Use of Increasingly Complex Text to Advance ELs Knowledge and Academic Language

Transcription

1 Section I: Linguistically and Culturally Diverse Literacies Use of Increasingly Complex Text to Advance ELs Knowledge and Academic Language Literacy Research: Theory, Method, and Practice 2015, Vol. 64, ª The Author(s) 2015 Reprints and permission: sagepub.com/journalspermissions.nav DOI: / lrx.sagepub.com Lisa M. O Brien 1 and Christine M. Leighton 2 Abstract This mixed-methods study explored joining use of increasing-complex text with sound instructional practices on English Learners (ELs) academic language and conceptual knowledge. Findings showed one-week postintervention, ELs achieved significant academic vocabulary gains such that there were no differences between ELs and general education (GE) students. Moreover, six weeks postintervention, ELs academic language and conceptual knowledge closely approximated that of their GE peers. Findings suggest that carefully scaffolded use of increasingly-complex text holds promise in advancing ELs conceptual knowledge and academic language. Given the widespread use of simplified text as ELs primary reading curriculum a practice that unintentionally delimits language and knowledge learning opportunities, and in turn, contributes to persistent achievement gaps efforts aimed at identifying practices that build academic language and conceptual knowledge and in turn, enable access to complex text are of particular importance. If replicated in larger studies, teachers will have a clear alternative to such practices. Keywords expository text, academic vocabulary, disciplinary literacy, SEI 1 Boston University, Boston, MA, USA 2 Emmanuel College, Boston, MA, USA Corresponding Author: Lisa M. O Brien, Boston University, 2 Silber, Boston, MA 01741, USA. lmobrien5674@gmail.com

2 170 Literacy Research: Theory, Method, and Practice 64 The number of English-language learners (ELs) in U.S. schools is rapidly increasing with an estimated 4.7 million enrolled during the school year (Nachazel & Yohn, 2012); unfortunately, decades of research document that ELs are consistently underserved (e.g., Hemphill, Vanneman, & Rahman, 2011). Many ELs begin school with large vocabulary differences (August, Carlo, Dressler, & Snow, 2005; Mancilla- Martinez & Lesaux, 2011); which over time affect comprehension development and long-term literacy achievement (Mancilla-Martinez & Lesaux, 2010, 2011). Among children achieving reading comprehension proficiency in fourth grade, only 2% are ELs (National Assessment of Educational Progress, 2011). Increasingly, scholars point to opportunity gaps accumulated inequities in access to essential educational resources as the root of achievement differences (Darling- Hammond, 2010; Milner, 2012). Although knowledge development occurs through myriad opportunities, rich experiences with expository text are key in building knowledge that supports expository text comprehension (Barber, Catz, & Arya, 2006; Graesser, McNamara, & Louwerse, 2003). Too often, though, little instructional time is devoted to expository and other informational text types (Jeong, Gaffney, & Choi, 2010), particularly for ELs. Without such access, the high rates of low comprehension proficiency for many ELs will likely persist. Given the evidence that many ELs demonstrate limited early knowledge and poor comprehension achievement and are often denied opportunities with complex text, research is needed to identify ways to support young ELs access to complex texts. Toward this end, this mixed-methods study explored instructional use of topically related and increasingly complex text as a way to accelerate young ELs conceptual knowledge and academic language. We compared knowledge and academic language outcomes for first graders attending a sheltered English immersion (SEI) classroom to their general education (GE) peers. We asked: 1. Are there differences in SEI children s knowledge and academic language after a 5-week read-aloud intervention using increasingly complex text as compared to GE children receiving the same curriculum using texts of comparable complexity? 2. What are the characteristics of academic language use and conceptual knowledge after the intervention for children with differing learning profiles? Research and Theoretical Perspective Complex expository text is particularly important for building the type of knowledge that enables expository text comprehension and anchors academic and disciplinary learning throughout the school years (Barber et al., 2006), because much of this knowledge is not readily accessible through everyday language use and experiences. Expository text discourse is replete with abstract concepts (e.g., fossilization, Graesser et al., 2003) and academic language characterized by specific organizational structures, lexical choices, and grammar negotiated by and situated in particular discourse communities (e.g., science; Schleppergrell, 2004; Snow & Ucceli, 2009). To learn

3 O Brien and Leighton 171 from expository text, children need knowledge of these linguistic features as well as the concepts embedded in the language (i.e., conceptual knowledge; Bransford, Brown, Cocking, Donovan, & Pelegrino, 2000). While all children need instructional experiences that develop the academic language and conceptual knowledge to facilitate access to complex text, this undertaking is particularly significant for ELs who are also developing overall oral English proficiency (i.e., everyday language already developed by their monolingual peers) along with academic language and conceptual knowledge. As the Common Core State Standards (CCSS, 2010; National Governors Association Center for Best Practices, Council of Chief State School Officers, 2010) call for use of complex texts at all grade levels, the stakes are high for ELs who may take 4 7 years to develop the academic language (Hakuta, Butler, & Witt, 2000) crucial to comprehending and learning from these texts. In attempts to promote success, teachers often provide ELs with large amounts of simple text (Fillmore & Fillmore, 2012), with the unintended consequence of restricting exposure to academic language and key knowledge-building opportunities afforded to their monolingual English-speaking peers. However, there is little extant evidence examining interventions that build academic language and conceptual knowledge and promote access to complex expository text with young ELs (e.g., Fillmore & Fillmore, 2012). This leaves young ELs teachers with little guidance on how to meet CCSS. As such, teachers are likely to continue avoiding the use of complex texts with their ELs altogether. To address this problem, we explored a read-aloud intervention aimed at supporting ELs access to complex expository text. We grounded our intervention in evidence showing the benefit of interdisciplinary approaches to knowledge and comprehension outcomes for monolingual English speakers (Barber et al., 2006; Halvorsen et al., 2012) and ELs (Cervetti, Bravo, Duong, Hernandez, & Tilson, 2008; Maloch, 2008), wide reading to vocabulary and conceptual knowledge (e.g., Kuhn et al., 2006), and recent theory of text complexity (Mesmer, Cunningham, & Heibert, 2012). We hypothesized for SEI children, interdisciplinary use of increasingly complex and topically related text would scaffold and accelerate access to the same text afforded to their monolingual English-speaking peers. That is, as SEI children were provided instruction with less complex text, they would acquire academic language and conceptual knowledge to scaffold their ability to access and learn from subsequent topically related and increasingly complex text. Although we explored the use of increasingly complex text with a small sample and did not explicitly measure comprehension, this study provides much-needed grounding for future research. Methods Setting and Participants The setting was three first-grade classrooms within the same school in a northeastern district serving 70 diverse children from low-middle income families. This school was

4 172 Literacy Research: Theory, Method, and Practice 64 Table 1. Demographic Characteristics of Children by Group. Demographics GE SEI n ¼ 31 n ¼ 10 Mean age in months Gender Female 20 7 Male 11 3 Ethnicity* White 14 0 Latino 12 9 African American 2 0 Asian 1 1 Other 2 0 English-language learner 0 10 First language* English 16 0 Spanish 12 9 Chinese 1 1 Other 2 0 Note. GE ¼ general education; SEI ¼ sheltered English immersion. *p <.01. the second lowest performing school within the district as indicated by statewide assessments. Enrollment in the study was voluntary and participating classrooms comprised two GE classrooms and one SEI classroom. Forty-one children participated (Table 1) with SEI children (n ¼ 10) demonstrating lower literacy skills (as measured by the Group Assessment and Diagnostic Evaluation [GRADE] word reading and comprehension subtests; Williams, 2001), vocabulary knowledge (as measured by the Peabody Picture Vocabulary Test-IV [PPVT]; Dunn & Dunn, 2007), and genre exposure (a composite measure of the proportion of expository texts in classroom libraries, children s self-selected texts, and teacher read alouds) at pretest than GE children (n ¼ 31). SEI and GE children indicated they were somewhat interested in the focal topic (as measured by a topic interest survey, Table 2). Intervention and Materials Intervention. To support the development of conceptual knowledge, all lessons adhered to three guiding principles: (1) Build conceptual knowledge by organizing instruction around a knowledge goal (Bransford et al., 2000). Such instruction may provide children opportunities to acquire networked knowledge (Nagy & Hiebert, 2011), and in turn, facilitate comprehension. (2) Build deep conceptual knowledge through multiple experiences in various contexts (Bransford et al., 2000). Multiple read alouds of topically related texts provide an ideal context for engaging with important concepts (e.g.,

5 O Brien and Leighton 173 Table 2. Participants Learning Characteristics by Group. Word Reading Stanine Listening Comprehension Stanine PPVT Standard Score Genre Exposure Topic Preference n M (SD) M (SD) M (SD)* M (SD)* M (SD) GE (0.47) 4.65 (0.34) (2.54) 0.40 (0.02) 0.97 (0.15) SEI (2.02) 3.70 (1.89) (10.43) 0.28 (0.01) 1.11 (0.93) Note. GE¼ general education; SEI ¼ sheltered English immersion; PPVT ¼ Peabody Picture Vocabulary Test. *p <.05. Table 3. Sample Three-Day Lesson Cycle. Day 1 Day 2 Day 3 Before Reading Before Reading Before Reading Introduce conceptual organizer Review taught words and content from prior Review taught words and content from prior Organize responses reading reading Introduce new words Introduce new words Explain purpose for Establish purpose and introduce text Explain purpose for reading reading Read aloud first segment Read aloud second segment Second read aloud Word explanations Word explanations Reread aloud entire Prompt discussion Prompt discussion text Teacher led word sort Teacher guided word Prompt discussion sort Organize responses Wrap-up Wrap-up Extension Recap Recap Prompt discussion in a Prompt application of Prompt application of new context taught vocabulary taught vocabulary Write about focal concepts Wrap-up Recap Prompt application of taught vocabulary Neuman, Kaefer, & Pinkham, 2014; Pollard-Durodola et al., 2011). (3) Promote concept development with talk and tasks to build on what students know while also reconstructing misconceptions (Bransford et al., 2000; Vosniadou, 2003). Effective discussion and tasks (e.g., semantic sorts) provide opportunities to process new information and clarify misconceptions (e.g., Heisey & Kucan, 2010). The intervention comprised 15 lessons distributed across 5 weeks. Lessons were grouped in sets of three per text that is, a lesson cycle (see Table 3 for a sample

6 174 Literacy Research: Theory, Method, and Practice 64 Table 4. Text Features for Sheltered English Immersion Texts by Lesson Cycle. Cycle Text Word Count Sentences Flesch Kincaid Reading Level Narrativity 1 Dinosaurs (national geographic) Learning from fossils Digging up dinosaurs Dinosaurs Dinosaur tracks Note. The texts for Cycles 4 and 5 were the same for both classrooms. Narrativity refers to the extent to which a text conveys a story. lesson sequence). The researcher provided all instruction and instruction occurred during regularly scheduled literacy lessons of min per lesson. For both groups, all lesson sequences focused on the same knowledge goal cast as a question: How do scientists collect and study fossils to learn about dinosaurs? All lesson sequences also included the same set of instructional activities. This included two read alouds of the focal text, explicit vocabulary instruction, talk, and tasks that facilitate knowledge development, and opportunities to apply new knowledge to differing contexts. Texts were read aloud twice as research suggests that optimal vocabulary knowledge development for primary grade children occurred with two to four interactive read alouds (see Biemiller & Boote, 2006 for a review). Because of the texts complexity, read alouds on the first and second day comprised Segments 1 (first half of text) and 2 (second half of text) of each text with questions posed at natural stopping points to scaffold talk and deepen understanding. Question development was guided by Beck and McKeown (2006). The entire text was read aloud on the third day with additional questions to scaffold talk and understanding. Additional key activities included the use of open word sorts, teacher student and student student talk, and prompting the use of taught words. For SEI children, the intervention also included scaffolding shown to benefit older ELs such as use of pictures with vocabulary instruction and scaffolded talk (e.g., sentence stems and revoicing) and writing with language-based supports (e.g., graphic organizers, sentence stems, and oral rehearsal, Baker et al., 2014). Materials. SEI read alouds comprised five topically related and increasingly complex expository text; GE read alouds comprised five topically related expository text of comparable complexity (see Tables 4 and 5 for text titles and complexity features). Read alouds in Week 5 used the same text. Cohmetrix was used to analyze text complexity features (Graesser, McNamara, & Kulikowich, 2011). Implementation fidelity. Ten percent of lessons were randomly selected and rated by two independent raters using an implementation fidelity check list with 97.53% agreement

7 O Brien and Leighton 175 Table 5. Text Features for General Education Texts by Lesson Cycle. Cycle Text Word Count Sentences Flesch Kincaid Reading Level Narrativity 1 Best book of fossils, rocks, 1, and minerals 2 Fossils part Fossils part Dinosaurs Dinosaur tracks Note. Narrativity refers to the extent to which a text conveys a story. (Cohen s k ¼.90) and lessons between groups were comparable, t(88) ¼ 1.60, p ¼.113. Data Sources and Collection Procedures Data sources were comprised of two researcher-designed measures of knowledge and academic language: Receptive Assessment of Science Vocabulary (RASV) and Conceptual Knowledge Interview (CKI). Administered within 1 week preintervention (Time 1) and 1 week postintervetion (Time 2), RASV was modeled after the PPVT and assessed the receptive knowledge of discipline-specific academic vocabulary for 20 taught words. RASV was reviewed by a content expert and piloted informally for face and content validity. RASV and PPVT were strongly correlated (r ¼.72, p <.001). CKI assessed conceptual knowledge and academic language and comprised two illustrations depicting key curricular concepts. Students were shown two illustrations depicting concepts and prompted to describe everything they saw (see Appendix A for a sample item). All CKIs were collected within 1 week postintervention and 6 weeks postintervention (Time 3). Analytic Process Quantitative Analyses To determine if there were differences in SEI and GE children s academic vocabulary growth, we conducted t-tests with RASV as the dependent variable. To determine if there were differences in children s conceptual knowledge depth (i.e., linked concepts) or academic language (i.e., proposition), we first analyzed children s CKI explanations qualitatively and converted qualitative data into quantitative scores which occurred in three stages. 1. In the first stage, we developed a coding scheme based on prior research (Schwendimann, 2014) and themes that emerged from the data. We segmented

8 176 Literacy Research: Theory, Method, and Practice 64 explanations into proposition values: full propositions comprising two concepts linked together (e.g., people digging fossils); 2. partial propositions comprising a concept that was not linked to another concept (e.g., sand, this is a scientist); 3. redundant comprising proposition values that were previously stated; and 4. not codable comprising proposition values lacking specificity (e.g., or something) or were inaudible or unintelligible. We then coded proposition values according to their relevancy to focal concepts as an indicator of conceptual knowledge depth. Relevant proposition values conveyed at least one relevant idea central to key concepts (e.g., The scientist was learning), and irrelevant proposition values conveyed concepts that were irrelevant to focal concepts (e.g., The man is wearing glasses). Proposition precision was coded along four dimensions as an indicator of conceptual knowledge and academic language: 1. precise proposition values comprising completely accurate information using precise language or referent that was previously stated (e.g., The paleontologist finds the fossil and They find fossils); 2. imprecise proposition values comprising completely accurate content but lacking precise language (e.g., The man is trying to find the bones); 3. partially inaccurate proposition values comprising some accurate content and some inaccurate content (e.g., Fossils lived a long time ago); and 4. inaccurate proposition values comprising completely inaccurate content (e.g., And they brushed their teeth). In the second stage, having agreed on an initial set of codes, we applied these codes to one CKI explanation and then modified the codebook such that it accurately portrayed proposition values in each explanation (with definitions and examples). We repeated until reaching coding saturation (Bowen, 2008). Next, each researcher independently coded a randomly selected subset of CKI explanations (10%). Agreement was 91.2%, Cohen s k ¼.84. Remaining data were then divided between the two researchers for independent coding. In the third stage, we converted these data into frequencies and proportions. We calculated proportions to make equitable comparisons as each explanation varied in length and, in turn, the total number of proposition values. To determine the relative amount and quality of children s conceptual knowledge and academic language, we calculated the proportion of each type of proposition to total proposition (excluding not codable and redundant proposition) based on mean values for each code. For example, we calculated the proportion of relevant proposition values to total proposition values (excluding not codable and redundant proposition values). We then regressed the proportion for each outcome on group (SEI ¼ 1; GE ¼ 0). To do so, we used the generalized linear model procedure with the logit option and

9 O Brien and Leighton 177 Table 6. Demographics and Learning Characteristics of Matched Pairs. Low Low Average Average High Average Characteristics SEI GE SEI GE SEI GE SEI GE Name Maria Celia Jose Daniela Ana Clara Ernesto Rick Gender Female Female Male Female Female Female Male Male EL Yes No Yes No Yes No Yes No L1 Spanish Spanish Spanish Spanish Spanish Spanish Spanish English Age (in months) PPVT GR-LC Note. GE children who spoke Spanish as their first language did not qualify for shelter English immersion instruction based on a statewide indicator of English proficiency. GE ¼ general education; SEI ¼ sheltered English immersion; L1 ¼ first language; EL ¼ English language learner; PPVT ¼ Peabody Picture Vocabulary Test; GR-LC ¼ GRADE listening comprehension. binomial family because of the restricted range of proportions (i.e., outcome values range from 0 to 1). Qualitative Analyses To explore the knowledge and academic language differences as well as the characteristics of academic language use after the intervention for children with differing learning profiles, we conducted additional qualitative analyses using multiple case study methodology (Yin, 2011) for eight illustrative cases from both the groups (SEI ¼ 4; GE ¼ 4). To do so, we first identified clusters based on the distribution of children s vocabulary knowledge (PPVT), listening comprehension (subtest of the GRADE), and topic interest demonstrated at pretest. For the PPVT, we found natural cutoff points resulting in three ranges: low average (below 90), average (between 90 and 110), and high average (above 110). For the GRADE, we did not observe natural cutoff points and, as such, identified scores (converted to stanines to have a mean of 5) of 5 or below as the lower range and 6 or above as the high range. We found topic interest scores (ranging from 1 to 3) to be evenly distributed and weakly associated with all outcome measures. As such, we did not match based on this variable. We then identified four matched pairs with differing learning profiles and each matched pair comprising an SEI and GE with comparable PPVT and listening comprehension scores at pretest (Table 6). We then used individual cases and four matched pairs as the unit of analyses. We also extended our examination to Time 3 (i.e., 6 weeks postintervention) to explore the extent to which different learning profiles maintained conceptual knowledge and academic language. We examined students explanations for word choice (i.e., quantity and quality of words used to convey focal concepts) including (1) use of targeted word (e.g., fossil, compare); (2) relevant but not targeted (e.g., dinosaur, find); and (3)

10 178 Literacy Research: Theory, Method, and Practice 64 Table 7. RASV Performance by Group. Group RASV Pretest RASV Posttest n M (SD) t M (SD) t GE (0.53) 2.47* (0.44) 1.97 SEI (0.80) (0.73) Note. GE ¼ general education; SEI ¼ sheltered English immersion; Missing data for one SEI student; RASV ¼ Receptive Assessment of Science Vocabulary. *p <.05. proxies for targeted words (e.g., look at for study, long time ago for ancient), proposition values (full or partial proposition), and level of precision (precise, imprecise, partially inaccurate, or inaccurate) for all matched pairs. Finally, we made comparisons within- and between-matched pairs. We focused these analyses on two concepts central to the intervention: fossils and paleontologist. Results Below we present the results of our exploration of differences in SEI and GE children s conceptual knowledge and academic language as well as the characteristics of academic language for children with differing learning profiles. To do so, we first present the quantitative findings examining differences in RASV gains pre- and postintervention for all study participants. We then present the mean proportion for proposition values, relevancy, and level of precision at Time 2 for all SEI and GE students. Then, we describe case study findings to illustrate our quantitative findings and explore academic language characteristics of students 1 week and 6 weeks postintervention. We compared Time 2 and Time 3 because evidence suggests that ELs often go through a silent period (i.e., ELs ability to express concepts in a second language emerges after building up competence in the second language via listening, Krashen, 1982, p. 27) and that deep knowledge is lasting knowledge (e.g., Bransford et al., 2000). These findings led us to explore if ELs explanations would reflect comparable conceptual depth and at least comparable if not greater academic language after the intervention and, in turn, permit us to more accurately portray students with differing learning profiles knowledge and academic language. Knowledge and Academic Language Differences Quantitative findings. At Time 1, SEI children demonstrated significantly less receptive knowledge of discipline-specific academic vocabulary than GE children (see Table 7). However, at Time 2, SEI children demonstrated greater gains such that there were no significant differences in SEI and GE children s vocabulary scores (see Table 8).

11 O Brien and Leighton 179 Table 8. Mean (SD) and Relationship between Proportion of Proposition Values and Their Relevancy and Level of Precision and SEI Children (n ¼ 39). SEI GE Mean (SD) Mean (SD) b z Proposition value Proposition 0.63 (0.33) 0.77 (0.24) Partial proposition 0.37 (0.33) 0.23 (0.24) Relevancy Relevant 0.95 (0.09) 0.91 (0.10) Irrelevant 0.05 (0.09) 0.09 (0.10) Level of precision Precise 0.16 (0.18) 0.26 (0.15) Imprecise 0.50 (0.24) 0.43 (0.17) Partially accurate 0.29 (0.26) 0.21 (0.18) Inaccurate (0.02) 13.97* Note. Missing data for two SEI students. SEI ¼ sheltered English immersion (n ¼ 9); GE ¼ general education (n ¼ 31). *p <.001. As we illustrate in Table 8, within 1 week of the intervention conclusion (Time 2), GE children s explanations had a greater proportion of propositions than SEI children s explanations, though not statistically significant. Approximately three fourths of GE children s explanations comprised propositions as compared to SEI children s explanation which comprised approximately two thirds of propositions. GE and SEI children s explanations comprised a comparable proportion of relevant proposition values. GE children s explanations comprised a greater proportion of precise proposition values than SEI children, though not statistically significant. SEI children s explanations comprised a greater proportion of imprecise and partially accurate proposition values than GE children, though not statistically significant. Finally, GE children s explanations comprised significantly more inaccuracies than SEI children. Case study findings across cases. To explore the differences in SEI and GE children s conceptual knowledge and academic language more in depth, below we report findings from case study (eight individual cases and four matched pairs) analyses. We first describe findings for academic vocabulary (receptive and expressive), and then we describe findings for academic language and conceptual knowledge for all eight students. All SEI and GE students demonstrated greater receptive knowledge of academic vocabulary (as measured by the RASV) at Time 2 with the exception of Clara (GE), who demonstrated decline in receptive academic vocabulary from Time 1 to Time 2. One SEI student (Ana) and one GE student (Daniela) achieved the largest gains (eight to nine words, see Table 9). These results are consistent with quantitative

12 180 Literacy Research: Theory, Method, and Practice 64 Table 9. RASV Performance by Matched Pairs. Low RASV-T1 RASV-T2 SEI Maria GE Celia Low average SEI Jose GE Daniela Average SEI Ana GE Clara High average SEI Ernesto GE Rick Note. SEI ¼ sheltered-english immersion; GE ¼ general education; RASV ¼ Receptive Assessment of Science Vocabulary; T1 ¼ Time 1; T2 ¼ Time 2. Table 10. Frequency of Word Choice in Matched Pairs Explanation of Fossil (Time 2 and Time 3). Time 2 Time 3 Group Relevant Target Proxy Relevant Target Proxy Low SEI Maria GE Celia Low average SEI Jose GE Daniela Average SEI Ana GE Clara High average SEI Ernesto GE Rick Total SEI GE Note. SEI ¼ sheltered English immersion; GE ¼ general education. findings as regardless of EL status, most students achieved gains in receptive academic vocabulary. To examine expressive academic vocabulary, we analyzed children s explanations at Times 2 and 3 for word choice used to convey the concepts of fossil (Table 10) and

13 O Brien and Leighton 181 Table 11. Frequency of Word Choice in Matched Pairs Explanation of Paleontologist (Time 2 and Time 3). Time 2 Time 3 Group Relevant Target Proxy Relevant Target Proxy Low SEI Maria GE Celia Low average SEI Jose GE Daniela Average SEI Ana GE Clara High average SEI Ernesto GE Rick Total SEI GE Note. SEI ¼ sheltered English immersion; GE ¼ general education. paleontologist (Table 11). For the concept of fossil, overall, at Time 2, SEI and GE students differed in the quantity and quality of words used to convey fossils. However, by Time 3, SEI children s word choice more closely approximated the word choice of GE students at Time 3. For example, at Time 2, SEI children s explanations comprised fewer relevant words than GE children s explanations, but overall at Time 3, SEI and GE students explanation was comparable in the amount of relevant words to convey the concept of fossils. Similarly, at Time 2, SEI children s explanations comprised more target words than GE children s explanations, whereas at Time 3 both the group s explanations comprised comparable amounts of target words. Take for example, Ana (SEI) and Clara s (GE) words used to convey the concept of fossils at Times 2 and 3. At Time 2, Ana s explanation included hardened, piece, andfossil, whereas Clara s explanation included parts of bone, missing, fossils, dinosaur, bone, andface. At Time 3, Ana s explanation included fossils, two, legs, ocean, skull, animal, long, neck, andtail, while Clara s included fossil, animal, whale, bone, humpback whale, face, tail, andlong. For the concept paleontologist, there were no observable differences between groups at Time 2 or Time 3. Across cases, SEI and GE children s explanations were fairly comparable in the quality and quantity of words used to convey the concept of paleontologist at Times 2 and 3. For example, at Time 2, José s explanation includes paleontologist and find, whereas Daniela s explanation includes man, discover, find, and put together. At Time 3, José s explanations includes search, paleontologist,

14 182 Literacy Research: Theory, Method, and Practice 64 Table 12. Overall Propositions and Precision for Fossil and Paleontologist (Time 2 and Time 3). Fossil Paleontologist Proposition Type T2 T3 T2 T3 Total propositions (full and partial) SEI GE Propositions (proportion) SEI GE Partial propositions (proportion) SEI GE Precise (proportion) SEI GE Imprecise (proportion) SEI GE Partially accurate (proportion) SEI GE Inaccurate (proportion) SEI GE Note. SEI ¼ sheltered English immersion; GE ¼ general education; T2 ¼ Time 2; T3 ¼ Time 3. find, scientist, work, and show, whereas Daniela s explanation includes scientist, find, wonder, man, and look at (proxy for compare). To examine academic language further and conceptual knowledge depth, we examined students explanations for the frequency and proportions of propositions values and the level of precision for proposition values conveying the concepts fossils and paleontologists at Time 2 and Time 3 (see Table 12). Recall that to produce a proposition a child must link two concepts together, which requires understanding of relations between concepts (i.e., conceptual knowledge depth) and knowledge of how academic language works (i.e., academic language knowledge). When discussing the concept of fossils, at Time 2, SEI and GE students explanations comprised comparable amounts of full and partial propositions that were maintained in their explanations at Time 3 (Table 12). Differences were evident in the precision of these propositions (see Table 13). Although SEI and GE students explanations were comparable in precision, GE students explanations comprised more imprecise statements while SEI students explanations comprised more partially accurate statements. By Time 3, however, SEI students explanations increased in precision whereas GE students explanations decreased in precision such that SEI

15 O Brien and Leighton 183 Table 13. Proportion of Propositions and Precision for Fossil Between T2 and T3 for Matched Pairs. Propositions Partial Propositions Precise Imprecise Partially Accurate T2 T3 T2 T3 T2 T3 T2 T3 T2 T3 Low SEI Maria GE Celia Low average SEI Jose GE Daniela Average SEI Ana GE Clara High average SEI Ernesto GE Rick Note. SEI ¼ sheltered English immersion; GE ¼ general education; T2 ¼ Time 2; T3 ¼ Time 3. students statements comprised double the proportion of precise statements as GE students. SEI students explanations comprised greater partially accurate statements than GE students, whereas GE students explanations comprised greater imprecise statements than SEI students. When discussing the concept of paleontologist (Table 12), similarly at Time 2, SEI and GE students explanations were comparable in both the proportion of full and partial propositions and this was maintained at Time 3. As with fossils, differences were also evident in the precision of SEI and GE students statements (see Table 14). At Time 2, SEI students statements comprised more than twice as many precise statements than GE students statements. This level of precision was not maintained at Time 3, and SEI and GE children were somewhat comparable in the precision of their explanations. Consistent with explanations of fossil, SEI students explanations comprised greater partially accurate statements than GE students, whereas GE students explanations comprised greater imprecise statements than SEI students. Looking across results, several findings emerge: 1. Although GE children demonstrated greater receptive knowledge of academic vocabulary prior to the intervention, within 1 week postintervention (Time 2), there were no significant differences. Examination of eight illustrative cases shows that while GE children demonstrated greater knowledge of expressive academic vocabulary at Time 2 (for fossils) no differences were evident by 6 weeks postintervention. 2. Both the groups produced explanations with greater academic vocabulary for fossils than paleontologist at Times 2 and 3.

16 184 Literacy Research: Theory, Method, and Practice 64 Table 14. Proportion of Propositions and Precision for Paleontologist Between T2 and T3 for Matched Pairs. Propositions Partial Propositions Precise Imprecise Partially Accurate T2 T3 T2 T3 T2 T3 T2 T3 T2 T3 Low SEI Maria GE Celia Low average SEI Jose GE Daniela Average SEI Ana GE Clara High average SEI Ernesto GE Rick Note. SEI ¼ sheltered English immersion; GE ¼ general education; T2 ¼ Time 2; T3 ¼ Time Both quantitative and qualitative analyses revealed no differences in the proportion of full and partial propositions in SEI and GE students explanations. 4. Although quantitative findings revealed no significant differences in SEI and GE students precision, qualitative findings show differences by group and concept. a) For fossil, at Time 2, SEI students explanations comprised fewer precise statements than GE students explanations, but by Time 3, SEI students explanations comprised twice as many precise statements as GE students. b) For paleontologist, SEI students explanations comprised more precise statements than GE students at both Time 2 and Time 3, although by Time 3, SEI students had decreased their level of precision so that it was comparable to GE. c) SEI students explanations were comprised of greater partially accurate statements than GE students. Characteristics of Academic Language and Conceptual Knowledge by Learning Profiles To determine the characteristics of students conceptual knowledge and academic language in their explanations, we compared matched pairs word choice, number of propositions, type of propositions, and level of precision related to fossil and paleontologist at Time 2 to these same outcomes at Time 3. Only GE students made inaccurate statements when discussing both the concepts, but this proportion was quite

17 O Brien and Leighton 185 low; therefore, inaccurate statements were not considered when analyzing results between cases. Differences in word choice by learning profile were uneven (Tables 10 and 11). For the concept of fossil, there was no observable pattern by learning profile. For paleontologist, however, by Time 3, the low and low-average-matched pairs increased their number of relevant words and targeted words, with Maria (SEI, low) and José (SEI, low-average) outperforming Celia and Daniela (Table 6). At Time 2 when discussing the concept of fossil, there were differences in propositions and the precision of those propositions by learning profile. Average and highaverage SEI students comprised a higher proportion of full propositions and a lower proportion of partial propositions than low and low-average SEI students explanations. Low and low-average SEI students were less precise than their matched pairs with fewer precise statements and more partially accurate statements, while average and high-average SEI students were more precise than their matched pairs with more precise statements and no partially accurate statements. Closer examination of low and low-average SEI students explanations illustrates how they used language to convey their understanding. Maria (low SEI) described a picture of paleontologists excavating fossils stating, I can see a dinosaur. This initial statement is a partially accurate, partial proposition as the picture displays a dinosaur fossil, not a dinosaur. She later explained, the fossils of the dinosaur, using precise language and elaborating the concept of fossil as remains from an ancient creature. In this way, her partial proposition and partial accuracies appear to scaffold a more precise proposition in later statements. José s (low-average SEI) explanation of a complex concept associated with fossils (i.e., extinction) was less precise than Daniela s explanation of a less complex concept associated with fossils (i.e., fossil features convey attributes about dinosaurs). When explaining fossilization, instead of saying, Dinosaurs died because they were stuck in mud and couldn t find food. Over time, the mud hardened and the dinosaur remains fossilized (a more precise explanation), José said, then when rocks came, they couldn t find fish to eat; there were mud in there. Further this made the bones stick together. Daniela, on the other hand, more precisely indicated bones of dinosaurs show which dinosaurs may have been plant and meat eaters in the past. At Time 3 when explaining fossils, there were differences in precision by learning profiles as low and low-average SEI students explanations increased in precision from Time 2 to Time 3. This increase in precision is exemplified by José (lowaverage SEI), who at Time 2 stated, I see fossils. Then at Time 3 said, they re searching to find dinosaurs fossils and it s a velociraptor. Average and highaverage SEI students precision scores were comparable. At Time 2 when discussing the concept of paleontologist, most students used only full propositions regardless of learning profile. However, there were differences in precision of those statements by learning profiles: high-average students explanations comprised the most precise statements, followed by average, low-average, and low students. Differences in precision between Time 2 and Time 3 were evident for the

18 186 Literacy Research: Theory, Method, and Practice 64 low- and high-average matched pairs. Low-matched pairs increased in precision while high-average matched pairs decreased in precision. In sum, results indicate: 1. There were differences in some outcomes by learning profiles. Low and low-average SEI students demonstrated an increase in precision 6 weeks postintervention when conveying the concept fossil and low-matched pairs when conveying the concept paleontologist. This was not necessarily the case for average and high-average students. 2. Precision decreased when SEI students were explaining more complex concepts as compared to explanations of less complexity. 3. Partial propositions and partially accurate statement may have served to scaffold subsequent full propositions with greater precision. Discussion and Significance With this mixed-methods research study, our purpose was to explore instructional use of topically related and increasinglycomplextextasawaytoadvanceyoung ELs knowledge and academic language in an everyday classroom context. This investigation was prompted by evidence indicating a root of ELs poor achievement is the widespread use of simplified text (Fillmore & Fillmore, 2012). We set out to explore the use of increasingly complex expository text joined with sound instructional practices as a way to bolster ELs conceptual knowledge and academic language development. To do so, we studied two groups of first-grade children: 10 SEI children who received a 5-week read-aloud intervention using increasingly complex text and 31 GE children who received the same intervention using texts of comparable complexity. We compared gains in receptive knowledge of academic vocabulary and conceptual knowledge and academic language outcomes of SEI and GE children within 1 week (Time 2) of receiving a 5-week read-aloud intervention. We also compared these same outcomes at Times 2 and 3 for eight illustrative cases. Taken together, findings suggest joining increasingly complex and topically related text with sound instructional practice helped to advance the academic language and conceptual knowledge for SEI students we studied, as by 6 weeks postintervention, SEI students academic language and conceptual knowledge closely approximated that of their GE peers. This was evident in nonsignificant differences in receptive academic vocabulary between groups at Time 2 as well as comparable proportions of full and partial propositions at Times 2 and 3 for eight illustrative cases. Moreover, although (within the illustrative case data set) GE students demonstrated greater expressive academic vocabulary knowledge, virtually no differences between SEI and GE students were observed at Time 3. For fossils, SEI students explanations were less precise than those of GE students at Time 2, but by Time

19 O Brien and Leighton 187 3, precision increased to two times beyond that of GE students. Differences were observed in partially accurate statements as more of SEI students statements were partially accurate than GE students. To understand and explain these outcomes, we turned first to theory of second language acquisition. Krashen (1982) explains that ELs go through a silent period in which they are listening and understanding language. As they develop competency in a new language, spoken language emerges. Our findings suggest that the EL students we studied also went through a silent period in which additional time to develop overall language proficiency permitted them to convey their understanding. Although at Time 2, SEI students produced explanations that were somewhat less developed than their GE peers, at Time 3, they were comparable. Our anecdotal observations and additional case study findings lend support to this interpretation. During lesson cycles, SEI students tended to engage far less in discussion during the first reading of a text than the second reading. Moreover, SEI students with less-developed vocabulary knowledge (low and low-average students) demonstrated the largest increase in precision from Time 2 to Time 3. These data suggest that providing ELs with time is essential for ELs to convey their understanding, particularly for those with the lowest initial English abilities. We also acknowledge that time alone may not explain effect evident in the present study. As suggested by our anecdotal observations during the intervention (e.g., students self-selected topically related text for independent reading), it may also have been the case that they sought out opportunities beyond the intervention that also extended their knowledge and language growth. In addition to learning profiles, within the case study data set, we also found that some outcomes varied by EL status and concepts (both topic and complexity). SEI students relied much more on target words and less on proxies to convey their understanding than GE children. It may have been that because SEI students were developing conceptual knowledge for abstract concepts far removed from everyday experiences and were also developing everyday language, they had to rely on target words to convey their understanding (not everyday language) in their explanations. Another possible explanation may reside in the nature of the students. Because SEI students were actively acquiring English vocabulary and their classroom teacher emphasized learning English, SEI students may have been predisposed to focus on learning words i.e., they were word conscious. Through our anecdotal observations, we noted that with each new lesson, SEI students were quite eager to see new vocabulary words, while GE students did not display this same level of enthusiasm. These findings suggest that ELs can and aspire to acquire complex concepts and academic vocabulary representing those concepts; as such there is no need to withhold such concepts and language as part of their curriculum. With respect to differing concepts, we found that most students achieved better outcomes with the concept of fossils than paleontologist. We see two interrelated

20 188 Literacy Research: Theory, Method, and Practice 64 explanations for these differences. First is in the nature of the intervention. The expository texts and questions used to scaffold discussion in the intervention focused more explicitly on factual details related to fossils than on paleontologist. Thus, despite the fact that paleontologist was explicitly taught and frequently embedded in discussion, much of students experiences included explicit emphases on fossils. In turn, developing the concept of paleontologist was more of an implicit task that is, more challenging and less conducive to promoting conceptual knowledge depth. This leads to the second explanation students conceptual knowledge depth likely plays a role. Because most students acquired deeper knowledge of fossils (as evidenced by more propositions and greater precision in explanations of fossils than paleontologist), we suspect they were able to draw from that knowledge in their explanations, whereas this was not the case when discussing paleontologist. Relatedly, we found that when students were explaining more complex concepts, explanations were less precise as compared to explanations of less complexity. Thus, findings also suggest that concept complexity and quality of conceptual knowledge bear great importance as children are developing academic language. It may also be the case that as students develop academic language, as suggested by prior research, they initially relied on partial propositions to help them reach a precise explanation (van Dijk & Kintsch, 1983); these statements may be seen as stepping stones to helping the student fully articulate their knowledge with precision. Thus, it may be important to encourage and allow ELs to go through this process when conveying their understanding. Although future research is needed, we find the gains made by ELs we studied after a brief intervention duration (i.e., 5 weeks) encouraging, particularly in light of evidence indicating ELs need 4 to 7 years to develop academic language proficiency (Hakuta et al., 2000). Given the small sample and focus on a single topic and grade, findings must be interpreted with caution. Future research is also needed to examine the effects on ELs comprehension and the relative contributions of explicit instruction, text, and authentic writing. Notwithstanding, trends evident in the present study suggest carefully scaffolded use of increasingly complex text holds promise in advancing ELs conceptual knowledge and academic language. If replicated in larger studies, teachers will have a clear alternative to the widespread practice of addressing literacy and language differences by offering simplified texts. This alternative may help teachers level the playing field, by providing ELs access to curriculum and, in turn, the linguistic and conceptual knowledge that anchors ongoing learning and access to complex text. In doing so, we hope to lay much needed groundwork for future research and contribute to efforts that support teachers in bolstering literacy and disciplinary knowledge during EL s early school years, providing a rich foundation to anchor learning throughout their school years, contributing to higher levels of EL achievement over the long term.

21 O Brien and Leighton 189 Appendix A Sample Conceptual Knowledge Item Administration protocol. Describe what you see in this picture. Tell me everything you can about what you see. 1. Once the child stops talking, probe for additional information by saying: Can you tell me anything else? 2. After the child has either added additional information or indicated he/she is complete, use the following probes to elicit addition information under the following circumstances: 3. If the child does not talk about the fossil, point to the fossil and say: What can you tell me about this? 4. If the child does not talk about the paleontologist, point to the paleontologist and say: What can you tell me about this? 5. If the child does not refer to any tools, point to the box of tools and say: What can you tell me about this? Declaration of Conflicting Interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.