development (i.e., K-3) and across more dimensions of reading (e.g., fluency and comprehension) than those in most studies of early intervention.

The type of longitudinal research required to document the effects of long-term early intervention engenders unique problems. For example, appropriate control groups for longitudinal studies have been difficult to establish, because teachers work together in schools. Thus, it is rarely possible to randomly assign students to classes within schools without contaminating the control classrooms over time. Moreover, children experience new teachers every year in most schools. To document progress within schools, students would need to show change over their previous performance and over the performance of other students in their schools in previous years. Studying the potential effects of long-term intervention, then, suggests the use of historical control groups, in which data are collected on reading performance prior to and during the implementation of new instructional practices over a long period of time. It would also be useful to know whether intervention affects some groups of children differentially, such as children from low socioeconomic status (SES) households, children with low levels of early reading skills, or children eligible for special education.

Personnel in two schools with differing socioeconomic characteristics worked with research personnel to learn how to implement scientifically based reading instruction and to use measures of the progress of their students to modify instruction. In keeping with the notion of helping children to attain grade-level expectations on key early reading skills, the dimensions of the study included ongoing professional development for teachers, measurement of students' reading progress from kindergarten through the end of third grade, and additional intervention for students who failed to show grade-level progress. Each of these features is described hereafter.

Professional Development

Researchers who study effective professional development have suggested that sessions should include modeling intended behaviors, discussing implementation issues, planning for adopting new strategies, and providing ongoing feedback on observed instructional changes (Darling-Hammond, 2000; McCutchen et al., 2002). More lasting effects may be achieved by linking professional development activities to current research, to concrete tasks, and to problems of practice (Gersten, Chard, & Baker, 2000). Moreover, Donohoe (1994) suggested that successful efforts may require an outside force working in partnership with schools to prompt, guide, and structure effective change in practice. In their study of sustaining improved practice over time, Gersten et al. (2000) found that "practices that helped teachers succeed with low-achieving students promoted the continued use of those practices" (p. 446). This advice is particularly pertinent to efforts aimed at reducing the incidence or severity of RD.

The research team in this study provided professional development (PD) in two areas. The first area-scientifically based reading instructionconsisted of component-centered PD for staff across both schools, in which teachers learned about findings on effective reading instruction from the National Research Council (Snow, Burns, & Griffin, 1998) and the National Reading Panel (2000). The focus of the content of PD shifted across research years, as an additional grade was added each year (i.e., K-1 in Year 1, K-2 in Year 2, K-3 in Years 3-4). The second area of PD was the interpretation of assessment results for students, based on the printouts that teachers received of their class with each student's scores across each measure over time. The researchers suggested benchmarks for average performance, and the teachers linked instructional activities to the benchmarks.

Measurement of Progress

Assessment that begins in kindergarten and continues through third grade raises additional challenges. We wanted to measure each of the esse\ntial reading components, but these components differ in importance across grades. For example, phonemic awareness, phonics, and listening vocabulary are important predictors and components of reading acquisition in kindergarten and first grade, but the influence of these components on reading achievement is usually superseded in second and third grades by reading fluency, vocabulary, and reading comprehension (Willson & Rupley, 1997).

According to Wagner et al. (1997), early screening of phonological awareness and letter knowledge to identify students who may develop reading difficulties is supported by the existence of stable individual differences in phonological language ability that influence the acquisition of reading skills. The predictive validity of phonemic segmentation and rapid letter naming measured in kindergarten and first grade has been well established (Good, Simmons, & Kame'enui, 2001; O'Connor & Jenkins, 1999; Torgesen, Wagner, & Rashotte, 1994). Good et al. (2001) found strong predictive correlations between scores on these measures in kindergarten and reading ability 3 years later. In the current study, we screened kindergartners on segmenting and letter naming in October and began to monitor progress in January to determine the students whose slow rate of progress suggested risk. To adjust for changing requirements of reading and skills of students over time, we followed students' progress by using measures of word recognition, fluency, and reading comprehension midway through first grade, and we continued these measures through the end of third grade.

Intervention for Students

To improve reading outcomes, students known to be lagging behind their peers should receive instruction that differs from the routines that were ineffective. It could be that their current instruction is appropriate, but they need more of it; or perhaps the grouping structure (e.g., whole-class instruction) offers too few opportunities to engage in essential reading tasks (e.g., identifying words, reading aloud, or generating the main idea of a passage). Alternatively, some students may not have acquired concepts or understandings that their grade-level peers learned some time ago, which suggests that they need a different content in their daily instruction. In this study, students who were identified as at risk for reading difficulties were provided with additional instruction by a member of the research team. This instruction is described in the next section.

TABLE 1

Study Design: Student and Teacher Participants Across Conditions and Years

Although there is strong support for attempting to decrease reading difficulties in kindergarten and first grade by increasing students' competence in the underpinnings of reading-namely, phonemic awareness and letter-tosound mappings-few studies have monitored the long-term effect of sustained intervention efforts across the primary grades. Our questions across the years of the study were, Given ongoing professional development in effective teaching of early literacy, what do teachers implement in their classrooms? What effect does professional development of teachers have on the literacy development of students? What is the effect of combined professional development and direct intervention on students' literacy outcomes? What proportion of students continues to struggle with reading despite early and ongoing intervention?

Method

Research Design

We used a longitudinal lagged design, in which teachers implemented changes in cohorts over the 4 years of the study, and control data were collected in the first year from the second and third graders in the same schools who did not participate in the intervention. Although the overall intervention plan included multiple cohorts over several years, only the students and teachers in the first cohorts (Layer 1, Layer 2, and Control) during the first year of data collection were included in the study we report here. Table 1 shows the combinations of grade levels and conditions for all of the participants whose scores were included. In the first year, teachers in kindergarten and first grade received professional development (PD; Layer 1), but only the kindergartners received additional intervention (Layer 2). As this group of kindergartners continued into Grades 1, 2, and 3, we called this group Layer 2, because they were eligible for two layers of intervention throughout their primary years: Layer 1, which included improved teaching through PD, and Layer 2, which involved additional intervention provided by the research team. Although the first-grade teachers also received PD in Year 1, their students did not receive additional intervention. We called this group Layer 1, because the only intervention that students received through this research was the instruction provided by their general education teachers in Grades 1, 2, and 3, along with the usual resources for intervention available in the school.

It is important to note that by the end of the study, the second- and third-grade teachers had participated in all three conditions: the control condition in Year 1, the PD-only condition (Layer 1) in Year 2 or 3, and the PD plus intervention for eligible students (Layer 2) in Year 3 or 4. By using this design, we compared reading outcomes of students who received their instruction from the same teachers in the same schools before and after making one and two layers of changes in instructional practices. By gathering control data before the teachers received professional development, we could be more confident that observed changes would be due to changes in practice and in layers of support rather than to changes in teachers, school climate, or administration.

Participants

The principals of two schools and all general education (n = 16), remedial (n = 2), special education (n = 2), and speech teachers (n = 2) agreed to participate. Student participants for the intervention groups included all students in kindergarten and Grade 1 (103 kindergartners and 103 first graders) who were enrolled in the two schools described later during the first year of the study, whether they were served in general or special education. Children in Grades 2 to 3 who were in these schools in Year 1 formed the control group (101 students in second grade and 102 in third grade). In Years 2, 3, and 4, we followed the first cohort of students in each intervention through their third-grade year. Attrition over time among the treated groups reduced the number of students to 90 by the end of Year 4.

TABLE 2

Schedule of Administration of Measures to Student Participants from the Beginning of Kindergarten Through the End of Third Grade

School 1 was located in an industrial area of a city in the Northeast that draws from a primarily low SES community. Fewer than 10% of the parents reported any college education. Ethnicity statistics reported 12% African American, 2% Hispanic, 3% Native American, and 83% European American. This school employed two full- time special educators (primary and intermediate grades), who work primarily with students with learning disability (LD) or mild mental retardation (MR), along with two full-time Title 1 teachers who assist students in inclusive classes and provide pull-out small- group reading instruction. In 1999 and 2000, 15% to 16% of the students in Grade 3 received special education services.

School 2 was a university-affiliated laboratory school located in an urban environment. Many of the students' parents were highly educated, and most paid tuition for their students to attend, although 12% were supported through tuition aid. Ethnicity statistics are 15% African American, 57% European American, and 28% Other. In 1999, 8% of students in the third grade were eligible for special education; in 2000, 11% were eligible. Across the two schools, only three students had first languages other than English, and their scores are not included here.

Measures

Table 2 shows the schedule of administration of the measures described in this section.

Prereading Measures. In October, January, and May of kindergarten, and in October of first grade, we administered measures of phonemic segmentation and letter naming (O'Connor & Jenkins, 1999). To assess segmenting, students listened to a three- phoneme word and were asked to say the sounds they heard in the word. Each of the 10 items was scored as the number of phonemes correctly identified. After the item was scored, feedback was provided, which included the correct response modeled by the tester prior to administering the next item. The split-half reliability (Spearman-Brown, corrected) was .96 for segmenting. For students who scored less than 27 on the segmenting test in October of first grade, we administered this test again with a new set of 10 words in January.

To assess letter knowledge, examiners showed students a card with the 52 upper- and lowercase letters randomly ordered in large type, and asked them to name as many of the letters as they could in 1 min. Interrater reliability on this task for a sample of 30 students was .94. We report the number of correctly named letters per minute. The prereading measures were read-ministered about every 3 weeks for high-risk students in Layers 1 and 2.

Reading Measures Collected From Experimental Students. In October, January, and May of Grades 1 to 3, we administered three subtests of the Woodcock Reading Mastery Tests-Revised (WRMT-R; Woodcock, 1998) to assess reading progress. The Word Identification subtest required students to identify words in isolation; the Word Attack subtest required students to apply phonic and structural analysis to pronounce pseudowords; and the Passage Comprehension subtest required students to silently read one or two sentences with a missing word signaled by a blank space and to supply a word that made sense in that space. We used alternate forms of the WRMT-R for each administra\tion. Split-half reliability estimates for Grade 1 on these subtests ranged from .94 to .98.

We collected oral reading fluency rates individually using a different grade-level passage at each testing point, beginning with January of first grade and continuing three times per year through the end of third grade. Interim fluency rates were collected about every 3 weeks for students in the high-risk group. Testers showed the child the passage and asked the child to begin reading when ready. As soon as the child began reading, testers timed the child reading aloud for 1 minute, following procedures outlined by Deno (1985). We marked errors on the score sheet and recorded the number of words read correctly in one minute. Interrater reliability on this task for a sample of 30 students was .91.

Receptive Vocabulary. The Peabody Picture Vocabulary Test, third edition (PPVT-III; Dunn, Dunn, & Dunn, 1997), is a norm-referenced test that we administered in March during the first and fourth year to students in Layers 1 and 2, and in March of Grades 2 or 3 for the control students. Each item consists of four black-and-white drawings arranged on a page. Participants were asked to identify, either by pointing or by saying the number that corresponded to the drawing, which of the four illustrations best represented the stimulus word presented orally by the examiner. Reliability (internal consistency) using coefficient alpha for students between the ages of 7 and 11 ranged from .95 to .96, with split-half reliability between .92 and .96.

Measures Collected from Control Students. In March of the first year of this study, we administered the PPVT-III to second- and third-grade students in the two experimental schools. In May, we administered the WRMT-R subtests and tested oral reading fluency on grade-level text. These students did not participate in the treatment described hereafter.

Documentation of Reading Instruction. Each year of the intervention, and in the first year for control classes, we surveyed teachers about their instruction, observed them teaching reading, and interviewed them about their instructional practices. On the surveys, used annually each March, we asked teachers to describe a typical hour of reading instruction, including the approximate time they would spend on each activity. We asked them to identify whether any features of the instructional activities differed from what they had done a year before. We also asked them to identify which parts of the lessons they thought were most beneficial for students overall and which parts were most beneficial for their students with reading difficulties.

Observations lasted 40 to 80 min and were conducted twice for the second- and third-grade teachers during their year as the control group (see Note) and three times for teachers in Layers 1 and 2. The purpose of these observations was to obtain a composite picture of student and teacher activities during reading instruction by recording classroom events as they occurred. Observers described teacher behavior and lesson content in a chronological format. Ongoing field notes described how the teacher delivered instruction, used materials, and grouped students, and also described the interactions between students and teachers and the level of engagement of the students with low reading scores. The process of gathering narrative field notes involved recording information by making entries at 5-min intervals. A checklist served as an accompaniment to recorded field notes and provided a quick record of classroom demographics, the materials used during reading instruction, general climate of the classroom, groupings of students, and the content of the reading lesson.

During the April observations, we documented activities that were consistent with and divergent from those we had observed in November and February and from the information they had provided a month earlier on the surveys. We followed up inconsistencies during the interviews. In the last year of the study, we videotaped 60 to 80 min of instruction in each of the participating classrooms for all but two of the teachers.

Data Collectors and Training

Beginning and midyear tests were administered by the authors, all of whom had extensive experience with these measures. They also delivered the PD and conducted small-group reading intervention in Layer 2. In May of each year, a set of testers (two education majors and a field supervisor in teacher preparation at the university) were hired to administer the outcome measures. These testers were not informed of the purposes of the project, and they had all administered the WRMT-R and the PPVT-III before on other projects. Two training sessions were conducted in which we reviewed procedures and taught testers to collect the oral reading fluency data. Testers were provided with stopwatches and several sample passages, and they learned to mark the passages and score them. They paired with the researchers to practice administering all of the measures, and the researchers presented scenarios that reinforced the decision rules for test basal and ceiling levels. Project personnel observed the testers during their first two administrations with students so that they could provide corrective feedback; however, none was needed.

Narrative observations were collected by the first author in Year 1, by the first and second author in Years 2 and 3, and by all authors in Year 4. In Years 2 and 4, the observers worked in pairs during the first three observations, so that discrepancies in field notes could be discussed and resolved. Because reliability was high (>.85), the remaining observations were conducted by single data collectors.

Layers of Intervention

Layer 1: Professional Development. During the first 3 years, teachers in appropriate grades, teaching assistants, and principals participated annually in three full-day (September, November, and March) and four 2-hour professional development sessions (October, December, February, and May). The content of these sessions shifted across years as teachers of students in later grades were added to the intervention participants. Each session began with the research evidence for emphasizing particular components of reading (e.g., phonemic awareness, phonics, and vocabulary in K-1; the alphabetic principle, vocabulary, word study, and fluency in Grades 1-2; multisyllabic word reading approaches and comprehension strategies such as developing the main idea, retelling, and summarizing in Grades 2-3), and continued with models of activities for teaching each component. During the last half hour of these meetings, groups of teachers generated instructional possibilities targeted to developing particular components of reading that were needed by some of their students.

In the first year, September content included teaching phonological blending and segmenting, letter names and sounds, and the alphabetic principle (Ball & Blachman, 1991; Hatcher, Hulme, & Ellis, 1994). We demonstrated activities from a kindergarten activity collection (Ladders to Literacy; O'Connor, Notari- Syverson, & Vadasy, 1998), and teachers discussed and rehearsed the activities in small groups. In November, we demonstrated strategies for phonics instruction (Beck & Juel, 1995; Carnine, Silbert, & Kame'enui, 1997; Iverson & Tunmer, 1993) and for integrating phonics with phonemic awareness (e.g., segment-to-spell; O'Connor & Jenkins, 1995; and word building; McCandliss, Beck, Sendak, & Perfetti, 2003). In March, we demonstrated strategies for teaching new vocabulary words and for promoting oral reading fluency, such as vocabulary expansion (Beck, McKeown, & Omanson, 1987) and rereading and partner reading (Dowhower, 1987; Mathes & Fuchs, 1993; Sindelar, Monda, & O'Shea, 1990). The afternoons were devoted to discussions among grade-level groups of teachers across schools and the research team; timelines for implementation; and schedules for coaching, data collection, and data sharing.

We provided ongoing support to school personnel through four additional meetings at each of the schools, in which we discussed the students' progress and how additional intervention could be orchestrated for the students making poor progress according to the collected data. In the weeks between inservice meetings, we dropped in on teachers' classrooms to provide feedback and instructional support for making modifications for individual students. These short sessions allowed unstructured time for questions, feedback, and discussion of students' progress.

In the second and third years, we focused professional development toward the second- and third-grade teachers who had served as controls in Year 1. In the first all-day session in Year 2, we reviewed the inservice content of the first year and extended strategies for developing word recognition and reading fluency (Carnine et al., 1997; Simmons, Kuykendall, King, Cornachione, & Kame'enui, 2000). During the last two days of professional development, we demonstrated linkages among reading fluency, accuracy, and comprehension (Durkin, 1993; Pressley, 1998; Shankweiler et al., 1999) and emphasized word analysis (i.e., decoding and multisyllable strategies, such as those developed by Lenz & Hughes, 1990, and Lovett, Steinbach, & Frijters, 2000) and comprehension strategies such as story grammar and generating main idea and summary statements (Jenkins, Heliotis, Stein, & Haynes, 1987; Jitendra, Hoppes, & Xin, 2000). Although we had planned these sessions with second- and third-grade teachers in mind, the K-1 teachers insisted on being part of these sessions as well. The four interim professional development meetings each year focused on data and activities for second- and third-grade students.

Although we taught teachers to use particular strategies and activities (e.g., oral segmenting, group responding, word analysis, main idea generation), teachers continu\ed to use the same background reading materials throughout the 4 years of the study (i.e., Harcourt Brace Jovanovich; Farr & Strickland, 1993; in School 1, and trade books in School 2).

Layer 2: Direct Intervention. Measures collected three times each year (and more frequently for high-risk students) allowed us to determine students' responses to Layer 1. In kindergarten, students who named fewer than 15 letters or 10 phonemes in January were considered to be at risk (see Good et al., 2001; O'Connor & Jenkins, 1999; for rationale). In first through third grades, students who scored 0.75 SD below average on the WRMT-R at any testing window were considered for additional intervention.

In the first year, 31 kindergartners scored below our cutoff scores for adequate achievement. In January, research personnel began providing instruction to these kindergartners in small groups of two or three students (following the recommendations for group size in Schumm, Moody, & Vaughn, 2000) for 10 to 15 min three times per week. In these groups, teachers scaffolded blending and segmenting activities by using smaller instructional sets and easier levels of tasks and providing more repetition and practice to develop key concepts. For most students, these small groups increased participation, response accuracy, and appropriate application to reading and writing task demands over their behavior in wholeclass groupings. By the end of kindergarten, nine students entered the average range on early reading measures and needed no additional assistance over the next 3 years. Among the other 22 students, some improved to the average range on phonemic and letter tasks, but they still needed more focused instruction to remain in the average range at subsequent testing points through the next 3 years.

In Year 2, we provided small-group instruction for 20 to 25 min three times per week to 17 first graders who met at-risk status during the October testing, including 15 students from the kindergarten group of the year before and 2 additional students who had demonstrated no difficulty with letter naming or segmenting in kindergarten but still had reading difficulties a year later. We used cumulative introduction to review short vowel sounds and supplemented classroom materials with sentences composed of decodable words. As students learned most of the letter sounds, we drew on more meaningful reading materials with primarily decodable text, such as the Bob Books (Maslen & Maslen, 1987), storybooks from Reading Mastery I (Engelmann & Bruner, 1995), and decodable mini- books from Open Court (SRA/ McGraw-Hill, 1998) to provide practice in applying learned skills while reading running text. Teaching in small groups allowed careful monitoring of individual progress and provided many more opportunities for students to respond, to read aloud, and to receive feedback during instruction. By the end of first grade, we had increased the intensity of Layer 2 intervention for 10 students by providing daily instruction in longer sessions, which for 3 students was delivered one-to-one. Analysis of this third layer of intervention is still underway; in this article, these students are included in the results for Layer 2.

In Years 3 and 4, we continued Layer 2 for 18 students in Grade 2 and 9 students in Grade 3. Some students in small groups spent roughly half of the time on word patterns, decoding, and spelling, followed by reading aloud in text containing those patterns. Other students attained skills near grade level (e.g., recognition of words and decoding) but still fell below desired reading rates. These students spent most of their small-group time reading and rereading text written at their instructional level to build fluency and comprehension.

Intervention Records and Observations. Two of the three small- group instructors for Layer 2 (a research associate and a doctoral student in special education) taught across the 4 years of intervention. In Year 3, we hired an additional instructor who was a certified special educator trained by the first author. These teachers recorded the content and attendance for each session and collected frequent oral reading fluency data. In the first 2 months of Layer 2 instruction each year, the first author observed each tutor two to four times. Thereafter, lessons were audiotaped monthly. When progress data indicated a low rate of growth for a particular student, the first author assumed a tutoring or observer role to identify the instructional features to adjust. These adjustments occurred for approximately one third of the students across the 4 years of the study.

Results

Teacher Implementation of PD Strategies

Field note records from the observations and interviews were transcribed into files and formatted according to NUD*IST 4. (Non- Numeric, Unstructured, Data, Indexing, Searching, and Theory Building; 1997, Melbourne, Australia: Qualitative Solutions & Research Pty, Ltd.) guidelines so that lines of text served as units of analysis. These documents yielded 1,911 lines of text. The research team began coding by first reviewing the documents and then developing an initial coding scheme through peer negotiation. Forty- six categories or nodes were assigned to text data. Categories included base data (data describing background or demographic information), strategies employed (data describing reading strategies used by teachers during the observations), instructional management (data describing how teachers managed and grouped students), and targeted student responses (data regarding the level of engagement of lower skilled students during reading instruction and teacher accommodations made for students). During this process, members of the research team made inquiries, offered impressions, and refined their notions regarding these data.

We have two sources for determining the changes that teachers made in response to professional development: self-reported changes from interviews and open-ended surveys, and changes documented through classroom observations. Eighteen classes were observed in Layer 1, which included two kindergartens (morning and afternoon sessions by the same teacher), two K-1 combinations, three first- grade, three second-grade, two combination Grade 1-2 classrooms, three third-grade, two Grade 3-4 combinations, and one learning support class. The other special education teacher was not observed; however, she completed the surveys and interviews.

Most of these classes were also videotaped in the final year. Research team members viewed video clip segments and coded them according to the hierarchy developed for the classroom observations; 100 of these video clips were transcribed for analysis. To address intercoder reliability of the video coding, 20% of the transcripts were selected for reliability analysis. Documents were coded by two researchers independently and compared. Intercoder reliability was calculated using the following formula: number of agreements/(Total number of agreements + disagreements) 100. With 416 agreements and 26 disagreements, intercoder reliability was calculated to be 94%.

Instructional Grouping. In the 10 control classrooms, 30% of teachers used small-group instruction for reading, all in combination classes. Following professional development, students were grouped in a variety of arrangements including whole group, small group, and paired practice. All of the Layer 1 teachers used at least two grouping arrangements for students during the observations, with 78% of teachers using small groups during part of the instructional session. More instructional time was spent in small groups (449 lines) than in any other arrangement.

Kindergarten and Grade 1. We do not have observations for kindergarten and first-grade classrooms prior to the intervention years, and so, for changes in practice, we relied on interview data, which we verified through observations during Layer 1. Teachers reported the greatest changes in phoneme awareness instruction, which consisted of shifting from an emphasis on the first sound only to all positions within words (181 lines). They also reported changing the timing of letter-sound instruction, with teachers in School 1 beginning 3 months earlier in the school year than prior to this study, and 1 month earlier in School 2. In Year 1, two of the kindergarten teachers followed the suggestion of Carnine et al. (1997) to separate letter shapes and sounds that students commonly confuse (e.g., using a, m, t, s, d, i as the first letters, and teaching 10 letters in between d and b and between i and e). In Year 2, all but one of the kindergarten teachers adopted the new order.

First-grade teachers reported their greatest changes in decoding instruction. Three teachers in School 2, along with their special education teacher, met to develop a common sequence of letter-sound instruction that included introducing short vowel sounds, phonograms, and vowel patterns. Their goal was for students to have consistency across settings. Teachers across both schools focused on letter-by-letter decoding early in the year in the first grades, formed two or three reading groups, and encouraged students to stretch sounds when blending and to segment sounds when spelling words. This instruction shifted toward reading connected text aloud in small groups during the last observation in April (162 lines). We observed fewer changes among teachers in School 2 than in School 1, mainly because School 2 teachers already grouped for instruction to accommodate the multi-age class structure.

Grades 2 and 3. In second and third grades, we were able to document changes through differences in the observations between the first and second years in Grade 2 and the first and third years in Grade 3. The shift toward the use of flexible grouping, small-group instruction, and reading aloud were the most dramatic changes. Only three teachers had grouped \students during the control year observations (where most teachers of combination classes provided reading instruction by grade level), whereas all but one teacher used small instructional groups during the intervention observations.

During control observations, one teacher encouraged oral reading of connected text; however, in the intervention classes, more than 80% of teachers included oral reading, with more than twice the time reading aloud (82 lines of text) rather than reading silently (33 lines). Most teachers also spent 87% of Layer 1 observations teaching word analysis strategies, such as the use of affixes and morphemes and the BEST strategy (Break the word apart, Examine the stem, Say the parts, Try the whole thing-a variation of Lenz & Hughes', 1990, DISSECT; 60 lines). More than half of these teachers reinforced these strategies in practice sessions with small groups of students with lower skills. We observed teachers in their control year drawing students' attention to words on the board or overhead projector, but we saw no direct teaching of approaches for reading multisyllabic words. Unlike first grade, changes in the second- and third-grade classes were similar across schools.

Across grades, teachers reported during interviews being more aware of the strategies they used in reading instruction. They talked about mindfulness of their instructional choices as a result of professional development and also of regularly scheduled meetings during which they discussed their teaching plans in relation to student progress.

Student Outcomes

The layered design of data collection allowed us to compare the control group (same schools prior to intervention) at the end of second and third grades with the effect of professional development alone (Layer 1) and the combination of professional development and direct intervention (Layer 2). We did not use hierarchical linear modeling or repeated-measures analyses in these comparisons because students in the control groups were measured at only one time point; thus, students in the third-grade control group were not the same students as those in the second-grade control group. Group differences were examined with a series of one-way analyses of variance (ANOVAs) on outcomes at each time point, with planned comparisons between the control and each treated group. Next, we examined treatment outcomes for specific disaggregated groups of students: (a) students identified as at risk, (b) students from low- income families, and (c) students with disabilities.

Student Outcomes Overall at the End of Grades 2 and 3. The columns in Table 1 show the groups of students included in these analyses. Descriptive statistics that show reading achievement at the end of Grades 1, 2, and 3 and the effect sizes (Cohen's d) of the layers versus control groups are shown in Table 3, along with demographic information on the students who formed each group. Because the kindergarten students received both Layers 1 and 2, they had no comparison group. Moreover, we lack pretest data from kindergarten and first grade to demonstrate comparability to the control group at that age; however, scores on receptive language (PPVT-III) did not differ in Grades 2 or 3.

TABLE 3

End-of-Year Outcomes by Grade and Group

ANOVA on second- and third-grade reading outcomes (Layer 1 versus Layer 2 versus Control) were all significant; Word Identification, Word Attack, Comprehension, and Fluency scores, respectively, for Grade 2 were F(2,281) = 7.23, 20.20, 11.73, 9.29, p < .01; Grade 3, F(2,283) = 4.48, 16.24, 9.97, 36.96, p < .01. Children in Layer 1 outscored the controls in all of the reading measures except Word Identification at the end of Grade 2. Children in classes where Layer 2 was available outscored the controls in all of the reading measures at both time points. By the end of third grade, effect sizes for professional development (Layer 1) over the controls for Word Identification, Word Attack, Reading Comprehension, and Fluency were .19, .34, .27, and .45, respectively-all in the small to moderate range.

Outcomes for Students at Risk. Although we began with 31 kindergartners at risk in Layer 2, we excluded from these analyses the students who caught up in kindergarten and needed no further assistance, which provides a conservative estimate of effects. Here, we analyze the combined outcomes for students at risk and students with disabilities (n = 22, 23, 23 for Layer 1, Layer 2, and control, respectively). We combined these groups because students with disabilities were identified at different time points over the course of the 4 years. All of the students in Layer 2 received additional direct intervention at some point during the K-3 years, and seven students received it for all 4 years. Many students in Layer 1 and the control group received Title 1 assistance beginning in first grade, and all eligible students received special education as they were identified. Means and standard deviations for high- risk students in each treatment across years are shown in Table 4.

TABLE 4

End-of-Year Outcomes for Students in the High-Risk Category by Grade and Year

As with the larger sample, significant differences in outcomes were found. Professional development (Layer 1) alone improved reading outcomes significantly over the control group for Word Attack, Comprehension, and Fluency in Grades 2 and 3, F(1, 43) = 19.52, 7.78, 15.14, respectively, for Grade 2; F(1,43) = 19.50, 5.28, 7.79, respectively, in Grade 3. The addition of direct intervention (Layer 2) showed larger effect sizes over the control group and significant differences across all reading areas, F(1, 44) = 21.36, 37.74, 24.86, 40.40, for Grade 2 (all p < .01); F(1,44) = 8.63, 35.95, 13.54, 18.87, in Grade 3 for Word Identification, Word Attack, Comprehension, and Fluency, respectively. Effect sizes for Layer 2 over the control group across these measures at the end of Grade 3 were moderate (.40, .67, .48, and .54, respectively). The effect sizes for Layer 2 were somewhat larger in second grade, which may reflect the increasing difficulty of reading processes in third grade.

Student Outcomes by School. The participating schools enrolled students who differed along several demographic features, including socioeconomic status (SES) of the family. Not surprisingly, students differed on most reading measures in kindergarten and first grade, with students in the more affluent school (School 2) scoring consistently higher. To estimate the effects of SES on treatment, we analyzed outcomes by school and treatment for the students in the high-risk group. The effects of school (Wilks' lambda = .734) and the treatment by school interaction (Wilks' lambda = .750) were significant, with the interaction due to the very low scores in School 1 for students in the control group. When we reanalyzed the interaction using only treated students (Layers 1 or 2), the results were no longer significant (Wilks' lambda = .858), which suggests that the treatments were effective for students of relatively higher and lower SES.

Identification of Reading Disability

In the 2 years prior to this study, the incidence of placement in special education averaged 15% (14.6% in the year the control group data were collected). Following 4 years of participation in this research, the rate of placement was 11.8% in the professional development condition (Layer 1) and 7.8% in professional development plus direct intervention (Layer 2). In each condition, one or two students were eligible under the category of mental retardation (MR) and the remainder under the category of learning disability (LD). The scores of students with LD in third grade are shown in Table 5. (The four students across conditions with MR are excluded from the tabled statistics.) Due to differing numbers of students with LD across groups (13 in control; 11 in Layer 1; 6 in Layer 2), we did not perform statistical tests on these scores. Although scores for students in Layers 1 and 2 appear similar, the importance lies in the declining number of students eligible for LD across conditions.

TABLE 5

Descriptive Statistics for Scores Disaggregated by Achievement Status and Treatment in Grade 3

Discussion

We began this study with the premise that we could reduce reading difficulties for students in the early grades by paying more attention to the foundational skills of phonemic awareness, phonics, word identification, and fluency. To accomplish that attention, we provided extensive PD to teachers, measured students' progress across the foundational skills, and provided more intensive instruction for students who continued to lag behind even as the instruction provided by their general education teachers improved. Our findings indicate that early and continuous intervention for students with high-risk profiles in kindergarten through third grade improves their reading outcomes. Although this validation is important, we were not surprised, because studies of early intervention over the past several years commonly have reported improvements for students in kindergarten and first grade. Nevertheless, some of the findings from this study were unusual.

We were surprised with the effect on students' reading achievement of our first layer of intervention, which was provided to students in the general education classrooms setting by their teachers. Although the intuitive connection between improved teaching and improved student achievement is appealing, few studies have reported such effects of professional development on achievement (National Reading Panel, 2000), because it can be difficult to isolate the effect of the PD from that of teaching competence in general. In our case, we cannot attribute students' literacy improvement to the relative effectiveness of teachers across intervention and control groups, because the teachers of the second and third graders were the same individuals regardle\ss of condition. Only one general education teacher was replaced during the 4-year intervention, and no changes occurred among the special educators.

Nevertheless, we identified a snag in our attempt to attribute improved student outcomes to PD alone. The first layer of intervention had two distinct features: professional development to increase the competence and confidence of teachers to teach reading, and periodic analysis of performance data on students' progress toward reading acquisition. During PD, teachers examined and discussed the trajectories of students and talked about their growth objectively and subjectively in comparison with other students. By the third year, teachers were skilled readers of data spreadsheets and could identify the students who were making slow growth in reading. These activities interfaced with the content of professional development, because we asked teachers to reflect on the needs of particular students in their classes as they discussed implementation.

Teachers responded to the challenge of changing their practices. Grouping for instruction was one dramatic area of change; however, it was initially difficult for teachers to manage. They worried about misbehavior among students who were not part of the current reading group, and it took a combination of our recommendations and discussions among the grade-level teams to work out the details. In kindergarten, grouping came easily because the teachers used learning centers routinely. In first grade early in the year, teachers kept their instructional groups short-usually 10 to 15 min per group following a whole-group activity. Toward the end of the year, the reading groups were 20 to 30 min each. The independent activities most commonly used were handwriting practice, rereading pages read during small-group instruction, and segment-to-spell with pictures. Two of the teachers in third grade used whole-group instruction for about half the time and then drew together a small group for focused practice on the decoding of multisyllabic words and reading aloud, while the rest of the class worked independently. At second and third grades, teachers generally had students who were not in their directed group engaged in story-related writing and partner reading. In addition to providing more focused instruction for small groups of students, several teachers used materials at lower levels of difficulty for struggling readers. It seems likely that controlling the reading level of materials offered more redundancy for high-frequency words, word patterns, and vocabulary, which some researchers (Dowhower, 1987; O'Connor, Bell, Harty, Larkin, Sackor, & Zigmond, 2002) suggested can lead to improved fluency. Fluency was the area with the largest effect size between Layer 1 and control students.

The instruction by the two special education teachers was already closely aligned to the instructional model used for PD. Both teachers used pull-out reading groups; however, one teacher supplemented general class instruction, and the other supplanted it. The change we identified was in teachers' attempts to increase the integration of reading approaches across general and special education after students were identified for special services. A second-grade teacher told us, "Before [this study], special ed was a totally separate thing-different methods, different materials. Now, [student name] carries his book from her class into mine, and says, 'Oh, I learned that from Ms. Z' [special education teacher]. Or Ms. Z tells me she showed [student name] how pairs of letters go together and he says, 'That's just like my spelling words.'" In the other school, general and special education teacher pairs developed a notebook kept on a shelf in the general education classroom for each teacher to jot down specific aspects of literacy instruction that the other could reinforce with particular students across the school day (e.g., word patterns, spelling words, new vocabulary). Teachers told us that the notebook helped them to shape coherent instruction for students who received services from multiple sources.

Not everyone jumped on the Layers bandwagon. Two teachers made no changes that we could identify. Both already used some of the features and strategies we had recommended in the PD, but they seemed resistant to less familiar practices that may have been useful in stimulating their poorer readers, such as segmenting words prior to spelling them or adding more opportunities for students to read aloud. Perhaps they recognized their own routines in our examples of research-based practices and so saw no need to change. We could not identify ways in which these teachers differed from those who made instructional changes; both had more than 10 years of experience (but not the most experience); both were well regarded by and strongly connected with other teachers in the school who made multiple changes. They attended all meetings and contributed to discussions. Nevertheless, we observed no implementation of any new instructional practice clearly related to the PD component of this study in these teachers' classrooms.

Layer 2

Layer 1 increased the participation of students with reading difficulties; however, additional assistance (Layer 2) was needed for 30% of students in kindergarten, 21% of students in Grade 1, 19% in Grade 2 (eight continuing to receive Layer 2 from first grade and two whose scores had been too high for eligibility a year earlier), and 10% in Grade 3. Although it is possible that the nine students identified early in kindergarten who caught up were helped by our intervention, it is just as likely that they represented the overidentification of LD common in prediction studies that have used measures similar to ours.

Layer 2 was not a "one size fits all" intervention. Small-group activities were tailored to the needs of the two or three students in the group. As these needs changed, groups were rearranged. Thus, not all students in Layer 2 received the same type or the same amount of reading instruction. Layer 2 did not always incorporate all of the components known to influence reading development. Some groups focused only on word-level strategies. Other groups included word analysis and strategies for reading comprehension. In second and third grade, some students received intervention that focused solely on increasing reading fluency. The flexibility of the content and scheduling of these small groups may have been a factor in their success. The duration of Layer 2 varied from 8 weeks to several years, depending on the students' progress. As students appeared to catch up to their classmates in reading, they were dropped from the Layer 2 interventions; as teachers learned ways to provide appropriate practice, some teachers were able to orchestrate these opportunities without pull-out assistance from project personnel.

Children "Fixed" at One Stage Falter at Another

Our second surprising finding was the transience of "catching up" for some students. We found segmentation to be relatively easy to teach and letter identification and letter-sound pairings straightforward; however, some students who were good at these skills still had difficulty blending known sounds into words or identifying irregularly spelled words. For example, two students were not identified in kindergarten because they learned phoneme segmentation and letter sounds without much difficulty, but in first grade, blending became critical to their success in decoding words. They joined a small-group intervention (Layer 2) that focused on blending, and one of these students rose to the average range by the end of first grade and stayed there at each testing interval thereafter. The other child needed an additional year of small- group intervention to catch up to grade-level performance. Other students-with the assistance offered through Layer 2-were able to keep up with their peers when reading generally consisted of one- syllable words. Because they caught up, we released them from Layer 2, only to catch them again as words became commonly multisyllabic and they needed specific strategies for reading these kinds of words.

It is important to note the fluidity of these groups. Only five students scored consistently below cutoff points at every testing cycle. Five others who received some Layer 2 intervention during each of the 4 years caught up on the measured reading skills at some time points, but without the supplemental small-group instruction, they were unable to maintain average performance commensurate with their peers.

Even students good at identifying individual words can have difficulty with speeded recognition, which affects reading fluency. That was the case for three students in our sample who, with an additional 20 min each day in fluency practice, were able to keep pace with grade-appropriate requirements for reading. Theory would suggest that it may be possible for students to master all word reading skills and still have difficulty with reading comprehension of longer texts, which is a hallmark of proficient reading in third grade and beyond. Although two groups of second graders and one group of third graders focused only on building reading fluency and comprehension, across the two schools, we found no students with grade-appropriate fluency and poor silent reading comprehension.

Response to Treatment

Researchers who monitor the progress of students in enhanced kindergartens and first grades (e.g., Torgesen, 2000) have found that not every child responds well to good interventions. For 19% of our students, intervention stretched over more than one school year. Four of the 18 first graders who fit the at risk category in K-1 improved to the average range as second graders. By third grade, only nine of the students identified in kindergarten continued to need interventions in addition to general education instruction. Of the sevenstudents who received Layer 2 for all 4 years, one entered kindergarten already diagnosed with developmental delay (later changed to LD), and six became eligible for special education: one with MR identified in first grade, three with LD in Grade 2, and two with LD in Grade 3.

We were encouraged by the finding that SES did not influence response to treatment significantly. Although we found differences that favored more affluent students at the beginning and end of the study, learning gains across schools were similar, and students without disabilities in the low SES school were above the national average by the end of third grade.

Our measures caught most students who eventually developed RD, but they were not able to identify kindergartners who would respond well to Layer 2 (defined as reading achievement that reached and maintained the average level) from those who were later identified with LD. By the end of first grade, however, all students who were later eligible for special education were receiving additional intervention, and most of these students were receiving daily assistance that was more intense than the Layer 2 intervention that was initially offered to them.

Achieving rates of fluent reading sufficient for maintaining reading comprehension can be another stumbling block for students with RD (Lovett et al., 2000), even as discrete skills such as the decoding of words improve. Among the control group, 21 students read fewer than 50 words per minute in grade-level text at the end of third grade, whereas 5 students in the PD condition and 2 who participated in Layer 2 read fewer than 50 words per minute by the end of the study.

Rates of Placement in Special Education

By assessing the effects and outcomes of each layer of intervention, we hoped for decreased placement in special education. Although the rates of placement in special education appear lower following Layers 1 and 2, these rates are not necessarily consistent in a school building from one year to the next. Therefore, we are not confident that this decrease is a result of the early intervention provided through this study.

Even so, students reading near the average range on the WRMT-R were referred to special education. One possible reason for these "nearly average" referrals was the discrepancy between the reading performance of students eventually labeled with LD and the average reading achievement of students in their classrooms. When we examined the spread of scores in third grade during the treatment and control years, this possibility warranted attention. Although scores of the top 25% of students across years appeared to be about the same, the scores of average students rose about half a standard deviation. Thus, students diagnosed as having RD were approximately 1.3 SD below the mean of their peers, which is about the same difference between students with and without LD in the control group.

A second possibility is that students were referred to special education for reasons other than their reading achievement. Two of the seven students with disabilities had already been diagnosed by first grade. We did not collect data on whether improving reading ability affected other academic or social areas related to disability, which could have been reasons for referral.

A third and, we believe, the most likely reason that students approaching average scores were referred for special education is the effort it took to keep these students at the "nearly average" level of reading achievement. In essence, we were already providing special education as we increased the intensity of our instruction by decreasing group size, increasing the frequency of pull-out instruction, and targeting instruction to specific areas of need identified through our periodic assessments.

Limitations

One problem with the interpretation of our results is the frequency of measurement for the experimental students. We might have strengthened our design by measuring students in the control years on the same schedule as during the treatment years. For this study, however, that would have led to a long delay in beginning the treatment, which neither we nor the schools were willing to accept. We attempted to counter this possibility by bringing in new testers for the end-of-year testing in all conditions to reduce the likelihood that students in the treatment years performed better because they were more familiar with the testers; nevertheless, we must acknowledge the potential confound with test practice.

The project design included fading researcher support from one year to the next (e.g., the focus of professional development shifted up one grade level each year), while continuing to provide students' progress data to teachers, which is known to improve the likelihood of maintaining changed practice over time (Gersten et al., 2000). However, we do not yet know the extent to which teachers will sustain the changes in practice that contributed to improvements in the literacy development of their students.

Finally, we lack confidence in the finding of fewer students being identified with LD by third grade, because referral and eligibility rates could rise as reading material becomes increasingly difficult in fourth grade and beyond. The toughest area of reading for us to repair was reading fluency, and recent research has suggested that the combination of fluency and comprehension assumes the functional load for reading during the intermediate grades.

Lingering Issues

Comparisons of reading programs for struggling readers (Hatcher et al., 1994; Juel & Minden-Cupp, 2000) have found that the best approaches were those that combined features of effective instruction in a consistent package delivered with sufficient intensity and duration. Although many students rotated in and out of the additional interventions offered in Layer 2, some students were identified as struggling readers early and received continuous intervention across 4 years. This level of intensity brought many students within the average range of reading on standardized measures; however, as the average performance of thei

Source: Journal of Learning Disabilities

More News in this Category