Education I NEED HELP ON MY HOMEWORK 3

Learning Disabilities Research & Practice, 19(4), 239–251 Copyright C© 2004, The Division for Learning Disabilities of the Council for Exceptional Children

English Reading Effects of Small-Group Intensive Intervention in Spanish for K–1 English Learners

Michael Gerber University of California, Santa Barbara

Terese Jimenez Loyola Marymount University

Jill Leafstedt California State University, Channel Islands

Jessica Villaruz, Catherine Richards, and Judy English University of California, Santa Barbara

In this article we report small, but statistically significant, effects of brief supplemental instruc- tion on English reading by Spanish-speaking kindergartners (N = 37) who performed poorly on a bilingual battery of phonological-processing tasks. Intervention design was compatible with the Reading First initiative and with research on use of multitiered intervention strate- gies for preventing reading failure among young monolingual students (e.g., L. S. Fuchs & Vaughn, 2003). We describe a Core Intervention Model (CIM) comprised of specific instruc- tional behaviors that teachers might easily learn and employ regardless of curriculum, and discuss implications of our findings for building multitiered preventative instruction for young English learners.

Scientific evidence supports as effective those early read- ing intervention programs that include such characteris- tics as (1) low child-to-teacher ratios, (2) structured and fast-paced designs, (3) an emphasis on fluency as a pri- mary goal, (4) a well-designed and effective regular class- room reading program, (5) evidence-based techniques, (6) direct and explicit instruction, and (7) ongoing assessments (Snow, Burns, & Griffin, 1998; Foorman & Torgesen, 2001; Justice & Pullen, 2003; National Reading Panel, 2000; Pikulski, 1994). These and other elements have been found to be effective for identifying risk for reading disabilities among young monolingual students and as a basis for design- ing interventions aimed at preventing that failure (Fletcher & Lyon, 1998; Fletcher et al., 2002; Lyon, 2003; Lyon et al., 2001; Torgesen et al., 2001). However, interpretation of the same performance on English-reading tasks by young English learners (ELs) is problematic because they have lim- ited exposure and proficiency in English; these children are often confused with students who have learning disabilities (Alanis, Munter, & Tinajero, 2003). Whatever individual dif- ferences may exist that would predict later reading difficulties are masked to some extent by variations in previous English- learning opportunities. If we could account for these varia-

Requests for reprints should be sent to Michael M. Gerber, 2327 Phelps Hall, Gevirtz Graduate School of Education, University of California, Santa Barbara, CA 93106. Electronic inquiries may be sent to mgerber@ education.ucsb.edu.

tions in some way, it is reasonable to expect that the same elements of early assessment and instruction that are effec- tive for monolingual students will be effective for ELs as well. What is missing in this body of research-based recommen- dations for reading practice is a better understanding of how to conduct monitoring and intervention programs for young ELs who are experiencing difficulties learning to read in English and who are easily confused with students with learn- ing disabilities.

Preventing Reading Failure

Special education policy since 2002 has been increasingly aligned with the No Child Left Behind Act (NCLB). Un- der NCLB, schools must engage in focused and account- able efforts to improve educational outcomes for all stu- dents but, especially, to close the gap in achievement for historically low-achieving students, including those who are disabled and those who are English learners. Fundamen- tal to this effort is a strong orientation toward preven- tion of reading failure. This orientation is emphasized by NCLB’s Reading First initiative. This initiative provides fund- ing and guidelines for establishing “high-quality, compre- hensive reading instruction in kindergarten through grade 3” and has been increasingly associated with a “three- tier model” of intervention (L. S. Fuchs & Vaughn, 2003; National Academy of Sciences, 2002; Vaughn, Bos, & Schumm, 2001).

240 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs

Three-Tier Model of Intervention

The central concept of a “three-tier model” of intervention is to monitor progress, frequently and continuously, using measures that are reliable predictors of later reading profi- ciency. By this means, teachers can identify those students who, despite good instruction, fall behind peers in rate as well as level of achievement. Generally speaking, teachers engage in primary (i.e., Tier 1) intervention as soon as their monitoring identifies students. Primary intervention may in- volve not only more instruction over a specified period of time (e.g., 10 weeks), but also some reconsideration and re- design of normal curriculum and instruction for those stu- dents who are failing to keep pace. Secondary (i.e., Tier 2) intervention occurs when monitoring reveals that students who received primary intervention nevertheless continue to lag behind peers. Secondary intervention implies instruction that occurs over and above normal but high-quality classroom instruction. It may involve tutorial or small-group instruction for 30–60 minutes each week on specific and important pre- reading skills (e.g., detecting rime or onset of words, segment- ing or blending phonemes, mentally manipulating phonemes by deletion and substitution). Tertiary instruction (i.e., Tier 3) is a highly intensive, for example, 50 hours or more over 10 weeks, of tutorial or small-group instruction for students that appear to resist good instruction, those who are persistent strugglers despite primary and secondary interventions.

Although researchers are steadily converging on bench- marks for identifying nonresponsive monolingual students, it is unknown at this time how to apply these identification and intervention strategies for young ELs. Many students who come to kindergarten lacking exposure to English phonology, alphabet, and vocabulary, for example, are unlikely to meet reading benchmarks under the progress-monitoring system described above. Therefore, it is certain that many will be perceived to be at risk.

It can be argued that erring on the side of overidentification for these children does no harm, but there is also a need to be economical with instructional resources allocated for inten- sive interventions. In addition, it is unclear whether the same instructional targets should have priority for students who are just learning English at school. How, for example, can ELs learn to associate English alphabet with English phonology when they do not yet know English phonology? Finally, the multitiered strategy described above depends critically on a high baseline level of instructional quality. It is only against that baseline that failures to progress can be interpreted mean- ingfully. However, we do not yet know to what extent what we perceive as high-quality instruction for monolingual students will be equally appropriate as well as effective for ELs.

Instruction and Preventative Intervention for At-Risk English Learners

Two decades of research have established that differences in phonological-processing abilities—particularly phono- logical awareness (PA)—are predictive of the degree to which monolingual students experience later success or failure in word reading (Ball & Blachman, 1991; Bradley

& Bryant, 1983; Christensen, 1997; Goswami & Bryant, 1990; Stanovich, 1986; Torgesen et al., 1999; Wagner & Torgesen, 1987). Initially, relatively few researchers seemed to recognize the implications for ELs, but there is an in- creasing number of empirical studies exploring to what ex- tent phonological-processing abilities represent core cogni- tive abilities, independent of one’s particular spoken language that might be available (i.e., transfer) as resources when learn- ing to read English as a second language (e.g., Bialystok, Majumder, & Martin, 2003; Cisero & Royer, 1995; Durgunoglu, Nagy, & Hancin-Bhatt, 1993; Geva, Wade- Woolley, & Shany, 1997; Gottardo, Yan, Siegel, & Wade- Woolley, 2001; Leafstedt, 2002; Riccio et al., 2001).

Moreover, while substantial research now has been de- voted to the scope and sequence of phonological prereading tasks that both promote and predict word-reading fluency for monolinguals, less has been written about how to frame such a curriculum for English learners or what specific in- structional behaviors will be effective for them. A new best- evidence synthesis of literature on reading programs for ELs (Slavin & Cheung, 2003) reveals a dearth of rigorous re- search on programs of reading instruction for ELs. Systematic study of early English-reading acquisition by ELs is similarly scarce. Researchers only recently began reporting longitudi- nal studies of English-reading development among groups of English learners (e.g., Gerber & English, 2003; Lindsey, Manis, & Bailey, 2003; Siegel et al., 2003; H. E. Swanson, Saez, Gerber, & Leafstedt, 2004).

Nevertheless, because interventions to teach phonolog- ical skills to monolingual English-speaking kindergartners have been successful in promoting improvement in word- reading ability (e.g., D. Fuchs et al., 2002; O’Connor, Jenkins, & Slocum, 1995), it seems reasonable to expect sim- ilar results for ELs by way of the cross-linguistic transfer of L1 phonological-processing skills to L2 reading. On the other hand, kindergartners in monolingual studies received instruction based on a phonological system with which they already were very familiar and highly practiced. That fa- miliarity might confer an advantage despite cross-linguistic transfer of some processing abilities. Therefore, in the study we report in this article, we assumed that training individu- als to improve performance on tasks requiring phonological- processing skills requires instruction that initially capitalizes on the phonological system with which students are most familiar. Therefore, our primary intervention for ELs who appeared to be at risk on bilingual monitoring measures was conducted in their home language (L1) to influence learning and performance in English (L2).

Curriculum and Instruction in Early Reading

Most studies of early intervention to prevent reading failure have focused on a specific scope and sequence of skills (i.e., a curriculum) determined to be critical to later word reading. Whatever individual adaptations that teachers make to indi- vidual differences in performance, attention, or behavior are generally not reported. For teachers of ELs whose students are also diverse both in first and second language proficiency, a distillation of critical features of effective instructional

GERBER ET AL.: EFFECTS OF SMALL-GROUP INTERVENTION 241

behaviors, independent of a specific reading curriculum, would be useful for raising the baseline of primary inter- ventions for ELs perceived as being at risk.

We may assume that extant studies of early reading inter- vention have employed an array of research-validated prac- tices and that among these are some principles of teaching that are drawn from the same knowledge base that informs instruction of students with learning disabilities. Specifically, research literature provides extensive evidence that students with learning and reading disabilities benefit from instruc- tional interventions that include elements of direct instruc- tion (H. L. Swanson & Hoskyns, 1999). In direct instruction, skills are rationally divided into hierarchical steps and are taught to a mastery criterion, using rapid pacing, large number of individual or group response opportunities in small, rela- tively homogenous groups. Thus, a direct instruction model fits well with the goals of multitiered interventions focused on a specific set of critical prereading and reading skills.

Furthermore, although one-to-one instruction has been used to produce maximum results in many recent studies of early reading interventions (e.g., Torgesen et al., 2001; Vellutino et al., 1996), this may not be a realistic goal for teachers and schools (L. S. Fuchs, 2002) with limited re- sources. Fortunately, reading instruction for small groups of students (i.e., three to five) can achieve results that may equal or surpass comparable one-to-one instruction (Elbaum, Vaughn, Hughes, & Moody, 1999; Foorman & Torgesen, 2001; National Reading Panel, 2000; Rashotte, MacPhee, & Torgesen, 2001; Vaughn, Hughes, Moody, & Elbaum, 2001). Again, a direct instruction model based on small-group in- struction not only helps maximize response opportunities for individual students deemed to be at risk, but also provides ex- plicit language models for students struggling with English skills.

Core Intervention Model (CIM)

In constructing an intervention model for this study, one that would be suitable for teachers to use in a multitiered system of instruction for at-risk ELs, we were guided by three consid- erations. First, we conceptualized intervention as necessarily compatible with current models of multitiered interventions (e.g., Reading First) in which supplemental instruction is pro- vided for students shown to be at risk based on low perfor- mance of critical prereading skills. Second, we wanted the intervention model to be simple in conception and implemen- tation to maximize its uptake by teachers, so we concentrated on instructional behaviors and techniques in addition to a simple developmental sequence of skills (Christensen, 1997). Third, we wished to test the hypothesis that cross-linguistic transfer of important phonological skills to L2 reading can be promoted by intervention in L1 prereading skills for young ELs.

Specifically, to provide supplemental, more intensive in- struction for students who appeared to be at greatest risk for experiencing difficulty learning to read English words, we designed a Core Intervention Model (CIM) based on general principles of direct instruction. Our CIM is designed for di- rect instruction of groups of four students who perform most poorly on phonological-skills screening. CIM teaching em-

phasizes example items for learning and practicing reading- related skills, rather a curriculum in its conventional sense. The sequence of teaching events is set by two considerations. The first was the relative fluency of students’ performances on items representing four, developmentally sequenced tasks— rime detection, onset detection, phoneme segmentation, and phoneme blending. The second was the sequence of steps in teaching the items, particularly the contingent correction, or reteaching, that was triggered by students’ errors in respond- ing. We used simply constructed materials to illustrate or sup- port the direct instruction teaching routines, again with the purpose of making the CIM maximally adaptive and usable for teachers. Following from principles of direct instruction, the CIM required teachers to be explicit and concise in ask- ing questions, to proceed at a rapid pace, and to monitor and contingently respond to errors with a specified subroutine whenever these occurred. That is, students were enthusias- tically praised for correct responses and incorrect responses triggered a series of defined prompts from the teacher that guarantee eventual success.

These explicit correction procedures for systematic reteaching are fundamental to our CIM. We call our correc- tion subroutine the Correction Staircase. The theory behind the Correction Staircase is that cognitively complex demands can be reduced systematically until students are able to re- spond correctly and then can be recomposed in “steps,” each providing a scaffold that supports the next higher step, until students are able to respond correctly to the original high- demand question.

Each teaching sequence in the CIM begins with what we call a “supply” question with an illustrative item (e.g., “What word rhymes with cat?”). These are called “supply” questions because students have to construct and provide a response, including orchestrating component processes for understand- ing the question, performing multiple, controlled operations in working memory as well as search of longer-term memory, while formulating and articulating a response. When students commit an error (or are unable to respond), teachers “step- down” the demand of the question by reducing it to a binary choice requiring recognition of correct answers; for example, “What rhymes with cat, calf or bat?” Students are praised for the correct response, as they are for each and every correct response throughout the use of the Correction Staircase. If students select correctly, teachers “step up” and again ask the original supply question.

If students cannot respond correctly to the binary choice question, teachers step-down to a model-lead step. At this step, the teacher first models the correct response (e.g., say- ing, “Bat rhymes with cat. What word rhymes with cat?”). Again, if the student is correct the teacher steps-up to the next more complex question (i.e., the binary choice) and contin- ues upward until the original supply question is presented and correctly answered. That is, teachers always begin and end a correction sequence with a supply question.

If students are still incorrect at the model-lead step, teach- ers step-down to the model-imitation step. At this step, stu- dents are prompted to imitate the teacher’s model of a correct response (e.g., “Say bat.”). Students are prompted until they repeat the correct answer. Again, as before, after any correct response, teachers immediately step-up to the next higher

242 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs

step until they culminate the sequence at the original supply- type question. If students respond to the supply question, the teacher might move on to the next question in his or her lesson sequence or present another, similar supply question to test for learning (e.g., “What rhymes with lip, lint or clip?”). If students are incorrect, the teacher again employs the Correc- tion Staircase as described above. Generally, teachers do not need to provide explanations to link the steps of correction. To do so, in fact, may increase the difficulty for ELs because it imposes additional language demands and may distract them from the focus on critical phonological discriminations that this technique promotes.

The staircase approach assures that students always are led to the correct answer, are given the opportunity to re- spond independently, and are reinforced positively for doing so. To use the staircase approach, teachers must understand its limited but important instructional objectives. Teachers must be able to create a lesson sequence in the set of illustrative items or problems that they choose and the fluency criteria (i.e., fast, accurate responding) for which they monitor so that they maximize the likelihood that students will generalize to untaught responses.

Summary

Although significant progress has been made in understand- ing the underlying mechanisms that contribute to read- ing disabilities in young monolingual students, it remains unclear how reading-specific disabilities can be meaning- fully discriminated from difficulties experienced when learn- ing to speak and understand English as a second lan- guage. Relatively recently, researchers have begun to explore cross-linguistic relationships between first language (L1) phonological-processing abilities and second language (L2) word-reading skills for young ELs. This research, together with the growing body of new longitudinal studies of reading development in young ELs, several of which are represented in this special issue, may offer insights into how—or if—we should discriminate young English learners from struggling readers whose home language is English. This research also might provide an empirical basis for applying or modifying those instructional strategies that are proving effective in pre- venting reading failure for many native English speakers.

The present study is part of an ongoing longitudinal re- search project that aims to extend our theoretical and prac- tical understanding of the role of phonological-processing abilities for young students acquiring second-language read- ing skills (Gerber & English, 2003). As part of this larger longitudinal project, we report here the effects of an in- tervention using our Core Intervention Model for a group of Spanish-speaking kindergartners who performed poorly on Spanish as well as English phonological tasks. Be- cause students were not familiar with English on entering kindergarten, we conducted the kindergarten intervention in Spanish. In first grade, when English phonics instruction be- gan and when students had learned enough English to be instructed in English, we conducted a new intervention in English. This article will present results to address the fol- lowing questions.

1. Is a brief intervention based on our Core Intervention Model sufficient to improve performance on English word-reading tasks by the 20 percent of English learn- ers who perform most poorly on bilingual measures of phonological-processing skills?

2. Do these at-risk English learners trained in phonolog- ical skills catch up to their not-at-risk peers over time?

METHODS

The present study was part of an ongoing longitudinal study, La Patera, of English word-reading acquisition by Spanish- speaking students who began school as English learners. In 2000–2001, La Patera recruited three elementary school dis- tricts in California that enrolled primarily Latino students for whom English was a second language.1 Latino students as a percent of school enrollment was 71 percent, 81 percent, and 84 percent in these districts; 43 percent, 49 percent, and 40 percent of students, respectively, were considered by the districts to have limited English proficiency. Over 70 per- cent of students in each district received free or reduced lunch. In a survey of the families of students, 75 percent reported family incomes below $29,000. Second-grade read- ing performance in these districts on statewide, standardized tests was markedly below expectations, particularly for stu- dents designated by districts as not being English proficient. An informal survey of teachers revealed that no consistent early reading curriculum was in use across districts, schools within districts, or teachers within schools. After receiving consent from schools, teachers, and parents, we initiated the longitudinal study with data from kindergartners in 23 intact classrooms. Our final kindergarten sample included data from 377 students.

As part of this longitudinal study, we identified students as being at risk in kindergarten based on bilingual measures of phonological-processing skills. In first grade, we included measures of word reading as part of our “risk” criteria. Each year of the project we planned an instructional intervention based on a Core Intervention Model for students at risk. Our purpose was to see if relatively brief, supplemental instruction would close the gap in English word reading between students considered at high risk and their higher performing peers.

Participants

To identify students at risk in the larger research project from which this study is extracted, we first identified 20 percent of our entire longitudinal sample (N = 377) that were low- est performing on a series of bilingual tests of phonological skills. To do this, we created a selection score constructed as the sum of unit-weighted factor scores derived from a principal components analysis of a battery of English and Spanish measures (see Gerber & English, 2003). Specifi- cally, our selection score was constructed from three factors (components) extracted from English and Spanish versions of 15-item phonological-awareness tasks. Tasks loaded on fac- tors representing cognitive demands and not language of task.2 For example, all four English and Spanish rime and

GERBER ET AL.: EFFECTS OF SMALL-GROUP INTERVENTION 243

onset tasks loaded together. The same was true for the two segmentation and two final phoneme tasks, respectively.

We were also concerned with the practical distinction be- tween those students from our entire sample presumed to be at risk for poor reading outcomes based on test performance and those students for whom teachers had the greatest concern about likely reading success. Therefore, we consulted with classroom teachers to give them an opportunity to agree, dis- agree, or substitute another student from their classrooms for whom they had greater academic concerns. Teachers agreed that 80 percent of the students who would have been iden- tified by test alone were at highest risk. These students be- came part of the Intervention Group. We permitted teachers to substitute their preferences to complete the target num- ber of at-risk students (i.e., 20 percent of the cohort). All the teacher-nominated students performed at least in the bottom 40 percent on phonological assessments.

By allowing teachers’ perceptions to enter into selection, we traded a degree of experimental control, that is, risk oper- ationally defined in terms of phonological measures, for eco- logical validity—teachers’ formative judgments about stu- dents they teach. As we report below, this decision might have been consequential to outcomes.

For this article, we report results for 43 students for whom we have complete kindergarten and first-grade data. Students in our final risk group came from five schools and 15 class- rooms. These students received interventions (INT) to accel- erate growth on English word-reading measures. As a nonin- tervention comparison group (NON), we randomly identified in each of these classrooms a sample of the remaining, usu- ally better-performing students (i.e., those whose selection score did not fall into the risk group). In this report, INT and NON refer only to group assignment of students in kinder- garten. At the end of kindergarten, we had complete data for 37 INT (54 percent female) and 45 NON (31 percent female) students, respectively. Of these students, we had 28 INT and 15 NON students available for analysis at the end of first grade.

When we looked at this final sample to see how students in the remaining first-grade groups had been categorized by kindergarten testing versus teacher judgment, we found that 23 of the 28 INT students (80 percent) had been selected jointly for intervention by test performance and teacher judg- ment. The remaining five INT students were selected only by teachers, but were in the bottom 40 percent on test per- formance criteria. Of the 15 NON students remaining at the end of first grade, there were six students who were nomi- nated in kindergarten either by test criteria (two) or by teacher judgment (four) alone. The researchers and the kindergarten teachers had previously agreed that these students would be on a wait list and would not receive intervention. The re- maining nine NON students at the end of first grade were not considered high risk by either kindergarten test criteria or teacher judgment.

The two groups of students were, on average, five years, six months of age (SD = 4.7 months) at the time of initial testing. Available school data allowed us to confirm that all except three students (one in INT and two in NON) were Latino/Hispanic. Although these three students’ ethnicity

could not be confirmed through school records, all were iden- tified as ELs by their teachers.

Pre- and Posttest Assessments

All students were assessed individually twice a year by teams of fluently bilingual undergraduates who were trained and su- pervised by graduate research assistants to perform all assess- ments. Students in both INT and NON groups were assessed before and after interventions each year for a total of four data points. We conducted preintervention assessments dur- ing the fall quarter of the academic year and postintervention assessments immediately following intervention completion in the spring. Assessments were conducted on school sites in the quietest available spaces. Most of the tests were individu- ally administered in three 20-minute sessions. Different test sequences, alternating between English and Spanish, were randomly assigned to students; however, we avoided sequen- tial administration of Spanish and English versions of the same task.

We always asked students about their language prefer- ence for speaking with the assessors. In addition, though, each assessor determined the dominant language by asking the student a series of questions such as: What language do your parents speak? Do you speak English? Do you speak Spanish? Which language would you like to speak today? If dominant language was unclear, the tester gave directions in both English and Spanish. If students’ behaviors led as- sessors to doubt expressed preferences, assessors provided test instructions and assistance in the language that seemed better understood. This occurred only a few times. Once an administration language was established, all test instructions, regardless of the language of test content, were given in the language that students understood best.

Assessors tallied scores and initialed record forms. These were then checked and rechecked by at least two other stu- dents, including a doctoral research assistant. Final data were entered into a database in which students were identified only by codes provided by the school districts. Data entry was spot- checked multiple times for entry errors.

Interventions

Intervention was considered a supplement to instruction stu- dents received in their respective classrooms. The same bilin- gual undergraduates who served as assessors were trained to use our Core Intervention Model to provide small-group di- rect instruction in Spanish to INT students. In first grade, intervention was presented in the students’ language of in- struction, which was English for all but two classrooms. Only 14 students continued to receive intervention instruction in Spanish in first grade.

CIM training consisted of a special two-hour workshop with modeling of direct instruction techniques and explana- tion of a desired scope and sequence of skills to cover for each session. Following this initial training, tutors engaged in su- pervised practice at a nonproject school site that also had

244 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs

a majority of ELs. When tutors demonstrated satisfactory mastery of the techniques by executing the critical teaching behaviors in the time allocated for each of the targeted skills (e.g., rime detection), they were permitted to begin interven- tions in the field with project students.

To maintain fidelity, interventions were monitored in the field by senior researchers or doctoral assistants who provided contingent feedback to the tutors to encourage desired inter- vention components and, when necessary, to correct tech- nique or pacing. Ten half-hour intervention sessions were scheduled for targeted students in each of the classrooms. Intervention training, monitoring of intervention tutors, and number and duration of intervention sessions were repeated in the same manner in first grade.

As previously indicated, the CIM focused on implemen- tation of direct instruction techniques, including an explicit correction procedure, the Correction Staircase, whenever stu- dents made errors or failed to respond. We employed these methods of instruction throughout each activity with students. In kindergarten, activities focused primarily on building stu- dents’ early phonological awareness skills of onset and rime. The goal of each activity was to increase students’ under- standing of these skills through multiple opportunities to re- spond to structured prompts. Tutors provided students with ongoing feedback to motivate and scaffold their responses, all through simple and fun activities.

During each activity, tutors prompted rapid and multiple individual and whole-group responses. Questions were re- peated several times to verify that appropriate responses were reliable. Tutors were guided by a list of skill objectives, some simple materials (e.g., pictures or chips to illustrate onset and segmentation), and sample prompts to illustrate each tar- geted phonological skill. Tutors modeled the skill, prompted response to a “supply”-type question, and assisted students to correct or practice the skill using the Correction Staircase and additional example items.

For instance, the objective of one lesson was to have students recognize phonemic similarities between words by identifying words with similar onsets. Necessary materials included picture cards of specific foods. Tutors reviewed pic- tured item names and modeled the activity. Later, tutors asked students to choose a pictured item that did not have the same beginning sound as remaining items (e.g., manzana, leche, melón). In a rime lesson, students created words that rhymed with a given word. Tutors showed students a pictured item and asked them to think of words that rhymed with that item; for example, “¿Qué suena como gol?” (What sounds like gol?). Students were later encouraged to make up words that rhymed with pictured items. Tutors first modeled making up rhyming words for given items; for instance, “¿Rol suena como gol?” (Does rol sound like gol?). In a phoneme segmentation ac- tivity, tutors created three-phoneme words using letter cards while showing pictures of the words to ensure understanding. After reviewing sample words and the task with considerable modeling, tutors asked students to identify sounds in words until words were completely segmented (e.g., /s/ /u/ /n/ for “sun”).

In a phoneme-blending task (used only in first grade), tu- tors used a selection of pictures of three- to four-phoneme words from a small bag. These pictures would be reviewed with students at the onset of the activity. Tutors orally pre-

sented students with individual phonemes for one of the pic- tured words hidden in the bag (e.g., /h/ /a/ /t/ for hat).

Background Instruction

Students who did not receive intervention received only the instruction normally provided by their teachers. Participat- ing districts, schools, and teachers provided entirely different instructional programs in the background during this study. There was no discernable consistency in programs, materials, or methods across any of these units of analysis. Instruction was provided in English in all but two classrooms. Two stu- dents came from one of these “primary language support” classrooms. Two students from the same classroom were in the INT group.

Dependent Measures

We used three tasks in both English and Spanish to measure phonological awareness (PA) in kindergarten and first grade: rime and onset detection and phoneme segmentation. Rime and onset tasks contained 15 items each, with three practice items in the kindergarten battery. Practice items were used to instruct as well as to introduce the task. Rime and onset tasks included three pictures per item. The assessor asked stu- dents which of two words either rhymed or started with the same sound as the prompt. In the segmentation task, students were shown a visual representation of a target item. Then they were asked to verbally segment into its individual phonemes the word provided by the assessor. Reliabilities (Cronbach al- pha) for the final forms of each of these tasks were as follows: Rime (English α = 0.77, Spanish α = 0.77), Onset (English α = 0.74, Spanish α = 0.65), and Segmentation (English α = 0.89, Spanish α = 0.92). To increase internal reliability, an additional five items were added in first grade.

As previously noted, rime and onset tasks in both lan- guages loaded together, as did both language versions for the segmentation task. Based on this structure of loadings and empirical research suggesting a developmental relationship among these tasks (e.g., Christensen, 1997), we created two composite measures of all Spanish and English tasks. One, early phonological awareness (EPA), derived from combina- tion of all rime and onset measures. The other, late phonolog- ical awareness (LPA), was formed by combining both lan- guage measures of segmentation.

In addition to the phonological measures, we also adminis- tered the English Word Attack (WA) and Letter-Word Identi- fication (WID) subtests from the Woodcock-Johnson Tests of Achievement III (2000). On the Word Attack subtest, students decoded a series of letters and nonwords. In the Letter-Word Identification subtest, students identified a series of letters by name and read (recognized) a list of real words.

RESULTS

Kindergarten students who did (INT) and did not (NON) re- ceive kindergarten intervention made statistically significant gains between the beginning (pretest) of kindergarten and

GERBER ET AL.: EFFECTS OF SMALL-GROUP INTERVENTION 245

TABLE 1 Results of Repeated-Measures Analysis of Variance for Groups ×

Time of Testing Within Subjects Between Subjects

Withing Subjects Between Subjects

F df p F df p

English Onset Main effect 84.99 3 0.00 2.78 1 0.10 Interaction 1.61 3 0.19

Spanish Onset Main effect 52.30 3 0.00 0.45 1 0.51 Interaction 0.54 3 0.65

English Rime Main effect 83.87 3 0.00 7.12 1 0.01 Interaction 1.88 3 0.14

Spanish Rime Main effect 134.39 3 0.00 1.96 1 0.17 Interaction 2.73 3 0.01

English Segmentation Main effect 120.59 3 0.00 1.77 1 0.19 Interaction 1.02 3 0.39

Spanish Segmentation Main effect 76.35 3 0.00 0.23 1 0.63 Interaction 2.30 3 0.08

Word Identification Main effect 72.14 3 0.00 1.44 1 0.24 Interaction 0.12 3 0.95

Word Attack Main effect 68.55 3 0.00 1.16 1 0.29 Interaction 1.31 3 0.28

Early PA Main effect 280.48 3 0.00 2.64 1 0.11 Interaction 0.91 3 0.44

Late PA Main effect 117.28 3 0.00 0.69 1 0.41 Interaction 1.67 3 0.18

the ending (posttest) of first grade. These results are shown in Tables 1–5.

At kindergarten posttest, INT students had substantially closed the performance gap with their higher-performing peers, and they had closed the gap on many measures by the end of first grade. In Table 1, we report results from a repeated measures, 2 (groups) × 4 (time of testing) analysis of variance for these measures. We present descriptive statistics for each of the four observation periods (Tables 2–5) between begin- ning kindergarten and end of first grade, along with results of posthoc, t-tests for group differences on each occasion. In performing posthoc tests, we were sensitive to the potential

TABLE 2 Beginning Kindergarten Scores

Intervention Nonintervention

Measure M SD N M SD N

English Onset 4.14 4.26 28 5.20 5.24 15 Spanish Onset 6.04 4.25 28 4.86 4.16 15 English Rime 4.39 3.79 28 8.00 3.96 15 Spanish Rime 4.42 3.77 28 7.80 4.50 15 English Segmentation 0.11 0.55 28 0.93 1.94 15 Spanish Segmentation 0.18 0.55 27 2.00 4.57 15 Early PA 19.00 9.12 28 25.87 11.71 15 Late PA 0.29 0.71 27 2.93 6.31 15 Word ID 5.96 2.84 28 8.93 3.20 15 Word Attack 0.00 0.00 28 0.00 0.00 15

Note. Underscoring represents statistically significant differences (p <

0.05). Word ID represents Word Identification, PA represents Phonological Awareness.

TABLE 3 Ending Kindergarten Scores

Intervention Nonintervention

Measure M SD N M SD N

English Onset 10.04 2.25 28 9.53 2.56 15 Spanish Onset 9.61 2.48 28 9.73 2.89 15 English Rime 9.61 2.97 28 9.93 2.71 15 Spanish Rime 8.96 3.09 28 10.67 2.74 15 English Segmentation 3.44 3.60 27 3.43 3.98 14 Spanish Segmentation 3.59 4.38 27 1.86 4.04 14 Early PA 38.21 6.47 28 39.87 7.58 15 Late PA 7.04 7.02 27 5.29 7.15 14 Word ID 10.43 4.35 28 12.50 4.99 14 Word Attack 0.25 0.84 28 1.07 1.94 14

Note. Underscoring represents statistically significant differences (p <

0.05). Word ID represents Word Identification, PA represents Phonological Awareness.

TABLE 4 Beginning First-Grade Scores

Intervention Nonintervention

Measure M SD N M SD N

English Onset 12.68 3.14 28 13.67 3.04 15 Spanish Onset 12.50 3.53 28 12.73 3.53 15 English Rime 13.46 3.13 28 15.20 3.71 15 Spanish Rime 15.25 3.91 28 15.33 3.70 15 English Segmentation 7.82 5.20 28 9.50 5.81 14 Spanish Segmentation 7.29 5.77 28 8.79 6.30 14 Early PA 53.89 8.55 28 56.93 9.85 15 Late PA 15.11 10.53 28 18.29 11.18 14 Word ID 16.46 6.36 28 18.60 5.36 15 Word Attack 4.93 3.20 28 4.93 2.46 15

Note. Underscoring represents statistically significant differences (p <

0.05). Word ID represents Word Identification, PA represents Phonological Awareness.

TABLE 5 Ending First-Grade Scores

Intervention Nonintervention

Measure M SD N M SD N

English Onset 14.25 3.46 28 16.93 2.52 15 Spanish Onset 13.86 2.89 28 12.87 3.54 15 English Rime 15.14 4.04 28 17.07 2.84 15 Spanish Rime 17.04 2.24 28 16.67 2.94 15 English Segmentation 14.07 4.91 27 14.60 3.20 14 Spanish Segmentation 14.61 4.90 27 13.00 4.46 14 Early PA 60.63 8.88 27 63.53 8.80 15 Late PA 29.04 9.16 27 27.60 7.39 15 Word ID 28.65 13.95 28 30.07 10.84 14 Word Attack 8.33 6.63 28 10.73 5.74 14

Note. Underscoring represents statistically significant differences (p <

0.05). Word ID represents Word Identification, PA represents Phonological Awareness.

246 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs

0

10

20

30

40

50

60

70

Pre K Post K Pre 1st Post 1st

N u

m b

er C

o rr

ec t

INT NON

FIGURE 1 Change in early phonological awareness.

risk of test-wise error, but have been conservative in inter- preting probabilities for observed mean differences. As an aid to visualizing these data, we present graphs in Figures 1 though 4 that show relative trajectories of growth across the four observations periods for EPA, LPA, Word Identification (i.e., reading real words), and Word Attack (i.e., decoding pseudowords), respectively. Examination of tabled data for each observation period aids in the interpretation of the graphical representation of growth, especially by disaggre- gating EPA and LPA into their separate English and Spanish components.

Group Differences Over Time

Generally speaking, these data indicate that INT students caught up with their better-performing peers by the end of

0

10

20

30

40

50

60

70

Pre K Post K Pre 1st Post 1st

N u

m b

er C

o rr

ec t

INT NON

FIGURE 2 Change in late phonological awareness.

first grade on all measures (except for English Onset) despite statistically significant differences in performance at the be- ginning of kindergarten.

Early PA

Repeated-measures analyses (see Table 1) to estimate the reliability of observed changes found statistically signifi- cant main effects on all measures of Early PA (all at ps < 0.01) for time of testing (ES = 2.0). Interaction effects for groups over time were found for Spanish Rime only ( p < 0.01), indicating greater gains for the INT group com- pared to the NON group despite the fact that the NON group (M = 56.9) continued to perform better at the end of first grade compared to INT (M = 53.9) on measures of Early PA.

GERBER ET AL.: EFFECTS OF SMALL-GROUP INTERVENTION 247

0

5

10

15

20

25

30

35

Pre K Post K Pre 1st Post 1st

N u

m b

er C

o rr

ec t

INT NON

FIGURE 3 Change in Word Identification.

Late PA

Repeated-measures analyses found statistically significant main effects for time on Late PA measures for both NON and INT groups ( ps < 0.01). That is, both groups showed significant gains in performance between kindergarten and first grade. We did not find statistically significant interaction effects for groups over time. In addition, there were no sta- tistically significant main effects differences for groups for either individual or composite Late PA scores.

Word Reading

Both groups’ gains over time on word-reading measures were statistically significant. However, there were no statistically significant between-group or interaction effects. Although

0

5

10

15

20

25

30

35

Pre K Post K Pre 1st Post 1st

N u

m b

er C

o rr

ec t

INT NON

FIGURE 4 Change in Word Attack.

NON students knew more words as measured by Word Iden- tification at the beginning of the study, no statistically signif- icant differences remained between groups by the end of first grade. At the beginning of kindergarten, both groups were similar in their relative inability to decode pseudowords as measured by the Word Attack task. Thereafter, though, they gained similarly over time, and differences in performance by the end of first grade were negligible and still statistically insignificant.

Group Differences at Each Time of Testing

Although the ultimate level of performance is important, it is also important to examine patterns of growth (see Figures 1– 4). For INT students to “catch up,” their rate of improvement, at some point at least, has to be greater than that of their

248 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs

initially-higher-achieving peers. Our results show, generally, that INT students gained skills at a rate similar to that of their higher-achieving peers. However, between the beginning and end of first grade, INT students gained ground (see Figures 1– 4 and Tables 4 and 5).

Early PA Skills

Posthoc analyses at each time of testing show NON students outperforming INT students on the EPA composite variable, due to differences on the Rime measures, at the beginning of kindergarten, but only on English Onset by the end of first grade. End of first-grade EPA and its component measures for the INT group indicated gains similar to those of the NON group from beginning first grade on all measures of Early PA (see Table 5).

Late PA Skills

NON students outperformed INT students at the beginning of kindergarten on late-developing phonological skills (LPA). Kindergarten pretest means for LPA were 2.93 and 0.3 for NON and INT groups, respectively. At the end of kinder- garten, following intervention, INT students passed NON stu- dents, although these were not statistically significant group differences (see Table 3). Average gain between kindergarten and first grade on all Late PA measures was greater for the INT (13.9) compared to the NON group (9.3) and, by the end of first grade, all INT and NON between-group differences disappeared in all measures and INT students were scoring higher on average (see Table 5).

Word-Reading Skills

There were statistically significant differences on English Word Identification (WID) favoring NON students at the be- ginning of kindergarten (NON M = 8.9, INT M = 6.0) (see Table 2). However, no statistically significant differences ex- isted for English Word Attack (WA), a very difficult skill for both groups at that time. However, although NON stu- dents gained more on WID between the beginning of kinder- garten and the beginning of first grade, the INT students made greater gains on WA. However, gains made by the end of first grade favored the NON group on both measures. Neverthe- less, no statistically significant group differences were found for either measure by the end of first grade.

DISCUSSION

Taken as a whole, we interpret these data as positive evi- dence that supplemental instruction can be effective in sup- porting learning by ELs identified by bilingual phonologi- cal skills tests and teacher judgment to be at high risk for English-reading failure. In this study, kindergarten INT stu- dents, selected by testing and teacher judgment from the bot- tom performing 20 percent of our larger longitudinal sample,

substantially closed the initial gap in performance on phono- logical and word-reading tasks between them and their better- performing peers. Therefore, we believe these data support use of brief, but intensive, interventions based on direct in- struction techniques for similar, high-risk students.

This said, we must acknowledge that our intervention pro- duced only a relatively weak effect and may have been abetted by background instruction and maturation. However, several factors increase confidence in our conclusions. First, extant research suggests that initially low-performing students will not keep pace or catch up with their higher-performing peers without effective intervention (Francis, Shaywitz, Stuebing, Shaywitz, & Fletcher, 1996). Second, we have conducted sub- sequent research (e.g., Leafstedt, Richards, & Gerber, this is- sue; Richards, 2004) that substantially replicates these find- ings with more extensive interventions.

Limitations and Implications

There are significant limitations to our study and caution should be exercised in interpreting our results. However, these limitations offer insights into the problem of identifying and remediating “risk” for young English learners. We will dis- cuss these limitations and their implications below.

Experimental Control

The study was not strictly experimental, making inferences concerning treatment effects more difficult. We decided as an ethical and economic consideration not to randomly assign students thought to be at high risk to a nontreatment con- trol condition. Instead, we adopted a strategy based on the evidence-based presumption that high-risk students could be expected not to catch up with higher-performing peers with- out intervention (Francis et al., 1996). Effectiveness of sup- plemental instruction was to be demonstrated if our interven- tion promoted growth that was equal to or better than growth for the comparison peers. It would have been better, in ret- rospect, to create an alternative treatment group and applied random assignment. We intend to attempt this approach with a new cohort.

Identification Criteria

Also as a matter of ecological validity, we permitted teacher concerns for reading failure—another kind of measurement, really—to enter into our test-based selection process. It is hard to avoid the fact that teachers identify risk by comparing students’ responsiveness to instruction to other students in their classrooms. Whatever criteria teachers use in making these judgments clearly overlapped with, but was not identical to, the phonological performance criteria we used to select our initial risk pool.

In a sense, students selected by prereading skill testing are at risk only in population (i.e., probabilistic) terms and may or may not be proven to be responsive to instruction. Teachers’ judgments, however, may incorporate not only an assessment

GERBER ET AL.: EFFECTS OF SMALL-GROUP INTERVENTION 249

of performance against a set of norms, but also an assessment of how specific students historically have failed to learn what teachers have attempted to teach. Interventions with our kind of test-nominated group, in fact, lead you to identify the es- tablished kind of risk present in such a teacher-nominated group. By allowing teacher-identified students into our in- tervention group, we may have inadvertently weakened the likelihood of treatment effects.

Length of Intervention

Another significant limitation was in the relative brevity of our intervention. Effects are found in longitudinal studies that deliver at least 10 hours of secondary, or supplemental, in- struction for students who initially perform poorly on phono- logical tasks, and larger effects require even more time (e.g., see Torgesen, 2002). In the current study, we delivered only three hours spread over approximately nine sessions. There- fore, it is quite possible that lengthier interventions would build on the small effects observed in this study. This is, in fact, what we have found in subsequent work (e.g., Leafstedt et al., this issue; Richards, 2004).

Contingency in Instruction

Most of the extant literature on early intervention for high-risk students concentrates on a guided curriculum and teaching by specialists, but has been relatively silent on what spe- cific instructional behaviors are, or should be, employed. In our study, we concentrated on direct instruction methods de- signed both to be effective and maximally efficient for class- room teachers or aides. Although we had planned a specific “scope and sequence” of phonological skills, that is, a mini- curriculum, we soon realized that there is an inherent conflict between delivering a mastery-oriented instruction that un- folds contingently for individual learners, on one hand, and attempting to keep to a time schedule for meeting a specific sequence of performance targets, on the other. Simply put, our instructional model required that teachers shift to new skills (e.g., phoneme segmentation) only in strict response to observable evidence that individual students were fast and accurate at a precursor skill (rime detection).

The net effect for already struggling students was to in- crease the amount of time that was spent on early skills (e.g., rime, onset) and decrease the amount of time left to instruct later skills (e.g., segmentation, blending). As long as the pe- riod of intervention was relatively fixed at 9–10 sessions, it was predictably impossible to respond to students individu- ally and simultaneously meet the same instructional targets for each. We think this conundrum is the core of the inferred “unresponsiveness” of some students even in small-group in- struction.

Background Instruction

One final limitation of this study relates to the lack of true experimental design. We also had no control over the back-

ground instruction that students were receiving and, although we were satisfied that it would produce no systematic con- founds, neither could we account for any of its specific effects. Teachers, after all, notice students who are not faring well and some—one would hope, most—modify their instruction to improve the situation. Extant research with monolingual students has increased the power of the independent variable (i.e., the intervention), in part, to overcome effects of back- ground instruction (see Torgesen, 2002, for discussion of re- search on very intensive, supplemental reading interventions by Torgesen, Vellutino, Foorman, Vaughn, and others).

Here, too, lies another conundrum for future research to address. If background, or baseline, instruction initiates a multitiered intervention system by its relative failure with some students, interventions will have to adjust their power in response. As background instruction improves, the students it nominates, by definition, will be more resistant to instruction. Future longitudinal research will need to describe and incor- porate in analyses those specific variations in background instruction that influence power of any particular multitiered intervention design.

Conclusions

Despite these limitations, we believe that this study begins to establish a scientific basis for teachers to differentiate among young English learners and a general scheme for provid- ing supplemental instruction for those who perform most poorly on measures monitoring acquisition of phonologi- cal and word-reading skills. The approach described in this study is consistent with models of multitiered intervention strategies for monolingual students, similar to those promoted by the Reading First initiative. Also, our Core Intervention Model is an example of what teachers might realistically be expected to implement with English learners in their own classrooms.

ACKNOWLEDGMENTS

The research reported in this article was supported in part by Grant T288S000327 from the Office of English Language Acquisition in the U.S. Department of Education. The opin- ions expressed do not necessarily reflect those of OELA or the USDOE.

The authors gratefully acknowledge the contributions to this research of Alexis Filippini, Ingrid Salamanca, Stacey Kyle, Maria Denney, Lee Swanson, Monica Ruz, Leilani Saez, Mayra Benitz, Elia Bustamante, Dulce Garcia, America Saucedo, Sara Fernandez Medina, and the teachers and prin- cipals from three participating school districts in southern California. The authors are also grateful to the anonymous reviewers and Joanne Carlisle, co-editor of LDRP, for their thoughtful criticisms and helpful recommendations.

NOTES

1. It is worth noting that three nearby districts declined to participate when they learned that we intended to

250 SPECIAL SERIES: READING RISK AND INTERVENTION FOR YOUNG ELs

conduct our kindergarten interventions in Spanish. Fol- lowing passage of California’s Proposition 227, bilin- gual education was eliminated. Despite our explana- tion that we did not intend to teach students to read in Spanish, but only to assess their phonological- processing abilities in their first language, these dis- tricts all explained their decision in terms of what they perceived to be a contentious political atmosphere fol- lowing their respective implementations of the new law. We do not believe that these reasons have any bearing on our testing results.

2. Subsequent principal components analyses for first- and second-grade performance yielded similar results.

REFERENCES

Alanis, I., Munter, J., & Tinajero, J. V. (2003). Preventing reading failure for English language learners: Interventions for struggling first-grade L2 students. NABE Journal of Research and Practice, 1, 92–109.

Ball, E. W., & Blachman, B. A. (1991). Does phonemic segmentation training in kindergarten make a difference in early word recognition and developmental spelling? Reading Research Quarterly, 26, 49– 66.

Bialystok, E., Majumder, S., & Martin, M. M. (2003). Developing phono- logical awareness: Is there a bilingual advantage? Applied Psycholin- guistics, 24, 27–44.

Bradley, L., & Bryant, P. (1983). Categorizing sounds and learning to read: A causal connection. Nature, 301, 419–421.

Christensen, C. A. (1997). Onset, rhymes, and phonemes in learning to read. Scientific Studies of Reading, 1(4), 341–358.

Cisero, C. A., & Royer, J. M. (1995). The development and cross-language transfer of phonological awareness. Contemporary Educational Psy- chology, 20, 275–303.

Committee on the Prevention of Reading Difficulties in Young Students. (1998). Preventing reading difficulties in young students. Washington, DC: National Academy Press.

Durgunoglu, A. Y., Nagy, W. E., & Hancin-Bhatt, B. J. (1993). Cross- language transfer of phonological awareness. Journal of Educational Psychology, 85, 453–465.

Elbaum, B. E., Vaughn, S., Hughes, M. T., & Moody, S. W. (1999). Grouping practices and reading outcomes for students with disabilities. Excep- tional Students, 65, 399–415.

Fletcher, J. M., Foorman, B. R., Boudousquie, A., Barnes, M. A., Schatschneider, C., & Francis, D. J. (2002). Assessment of reading and learning disabilities: A research-based intervention-oriented approach. Journal of School Psychology, 40, 27–63.

Fletcher, J. M., & Lyon, G. R. (1998). Reading: A research-based approach. In W. M. Evers (Ed.), What’s gone wrong in America’s classrooms? (pp. 49–90). Stanford, CA: Hoover Institute Press.

Foorman, B. R., & Torgesen, J. (2001). Critical elements of classroom and small-group instruction promote reading success in all students. Learn- ing Disabilities Research & Practice, 16(4), 203–212.

Francis, D. J., Shaywitz, S. E., Stuebing, K. K., Shaywitz, B. A., & Fletcher, J. M. (1996). Developmental lag versus deficit models of reading dis- ability: A longitudinal, individual growth curves analysis. Journal of Educational Psychology, 88, 3–17.

Fuchs, D., Fuchs, L. S., Thompson, A., Al-Otaiba, S., Yen, L., Yang, N. J. et al. (2002). Exploring the importance of reading programs for kindergart- ners with disabilities in mainstream classrooms. Exceptional Children, 68, 295–311.

Fuchs, L. S. (2002). Three conceptualizations of “treatment” in a re- sponsiveness to treatment model. In R. Bradley, L. Danielson, & D. Hallahan (Eds.), Identification of learning disabilities: Research to practice (pp. 521–529). Mahwah, NJ: Lawrence Earlbaum Associates.

Fuchs, L. S., & Vaughn, S. (2003). Redefining learning disabilities as in- adequate response to instruction: The promise and potential problems. Learning Disabilities Research & Practice, 18(3), 137–146.

Gerber, M. M., & English, J. P. (2003). Project La Patera. Final report. Santa Barbara, CA: Center for Advanced Studies of Individual Differences, University of California.

Geva, E., Wade-Woolley, L., & Shany, M. (1997). Development of reading efficiency in first and second language. Scientific Studies in Reading, 1, 119–144.

Goswami, U., & Bryant, P. (1990). Phonological skills and learning to read. Hillsdale, NJ: Erlbaum.

Gottardo, A., Yan, B., Siegel, L. S., & Wade-Woolley, L. (2001). Fac- tors related to English reading performance in students with Chi- nese as a first language: More evidence of cross-language transfer of phonological processing. Journal of Educational Psychology, 93, 530– 542.

Justice, L. M., & Pullen, P. (2003). Promising interventions for promoting emergent literacy skills: Three evidenced based approaches. Topics in Early Childhood Special Education, 23(3), 99–113.

Leafstedt, J. L. (2002). Crossover of phonological processing skills: A study of Spanish-speaking students in two instructional settings. Unpublished doctoral dissertation. University of California, Santa Barbara.

Leafstedt, J. M., Richards, C. R., & Gerber, M. M. (this issue). Effective- ness of explicit phonological awareness instruction for at-risk English learners. Learning Disabilities Research & Practice.

Lindsey, K. M., Manis, F. R., & Bailey, C. E. (2003). Prediction of first grade reading in Spanish-speaking English language learners. Journal of Educational Psychology, 95, 482–494.

Lyon, G. R. (2003). A definition of dyslexia. Annals of Dyslexia, 53, 1– 14.

Lyon, G. R., Fletcher, J. M., Shaywitz, S. E., Shaywitz, B. A., Torge- sen, J. K., Wood, F. B. et al. (2001). Rethinking learning disabili- ties. In C. E. Finn, A. J. Rotherman, & C. R. Hokanson (Eds.), Re- thinking special education for a new century (pp. 259–287). Washing- ton, DC: Thomas B. Fordham Foundation and the Progressive Policy Institute.

National Reading Panel. (2000). Teaching students to read: Summary re- port. Washington, DC: National Institute of Child Health and Human Development.

O’Connor, R. E., Jenkins, J. R., & Slocum, T. A. (1995). Transfer among phonological tasks in kindergarten: Essential instructional content. Journal of Educational Psychology, 87, 202–217.

Pikulski, J. J. (1994). Preventing reading failure: A review of five effective programs. Reading Teacher, 48(1), 30–39.

Rashotte, C. A., MacPhee, K., & Torgeson, J. K. (2001). The effectiveness of a group reading instruction program with poor readers in multiple grades. Learning Disabilities Quarterly, 24(2), 119–134.

Riccio, C. A., Amado, A., Jiménez, S., Hasbrouck, J. E., Imhoff, B., & Denton, C. (2001). Cross-linguistic transfer of phonological process- ing: Development of a measure of phonological processing in Spanish. Bilingual Research Journal, 25, 417–437.

Richards, C. R. (2004). Response to instruction: A study of phonological awareness instruction for kindergartener English learners. Unpub- lished dissertation, University of California, Santa Barbara.

Slavin, R. E., & Cheung, A. (2003). Effective reading programs for English language learners: A best-evidence synthesis. Baltimore, MD: Johns Hopkins University, Center for Research on Effective Schooling for Disadvantaged Students.

Snow, C., Burns, S., & Griffin, P. (Eds.). (1998). Preventing reading difficulties in young students. Washington, DC: National Academy Press.

Stanovich, K. E. (1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 26, 7–29.

Swanson, H. E., Saez, L., Gerber, M. M., & Leafstedt, J. (2004). Literacy and cognitive functioning in bilingual and non-bilingual children at or not at risk for reading disabilities. Journal of Educational Psychology, 96, 3–18.

Swanson, H. L., & Hoskyns, M. (1999). Interventions for students with learning disabilities: A meta-analysis of treatment outcomes. New York: Guilford Press.

Torgesen, J. K. (2002). Lessons learned from intervention research in read- ing: A way to go before we rest. In R. Stanthorp & P. Tomlinson (Eds.), Learning & teaching reading (pp. 89–103). British Journal of Edu- cational Psychology Monograph Series II: Psychological Aspects of Education. Leicester, UK: British Psychological Society.

GERBER ET AL.: EFFECTS OF SMALL-GROUP INTERVENTION 251

Toregesen, J. K., Alexander, A., Wagner, R., Raschotte, C., Voeller, K., & Conway, T. (2001). Intensive remedial instruction for students with severe reading disabilities: Immediate and long-term outcomes from two instructional approaches. Journal of Learning Disabilities, 34, 33– 58.

Torgesen, J. K., Wagner, R., Rashotte, C., Rose, E., Lindamood, P., Conway, T. et al. (1999). Preventing reading failure in young students with phono- logical processing disabilities. Journal of Educational Psychology, 91, 579–593.

Vaughn, S., Bos, C. S., & Schumm, J. (2001). Teaching exceptional, di- verse, and at-risk students in the general education classroom (3rd ed.). Boston, MA: Allyn & Bacon.

Vaughn, S., Hughes, M. T., Moody, S. W., & Elbaum, B. (2001). Instruc- tional grouping for reading for students with learning disabilities: Im- plications for practice. Intervention in School and Clinic, 35, 131– 137.

Vellutino, F., Scanlon, D., Sipay, E., Small, S., Pratt, A., Chen, R. et al. (1996). Cognitive profiles of difficult-to-remediate and readily reme- diate poor readers: Early intervention as a vehicle for distinguishing between cognitive and experiential deficits as basic causes of specific reading disability. Journal of Educational Psychology, 88, 601–638.

Wagner, R. K., & Torgeson, J. K. (1987). The nature of phonological process- ing and its causal role in the acquisition of reading skills. Psychological Bulletin, 101, 192–212.

About the Authors

Michael M. Gerber is Professor of Education, Gevirtz Graduate School of Education, and Director of the Center of Advanced Studies of Individual Differences, at the University of California, Santa Barbara.

Terese Jimenez is an Assistant Professor of Education in the School of Education at Loyola Marymount University.

Jill Leafstedt is Assistant Professor of Education in the Education Program at the California State University, Channel Islands.

Jessica Villaruz is a doctoral candidate in educational leadership and organization in the Gevirtz Graduate School of Education at the University of California, Santa Barbara.

Catherine Richards in an Assistant Professor of Education at California State University, Long Beach.

Judy English is a Postdoctoral Researcher in the Gevirtz Graduate School of Education at the University of California, Santa Barbara.

All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.