First-Grade Latino Students' English-Reading Growth

First-Grade Latino Students' English-Reading Growth 23

First-Grade Latino Students' English-Reading Growth in All-English Classrooms Paul Neufeld Simon Fraser University Steven J. Amendum Jill Fitzgerald Karren M. Guthrie The University of North Carolina at Chapel Hill Abstract Two main questions were addressed in this study: (1) How does firstgrade Latino English-language learners' growth in English instructional reading level and selected word-level reading subprocesses (ability to read words in isolation, phonemic awareness, and phonics) compare to their monolingual native-English-speaking peers' growth?; and (2) Does first-grade Latino English-language learners' English reading growth (instructional reading level and selected word-level reading subprocesses) vary according to their oral English language abilities? Participants were 47 students in two first-grade classrooms-28 were Latino English-language learners, and 19 were monolingual nativeEnglish speakers. At each of two points in time-mid-year and end-ofyear-three reading measures were administered to all participants and an additional four oral-English measures were administered to the Latino participants. To address the first research question, repeated measures analyses of variance were performed, first using Instructional Reading Level as the dependent variable, then with follow-up analyses to examine growth in word-level sub-processes of reading. The second research question was addressed using repeated measures analyses of covariance. Main findings were that language status (Latino English learners versus monolingual native-English speakers) was not related to Instructional Reading growth or growth in word-level subprocesses of reading, and Overall English Oral Ability was not related to Instructional Reading Level growth, but was related to word-level reading sub-processes.

Rationale Over the past few decades the numbers of English-language learners attending schools in the United States have increased markedly.

24 Reading Research and Instruction, Volume 46, No. 1 * Fall 2006

For instance, between the 1990-1991 and 2000-2001 school years the percentage of students nationwide classified as "Limited English Proficient" increased by 105% (Kindler, 2002). During the same period total school enrollment increased by only 12%. Also, the percentage of Latinos living in the United States increased by 355% between 1970 and 2004 (Kasarda & Johnson, 2006). In 2000, Latino students represented 17% of nationwide enrollments in Kindergarten through Grade 12 (ERIC Clearinghouse on Urban Education, 2001). Moreover, roughly one-third of the Latino population is below 18 years of age, and estimates are that Latino school enrollment will increase by 25% by the year 2025 (ERIC Clearinghouse on Urban Education, 2001). Of particular relevance to the present study, there is some evidence of a significant gap between English-language learners' reading achievement and that of their native-English-speaking peers. For example, although Latino students, on average, had higher National Assessment of Educational Progress (NAEP) reading scores in 1999 than in the 1970s, the most recent NAEP data available for Latino students indicate that on average, their performance continues to remain substantially lower than Caucasian students of European heritage (U.S. Department of Education, National Center for Education Statistics, 1999). Since reading is a pivotal proficiency for academic success, it is crucial that researchers and practitioners alike attain rich understandings of factors involved in Latino students' English reading development. Learning to read in a new language surely introduces a range of complications and challenges, for both students and teachers. However, while practitioners and lay persons widely believe that optimal levels of oral English are necessary for learning to read in English, the precise role of oral English proficiency in English reading growth is not clear. To our knowledge, there are no comprehensive or "grand" formal theories to guide understandings and hypotheses about relationships among key constructs, variables, or principles that could be involved (cf. Fitzgerald, 2003). Indeed, even without considering the linkages between oral learning in a new language and reading in a new language, no such "grand" theory exists just for how individuals learn a new language (Mitchell & Myles, 2004). A further complicating factor in English-language-learner reading research is that neither oral language nor reading is a unitary construct. Instead, each is multidimensional. Oral language entails such dimensions as knowledge of vocabulary meanings, syntax, phonology, morphology,


and pragmatics. Reading involves such abilities as word recognition, comprehension, and fluency. Moreover, oral language and reading abilities are dynamic, varying in importance at different developmental stages (e.g., Chall, 1996; Fitzgerald & Shanahan, 2000). A well-developed understanding of English-reading developmental trajectories for young English-language learners, especially as compared to their native-English-speaking peers, is important as it could inform teachers' instructional decisions. If the trajectory revealed that certain subprocesses of reading tended to typically develop at particular stages, then teachers might especially emphasize those sub-processes during certain phases of instruction-as a developmental "booster." Understanding sound and word-level sub-process trajectories is likely to be critical for understanding young English-language learners' reading trajectories. Such sub-processes have been shown to have preeminent importance for monolingual native-English-speaking children (Adams, 1990; Siegel, 2003) and are likely to be equally important for English-language learners. Similarly, a robust depiction of how relationships between oralEnglish and English-reading abilities might change over time could significantly impact teachers' expectations for students' progress as well their instructional planning. For instance, suppose that on average, young English-language learners' oral-English proficiency was initially unrelated to their overall reading achievement, but at the end of two years, it related. One instructional implication might be that teachers could consider early immersion in English reading for young learners-because English reading development might not be dependent upon oral-English level. But over time teachers might pay increasing attention to English-language learners' oral-English vocabulary and comprehension development, because those might weigh heavily in maintaining a positive English reading trajectory. To date, relatively little is known about English reading growth or about the impact of oral-English proficiency upon English reading over time for young Latino English-language learners, especially for those in English immersion classrooms. Probably owing in part to the lack of a grand theory as well as to the multidimensional nature of the two constructs, learning oral English and learning to read in English-as-a-new language have been studied, to borrow a phrase from Mitchell and Myles (2004), as "modular phenomena," with enclaves of investigators addressing different aspects of each construct. Findings from the limited number of studies conducted on Englishlanguage learners' English reading achievement development in relation to


their monolingual English-speaking peers' achievement development are inconsistent. Some research focusing on children in kindergarten through second grade has documented how, across a school year, English-language learners can accomplish gains in reading achievement similar to those of their typically developing, native-English-speaking peers (Aruajo, 2002; Fitzgerald & Noblit, 1999, 2000; Hutchinson, Whiteley, Smith & Connors, 2003; Lesaux & Siegel, 2003; Manis, Lindsey & Bailey, 2004; Muter & Diethelm, 2001; Weber & Longhi-Chirlin, 2001). However, other studies have documented a substantial lag in the reading achievement levels of young Latino English-language learners in English immersion classrooms as compared to their native-English-speaking classmates (Fitzgerald & Noblit, 1999; Hutchinson et al., 2003; Neufeld & Fitzgerald, 2001). Further, limited research has been conducted on English-language learners' reading subprocess development (e.g., phonemic awareness, phonics knowledge, and word recognition). Many of the studies conducted in this area to date are case studies of individual learners. Among the specific subprocesses that have been studied are: word-recognition strategies (Fitzgerald & Noblit, 1999, 2000; Geva, Yaghoub-Zadeh & Schuster, 2000; Hutchinson et al., 2003; Lesaux & Siegel, 2003; Manis et al., 2004; Weber & Longhi-Chirlin, 2001), phonological awareness assessed in literacy contexts (Fitzgerald & Noblit, 1999, 2000; Geva et al., 2000; Lesaux & Siegel, 2003; Manis, et al., 2004; Muter & Diethelm, 2001; Neufeld & Fitzgerald, 2001), phonics knowledge (Aruajo, 2002; Fitzgerald & Noblit, 1999, 2000), comprehension (Fitzgerald & Noblit, 2000; Hutchinson et al., 2003; Lesaux & Siegel, 2003; Manis et al., 2004; Weber & Longhi-Chirlin, 2001), writing (Aruajo, 2002; Fitzgerald & Noblit, 1999; Neufeld & Fitzgerald, 2001; Weber & Longhi-Chirlin, 2001), and spelling (Lesaux & Siegel, 2003; Weber & Longhi-Chirlin, 2001). In general, findings from the studies suggest that low- and highperforming kindergarten, first, and second-grade English-language learners develop in ways that are highly similar to those of their nativeEnglish-speaking peers. While the relationship between the oral-English and Englishreading abilities of some groups of English-language learners has been examined to a limited extent (Durgunoglu, Nagy, & Hancin-Bhatt, 1993; Gottardo, 2002; Quiroga, Lemos-Britton, Mostafapour, Abbott & Berninger, 2002), few studies have been conducted to examine Latino English-language learners' oral-English abilities in relation to their growth in English reading. The majority of prior studies involved static


correlations at one or two time points. Two major findings can be gleaned from the prior work. First, global English-oral proficiency does not tend to correlate with young English-language learners' global English-reading achievement (Durgunoglu, et al., 1993; Quiroga, et al., 2002). Second, global oral-English proficiency and oral-English vocabulary have been shown to be related to English-reading comprehension, and Englishphonological processing has been shown to be related to English-word reading (Chiappe, Siegel & Gottardo, 2002; Gottardo, 2002; Quiroga et al., 2002). In short, presence of Latino English-language learners is on the rise in U.S. schools, and learning to read English is critical for their future success in academia and society. Further, statistics suggest average gains over the course of the past two decades, but a considerable gap persists between Latino students' English-reading achievement and their monolingual peers' achievement. To date, limited findings provide some description of young English-language learners' early English-reading trajectories, and static correlations inform selected issues about the relationship between oral-English proficiency and English-reading abilities for English-language learners. However, there are no "grand" formal theories to guide hypotheses about trajectories or oral-to-reading relationships, nor are there studies documenting the growth trajectories for English-reading achievement or the growth pattern for oral-to-reading relationships for young Latino children. The present study is one step toward building better understandings of the trajectories for Englishreading achievement and selected reading subprocesses, as well as for selected oral-to-reading relationships. The following research questions were addressed in the present study: (1) How does first-grade Latino English-language learners' growth in English instructional reading level and selected word-level reading subprocesses (ability to read words in isolation, phonemic awareness, and phonics) compare to their monolingual native-English-speaking peers' growth?; and (2) Does first-grade Latino English-language learners' English-reading growth (instructional reading level and selected wordlevel reading subprocesses) vary according to their oral English language abilities?


Methods Setting School. Hope City Elementary School' was located in a rural region of the southeastern United States. The school employed 53 teachers and served approximately 800 students in kindergarten through fifth grade. The school's student body was 18% Latino, 37% African American, and 43% Caucasian American. The remaining 2% of the student body was classified as Other racial or ethnic group. Nearly, twothirds of students (61%) received free or reduced-price lunches. Classrooms and Participants. Participants were drawn from two intact first-grade classrooms that were selected by the school's principal in consultation with the participating teachers. There were four first-grade classrooms in the school. The principal of the school assigned all the firstgrade Latino children to the two classrooms in the present study. The study included all the students from both classrooms who were in attendance at the school for at least seven continuous months of the school year. There were 47 students in total, 25 from one classroom, and 22 from the other. Eighteen of the participants were female, and 29 were male. Nearly, three quarters (72%) of the students in the sample received free or reduced-price lunches. Ethnicities were: 59.60% Latino (N=28), 21.30% African-American (N=10), 17.10% Caucasian American (N=8), and 2.10% Native American (N=1). Twenty-eight of the participants were Latino English-language learners, and 19 were monolingual nativeEnglish-speakers. Of the native-English-speaking students, 52.60% (N=10) were African American, 42.10% (N=8) were Caucasian American, and 5.30% (N=I) were Native American. Of the Latino students, 68% had lived in the United States since birth, 29% had lived in the United States for more than two years but not since birth, and 3% had lived in the United States less than one year (71% of Latino students reported, N = 20). School officials assessed the oral-English skills of the Latino students at the start of the school year using a measure of Oral English Fluency (IPTI: Oral [Dalton, Tighe, & Ballard, 1991]). Students' scores ranged from 1, or "Non-English Speaker," to 5, or "Fluent-English Speaker," with a mean score of 2.79 (s.d.=1.18, n=24), or "LimitedEnglish Speaker."

1 Pseudonyms are used.


Both teachers were general educators with no specialized training in how to teach English-language learners. All of the instruction in both classrooms was conducted in English. In one class, 12 English-language learners received approximately 30 minutes of small group instruction from the school's English-as-a-Second-Language teacher twice each week. In the other class, the English-as-a-Second-Language teacher worked with mixed groups of English-language learners and native-English speakers in the regular classroom for 30 minutes twice each week. The students who worked with the English-as-a-Second-Language teacher varied from visit to visit. Both teachers described their instructional practices in literacy as balanced in the sense of attending to both sound- and word-level features of reading and writing as well as text-level processes. Both teachers reported frequently engaging students in activities focused on developing their understandings in phonological awareness, letter-sound relationships, word identification strategies, comprehension, and fluency. Both used small-group and whole-class instruction to teach reading. Writing was an important part of instruction in the eyes of both teachers, and each conducted daily writing lessons. Data Sources and Variables Createdfrom Them Overview At two points in time-mid-year and end-of-year-three reading measures (Oral Reading of Successively Difficult Passages [Bader & Weisendanger, 1994], Basic Sight Vocabulary test [LaPray & Ross, 1986], and Hearing Sounds in Words [Clay, 1993b]) were administered to all participants. At the end of the year four oral-English measures were administered to the Latino participants: IPTI: Oral (Dalton, Tighe, & Ballard, 1991), Picture Description Task (Snow et al., 1995), Noun Definitions Task (Snow et al., 1995), and Narrative Production Task (Snow et al., 1995). As Table 1 shows, from the three reading measures, we created four reading variables to represent our constructs of interest in the research questions. From Oral Reading of Successively Difficult Passages, we created two variables-Instructional Reading Level and Phonics Knowledge. From the Basic Sight Vocabulary test, we created the variable Reading Words in Isolation. From Hearing Sounds in Words, we created the variable Phonemic Awareness in a Literacy Context. From the four oral-English measures we created five variables. From the ITPI: Oral we created the variable Overall English Oral Ability. From the Picture Description Task we created the variable Decontextualized Descriptive Oral Language. From the Noun Definitions Task we created


Decontextualized Formal Definitional Oral Language and Decontextualized Informal Definitional oral Language. From the Narrative Production Task we created Syntactic Complexity. The first author administered the measures individually in English in counterbalanced fashion. Administration sessions lasted approximately 20-30 minutes per child. To our knowledge the validity of the measures used in the present study has never been evaluated statistically. The measures for the reading variables might be considered to have face validity because they are commonly used in early-grades classrooms or they are very similar to measures commonly used in early-grades classrooms. The reading measures have been used extensively in prior research, and the orallanguage measures have been used in prior research, but to a lesser extent. In the sections that follow, data sources are detailed, followed by a description of the variables that were created. Table 1 shows the variables, data sources, and administration procedures in abbreviated form. Additionally, reliabilities for all variables are shown in the table. Interested readers may find the table useful as an outline or advanced organizer for the material that follows. Table 1 Variables, Data Sources, AdministrationProcedures,andReliabilitiesa Administration Variable Data Source Procedure Reliability InstructionalReading Level (allparticipants) Instructional Oral Reading The student read aloud .92 (for Reading of successively difficult identifying Level Successively passages while the miscues), .92 Difficult examiner recorded (for determining Passages miscues on a separate instructional (Bader & copy of the passage reading level) Weisendanger, (Barr et al., 1995; 1994) Clay, 1993a). aExcept where noted, reliability estimates are interrater estimates based on random samples of 25% of available protocols.


Table 1 (continued) Variables,Data Sources, Administration Procedures,and Reliabilitiesa Administration Reliability Procedure Data Source Variable Selected Reading Subprocess Variables (allparticipants) The student read aloud, .72 (for Basic Sight Reading predicting graded lists of words Words in Vocabulary degree of reading level & ordered by test (LaPray Isolation within two difficulty. Ross, 1986) grade levels) as reported by the test authors (LaPray & Ross, 1986)

Phonemic Awareness in a Literacy Context

Hearing Sounds in Words (Clay, 1993b)

Phonics Knowledge

Oral Reading of Successively Difficult Passages (Bader & Weisendanger, 1994)

The student listened to two sentences spoken slowly, one at a time, and wrote what was heard. Using the examiner's markings of a student's oral reading of the Instructional Reading Level passage, and using Clay's (1993a) method, each miscue was evaluated for the student's use of visual cues to say sounds associated with the visual cues. Phonics Knowledge was the percentage of miscues showing use of visual cues.

.93 (for agreement within 2 score points) .96 for percentage of miscues indicating use of visual cues

aExcept where noted, reliability estimates are interrater estimates based on random samples of 25% of available protocols.


Table 1 (continued) Variables,Data Sources, Administration Procedures,andReliabilitiesa Administration Variable Data Source Procedure Reliability OralEnglish LanguageAbility (Latinoparticipantsonly) Overall IPTI: Oral The student answered Split-half, English (Dalton, orally presented questions .82; testretest, .93 Oral Ability Tighe, & (e.g., "What is your Ballard, 1991) name?") and looked at (Dalton, pictures and objects and Tighe & answered questions (e.g., Ballard, Pointing to a picture of an 1991) apple the examiner says, "What is this? This is an Decontextu alized Descriptive Oral Language (Snow et al., 1995) Decontextu alized Formal Definitional Oral Language (Snow et al., 1995)

Picture Description Task (Snow et al., 1995)

Noun Definitions Task (Snow et al., 1995)

Child looked at a colorful picture of a circus tent and surrounding activity and described the picture to the examiner.

.80 (for agreement within one score point)

Child defined 14 nouns .75 (e.g., alphabet, bicycle, bird, clock, umbrella). A response was coded as formal if it was marked by the presence of a superordinate (often followed by a relative clause). aExcept where noted, reliability estimates are interrater estimates based on random samples of 25% of available protocols.


Table 1 (continued) Variables, Data Sources, Administration Procedures,andReliabilitiesa Administration Variable Data Source Procedure Reliability Oral English LanguageAbility (Latinoparticipantsonly) Decontextu Noun See above. A response was .95 alized Definitions coded as informal if it did Informal Task (Snow et not include a superordinate Definitional al., 1995) and show an attempt to use Oral a relative clause. Language (Snow et al., 1995) Syntactic Narrative The student was shown .80 Complexity Production three pictures that (Snow et Task (Snow et comprised a simple al., 1995) al., 1995) wordless story about bears. The examiner asked the student to look at the pictures and tell a story represented in the pictures. aExcept where noted, reliability estimates are interrater estimates based on random samples of 25% of available protocols. Data Sources Oral reading of successively difficult passages (Bader & Weisendanger, 1994; Barr, Blachowicz & Wogman-Sadow, 1995; Clay, 1993b) (Administered to all participants). Students read aloud increasingly difficult graded texts while the examiner recorded miscues on a separate copy of the passage (Barr et al., 1995; Clay, 1993a). Basic sight vocabulary (LaPray& Ross, 1986) (Administeredto all students). Students read words aloud from grade-level lists ordered by degree of difficulty. Test authors provided criteria for "passing" a grade level. Hearing sounds in words (Clay, 1993b) (Administered to all students). Students listened to the following sentences dictated aloud slowly, one at a time, and wrote down what was heard (Clay, 1993a; Johnston, 1992). The dictated sentences were: "My friend wants to go and


see the big sailboat. It has a blue flag." Any representation for a spoken sound was scored as correct (Johnston, 1992). IPTI. Oral (Dalton, Tighe & Ballard, 1991) (Administered to Latino students only). The student answered orally presented questions (e.g., "What is your name?") and looked at pictures and objects and answered questions (e.g., Pointing to a picture of an apple the examiner says, "What is this? This is an

__

."_).

Picture description task (Snow, Tabors, Nicholson & Kurland, 1995) (Administered to Latino Students only). Students looked at a colorful picture of a circus tent and surrounding activity and were asked to describe the picture aloud in English in enough detail so the examiner could recognize it. Student responses were audio taped and later transcribed for analysis. Noun definitions task (Snow, Tabors, Nicholson & Kurland, 1995) (Administered to Latino students only). Students defined 14 nouns (e.g., alphabet, bicycle, bird, clock, umbrella). The examiner read each noun aloud in turn, and the student provided a definition, or stated that he/she did not know the meaning of the word. Student responses were audio taped and later transcribed for analysis. Narrative production task (Snow, Tabors, Nicholson & Kurland, 1995) (Administered to Latino students only). Students were shown three pictures that comprised a simple wordless story about bears flying a kite. The examiner asked the student to look at the pictures and tell a story represented in the pictures. Student responses were audio taped and transcribed for analysis Variables and Associated Reliabilities In this section, the variables are described, and the reliability estimates associated with each variable are reported. Except where noted, reliability estimates are interrater agreements of two trained raters who independently scored a random sample of 25% of available protocols. InstructionalReading Level Using Clay's (1993a) method, Instructional Reading Level was the highest-level passage from the Oral Reading of Successively Difficult Passages data source at which the student read with at least 90% word recognition accuracy. A score of "0" indicated that a student did not pass even the lowest reading passage; .25 indicated approximately a pre-primer level, which is, for typically developing students, achieved around the


beginning of first grade; .50 indicated approximately a primer level, achieved by typically developing students around the middle of first grade; 1.00 indicated approximately end-of-first grade level; 2.00 approximately second grade level; and so on. The interrater reliability estimate was .92 for identifying miscues and .92 for determining Instructional Reading Level. Reading Subprocess Variables Reading words in isolation. Using criteria provided by the test authors for passing graded lists, grade equivalent scores were derived from the Basic Sight Vocabulary measure. A score of "0" indicated that a student did not pass even the lowest reading passage; .25 was approximately a pre-primer level; .50 was approximately a primer level; 1.00 was end-of-first grade level; 2.00 second-grade level; and so on. The reliability estimate given by test authors for predicting within two grade levels the accurate level was .72 (LaPray & Ross, 1986). Phonemic awareness in a literacy context. This variable was created from the Hearing Sounds in Words data source. Any representation for a spoken sound was scored correct (Johnston, 1992). The possible total raw score was 45, and scores were converted to percent correct. The interrater reliability estimate, within two score points for agreement, was .93. Phonics knowledge. The Phonics Knowledge variable was created from the Oral Reading of Successively Difficult Passages data source. Using the examiner's markings of a student's oral reading miscues from the Instructional Reading Level passage, and using Clay's (1993a) method, each miscue was evaluated for evidence the student used visual cues (i.e., letters) to identify sounds associated with that cue. Then, the percentage of miscues showing use of visual cues was calculated and labeled Phonics Knowledge. The interrater reliability estimate for Phonics Knowledge was .96. Oral English LanguageAbility Overall English oral ability. Generated from the IPTI: Oral (Dalton, Tighe & Ballard, 1991), using the test authors' scoring system, a total score was awarded of A, B, or C, equivalent to "non-English-," "limited-English-," or "fluent-English-speaking," respectively. The letters were assigned numerical values from 1 to 3. The split-half reliability estimate provided by the test authors was .82, and the test-retest estimate was .93 (Dalton, Tighe & Ballard, 1991).


Decontextualized descriptive oral language (Snow, Tabors, Nicholson & Kurland, 1995). Using procedures outlined by Snow and colleagues (Snow, Tabors, Nicholson & Kurland, 1995) a composite score was obtained from the PictureDescription Task as an indicator of a child's oral decontextualized language. Decontextualized language is language used in literate contexts and is characterized by "specific features that differ from much oral language . . . [such as] the use of sophisticated vocabulary and procedures for making information linguistically explicit rather than relying on shared physical and social contexts . . . ." (Snow, Tabors, Nicholson, & Kurland, 1995, p. 38). To begin, for each description, we counted specificity markers (adjectives, verbs, and locatives), specific vocabulary (e.g., circus), and number of words spoken. Then, using a conversion scale provided by the authors, the raw scores for specificity markers were changed to converted scores, which could range from 0 to 4, and the number of words spoken was changed to converted scores, which could range from 0 to 11. Finally, we summed the specificity marker converted score, the specific vocabulary score, and the number of words spoken converted score to obtain the final composite scores. The possible score range was from 0 to 28. The interrater reliability estimate, within one point, was .80. Decontextualizedformaldefinitional orallanguage (Snow, Tabors, Nicholson & Kurland, 1995). This variable was created using the following five-step procedure provided by Snow and colleagues (Snow,Tabors, Nicholson, & Kurland, 1995). In Step 1, responses from the Noun Definitions Task were coded as formal or informal. A response was coded as formal if it was marked by the presence of a superordinate (often followed by a relative clause). A superordinate is a word whose meaning encompasses the meaning of another more specific word. For example, "animal" is a superordinate of "dog." An example of a response coded as formal for the noun bird is, "It is a flying animal with wings." Conversely, a response was informal if it did not provide a superordinate and show an attempt to use a relative clause. An example of a response coded as informal for the noun bird is, "It flies and lives outside." In Step 2, for formal definitions, Quality subscores were assigned for aspects of the participant's definition-syntax, superordinates, relative clauses, synonyms, definitional features, function, other description, examples, and comparisons. Quality subscores could range from 0 to 2, 3, 4 or an unlimited number (for definitional features) depending on the definitional aspect. In Step 3, for all responses coded as formal, the associated Quality


subscores were summed. One subscore, definitional features, was the total number of descriptors mentioned, so definitional features could be unlimited in number. In Step 4, the summed Quality subscores for syntax, superordinate, relative clause, synonyms, definitional features, function, other description, examples and comparisons were summed across all formal definitions. In Step 5, the Decontextualized Formal Definitional Oral Language score was the total sum from Step 4 divided by the total number of definitions coded formal, and the result was rounded to the nearest whole number. The lowest possible score was 0, and the maximum possible score was unlimited. The interrater reliability estimate was .75. Decontextualized informal definitional oral language (Snow, Tabors, Nicholson & Kurland, 1995). A description of Snow et al.'s (1995) five-step procedure for creating this variable follows. In Step 1, responses from the Noun Definitions Task were coded as formal or as informal as described in the section on Decontextualized Formal Definitional Oral Language. In Step 2, for informal definitions, Quality subscores were assigned for aspects of the participant's definition-synonyms, definitional features, function, other description, examples, and comparisons. Quality subscores could range from 0 to 2, 3 or an unlimited number (for definitional features) depending on the definitional aspect. In Step 3, for all responses coded as informal, the associated Quality sub-scores were summed. One subscore, definitional features, was the total number of descriptors mentioned, so definitional features could be unlimited in number. In Step 4, the summed Quality subscores for synonyms, definitional features, function, other description, examples and comparisons were summed across all informal definitions. In Step 5, the DecontextualizedInformal Definitional Oral Language score was the total sum from Step 4 divided by the total number of definitions coded informal, and the result was rounded to the nearest whole number. The lowest possible score was 0, and the maximum possible score was unlimited. The interrater reliability estimate was .95. Syntactic complexity (Snow, Tabors, Nicholson & Kurland, 1995). This variable was created from the Narrative Production Task (Snow et al., 1995) using the three-step procedure provided by the authors. In Step 1, the total number of words was counted. In Step 2, clauses were identified. In Step 3, the number of words per clause was calculated and then converted to a score scaled from 1 to 4. The interrater reliability estimate for the converted scores was .80.

38

Reading Research and Instruction, Volume 46, No. I * Fall 2006

Results The data were analyzed using a series of repeated measures analyses of variance. In the sections that follow, the analytical models used to address each of the research questions are specified and the results are presented. Research Question I To address the first research question-How does first-grade Latino English-language learners' growth in English instructional reading level and selected word-level reading subprocesses (ability to read words in isolation, phonemic awareness, and phonics) compare to their monolingual native-English-speaking peers' growth?-two repeated measures analyses of variance were performed, first using Instructional Reading Level as the dependent variable, then with follow-up analyses to examine growth in specific word- and sound-level subproccesses of reading (ability to Read Words in Isolation, Phonemic Awareness in a Literacy Context, and Phonics Knowledge). In the following sections, first each analytical model is fully specified, and then the significant results are detailed. Is Language Status Related to InstructionalReading Level Growth? A repeated measures analysis was performed with: language status (Latino English-language learners and monolingual native-Englishspeaking peers) as the between-subjects factor, time as the within-subjects factor (middle and end of the year), and Instructional Reading Level as the dependent variable. The Test of Compound Symmetry was not significant, F(3, 162000) = 2.156, p > .09. Consequently, a univariate analysis was done. Language status was not related to Instructional Reading Level growth. There were no significant interactions involving language status, and no main effect of language status. Means and standard deviations are shown in Table 2. There was a time main effect, F(1, 30) = 23.42, p