NGSI paper

A version of this paper is published as: Buxton, C. & Caswell, L. (in press). Next generation sheltered instruction to support English learners in secondary science classrooms. Science Education. Next Generation Sheltered Instruction to Support Multilingual Learners in Secondary Science Classrooms Cory A. Buxton1 & Linda Caswell2 1 Oregon State University, 2 Abt Associates Abstract Using findings from a four-year research and development effort, we propose an updated model of sheltered instruction for science classrooms that leverages the opportunities provided by the Next Generation Science Standards (NGSS) to better support multilingual learners in middle and high school science. Using data from teacher implementation logs and interviews, we examine how secondary grades science teachers’ participation in our professional learning framework and subsequent enactments of an initial set of project instructional practices led these teachers to articulate nuanced and strategic adaptations to the initial practices to better meet the needs of the multilingual learners in their classrooms. We found that teachers moved away from implementing sheltered instruction practices based on general principles or formal classifications, such as students’ ESOL level, instead taking a more nuanced approach to adopting, adapting and rejecting practices based on perceived needs, assets, and personal knowledge about individual multilingual students in the specific disciplinary context of science. Based on this analysis, we conceptualize and refine what we now see as a new and exploratory set of Next Generation Sheltered Instruction practices for integrating science and language to support multilingual learners in secondary science classrooms in the age of the NGSS. Next Generation Sheltered Instruction to Support Multilingual Learners in Secondary Science Classrooms Purpose The Next Generation Science Standards (NGSS lead states, 2013) are shifting the goals of science learning for the majority of K-12 students in U.S. schools, as most states have adopted new science standards since 2013 that are either adoptions or adaptations of the NGSS (Carnegie Corporation of New York, 2017). This transition from the previous generation of state science standards to the next generation of NGSS-aligned standards has the potential to substantively shift how scientific meaning making and communication are conceptualized in schools. The previous generation of state science standards can be typified as a combination of discrete concepts and processes of inquiry (conceptualized separately), which are expressed through academic language defined primarily in structural terms of academic vocabulary and grammar norms (Bunch, 2013). In contrast, the NGSS represent science as three-dimensional learning by blending disciplinary core ideas, science and engineering practices and crosscutting concepts, while viewing language (at least implicitly) in functional, rather than structural, terms that highlight the role of language for meaning making (Lee, 2017). Shifting how science learning is conceptualized for all students provides an opportunity to promote a parallel shift in how we conceptualize the support and resources that multilingual learners A wide variety of terms are currently in use to describe individuals who are being educated in predominantly English language settings, but whose home language is other than English. These terms include English language learners, English learners, and emergent bilingual learners, among others. We have likewise used various terms over time as this young field of study continues to evolve. We have currently settled on the term multilingual learners because, for us, it represents a non-hierarchical perspective that, as learners, we always flexibly use the full range of language resources we have available to make and communicate meaning. Our language skills are always emerging and growing, so adding the term “emergent” to multilingual learners seems unnecessary. We have further decided to avoid the acronyms that are also commonly in use (EL, ELL, EBL, etc.) as we have come to view these as potentially dehumanizing through their overuse. need to thrive in the science classroom. Sheltered instruction (e.g., Echevarria, Vogt & Short, 2016), which largely aims to simplify and concretize the language and examples that teachers use, has been the dominant approach to supporting content area teaching in linguistically diverse classrooms since the requirements of the No Child Left Behind Act (NCLB, 2001) focused greater attention on all students receiving grade-appropriate content instruction. We argue that new frameworks of linguistic support and enrichment, tied more explicitly to the new goals of science learning under NGSS, are needed to help science teachers think differently about the language affordances and demands inherent to science and to promote new strategies for building on the linguistic assets that all students bring to the science classroom. However, simultaneously reconceptualizing both the science content and the language of science raises a range of challenges for both students and teachers. While a number of the strategies and practices advocated in the NGSS as beneficial for all students can, indeed, prove helpful to multilingual learners and other students traditionally marginalized in science education, there are unique assets and challenges that multilingual learners bring to the science classroom that require differentiated attention on the part of teachers (National Academies of Sciences, 2018). Further, as we have written elsewhere (Authors, 2018), lessons from efforts to support multilingual learners in science classrooms can enrich science learning for all students. Thus, the current reform context offers a unique opportunity to transform inclusive classroom practices in ways that can broaden science participation through robust classroom engagement of all learners (Ferrini-Mundy, 2013). In this paper, we use findings from a four-year research and development effort, which began with one year of co-planning with teachers followed by three years of data collection, to consider the instructional choices that secondary grades science teachers made as they enacted our project’s initial set of instructional practices in support of their multilingual learners. These initial instructional practices had been co-developed in a prior project that predated NGSS. The state in which both projects took place was one of the 24 states that have not adopted the NGSS but that enacted new K-12 science standards that were adapted from the NGSS. In this paper, we examine how teachers’ participation in our professional learning framework influenced their instructional choices as their ideas (and ours) about the relationship between science and language evolved. Building on our analyses of teacher enactment logs and interviews as teachers adapted our initial instructional practices, we use this study to propose a new and exploratory set of practices for integrating science and language to support multilingual learners in secondary science classrooms in the age of NGSS; practices we refer to as Next Generation Sheltered Instruction (or NGSI). Specifically, we address the following research questions: What choices around our project’s initial instructional practices did secondary grades science teachers enact to foster science meaning making with their multilingual learners and all students? What new instructional practices for integrating science and language for multilingual learners in secondary science classrooms can be proposed based on our project teachers’ instructional adaptations in response both to new standards and to their own professional learning? Content and Language Integration As we conceptualized our approach to supporting secondary science teachers working with multilingual learners, we grappled with ongoing debates related to content and language integration. One of these debates has to do with the question of language registers and the role of “academic language” in school learning (Walldén, 2019; Cummins, 2000). Our use of the term register is distinct from the more general focus on the formality or informality of discourse (e.g., academic versus colloquial register). We consider register to be the context of communicative situations that includes a topic (field), audience (tenor) and type of communication (mode) among participants (Halliday & Hasan, 2006). With this understanding, we can see that multiple registers are required in the NGSS classroom as the nature of interactions and topics change (e.g., from direct negotiation with lab partners during a science investigation, to explanation of a phenomenon in response to a teacher’s question, to a written lab report summarizing findings). In most school contexts, however, the more common binary view of register holds, in which colloquial language registers are viewed as acceptable for non-academic purposes but are viewed as unsatisfactory and in need of replacement for academic purposes (Harman & Khote, 2018). We find that a restricted view of language register is at the heart of confusion about what counts as academic language, how much of this language is content specific as opposed to generically academic, and how much class time, in a content course such as science, should be dedicated to explicit instruction around academic language (Duff, 2010). For example, Krashen and Mason (2015) argue that explicit teaching of academic language is both misguided and unnecessary because the most effective way to develop any linguistic register is to read and talk about high interest materials that exemplify the use of that register. This perspective seems to align with ideas promoted by some proponents of the NGSS, who argue that students will acquire the language of science implicitly when given access to rich experiences with science phenomena that reform-based science teaching affords (e.g., Stage, Asturias, Cheuk, Daro & Hampton, 2013). From a different perspective, Flores and Rosa (2015) argue that explicit academic language instruction is misguided, because no matter how much academic language in English that multilingual learners adopt, their language will still be seen as deficient due to raciolinguistic norms that hold sway in the U.S. education system and society broadly. While we agree that language develops through use, need, and interest, and that NGSS-based instruction can certainly support these goals, we see a parallel here to the whole language versus phonics debates that took place in the teaching of reading (Goodman, 2014). While whole language approaches have been shown to work well for some learners, especially those with access to more robust reading resources outside of school (Cook, 2013), most multilingual learners benefit from modest amounts of more structured and explicit reading instruction (Cheung & Slavin, 2012). If this parallel holds for the process of learning the language of science, we should infer that multilingual learners will benefit from a certain amount of explicit focus on functional ways that the language of science is used to make meaning. For this to occur, however, both teachers and students need to consider the relationship between science and language in new ways, including gaining an awareness of raciolinguistic norms. Supporting Teachers in Rethinking the Relationship Between Science and Language It has long been known that large-scale changes in student learning goals, such as those changes advocated in the NGSS, require new visions of support and guidance for teachers of all experience levels (e.g., Ball & Cohen, 1999; Borko, 2004). In recent years, several frameworks have been proposed to specifically aid science teachers in working more effectively with multilingual learners. One model, proposed by Bunch (2013), focuses on the purposeful integration of language and literacy into disciplinary content instruction, supporting what Bunch refers to as science teachers’ pedagogical language knowledge (PLK). Bunch argues that while there is broad agreement that teachers of multilingual learners require new and deeper understandings about language, there is less agreement about what the exact nature of this enhanced linguistic understanding entails, as well as how it can best be developed. He proposes that the language demands associated with a given set of disciplinary standards can only be understood by first looking at the key practices that are central to those standards. Thus, in the case of the NGSS, this means that there are specific linguistic understandings that teacher must develop around the three dimensions of science learning that would be a central part of developing science PLK. Bunch calls on the field to propose, test, and refine new models to support science teachers in developing their science PLK, and along with colleagues has been engaging in such an effort (Lyon, Tolbert, Stoddart, Solis & Bunch, 2016). In another framework for improving the preparation of content area teachers working with multilingual learners, Lucas and Villegas (2013) proposed a model for the preparation of linguistically responsive teachers. They concluded that even experienced teachers need a number of years to develop the knowledge and skills to support linguistically responsive teaching in secondary content areas, such as science. More specifically, for Lucas and Villegas, linguistically responsive science teachers need to develop: (a) multiple ways of learning about and building on their students’ linguistic backgrounds; (b) current understandings of second language acquisition; (c) the ability to identify the specific language demands of a given classroom task; and (d) multiple ways of scaffolding content-specific language supports for multilingual learners. However, as Johnson, Bolshakova & Waldron (2016) point out, large scale efforts to support science teachers in developing these types of linguistically responsive skills must overcome a host of potential obstacles ranging from culture to resource allocation to assessment priorities. Finally, as a third example, Turkan, De Oliveira, Lee and Phelps (2014), developed an analytic framework they refer to as disciplinary linguistic knowledge (DLK) to describe the teacher knowledge base needed to apply current understandings of the role of language in teaching disciplinary content. More specifically, DLK highlights two components of teachers’ knowledge about disciplinary discourses, such as the language of science: (1) teachers' ability to identify the linguistic features most salient to the discipline, and (2) teachers' ability to model how to communicate disciplinary meaning through engaging students in using the language of that discipline. Similarly, MacDonald, Miller & Lord (2017) have focused explicitly on strategies for changing science classroom interaction patterns to be inclusive of multilingual learners when shaping opportunities for students to do and talk science. In sum, there is a growing consensus among researchers that the integration of disciplinary practices and classroom discourse must be central to the work of all secondary grades content area teachers, yet in the case of science teaching with multilingual learners, models for how to develop these skills remain emergent (Lee, Quinn, & Valdés, 2013; Tong et al., 2019). The [XX] Project attempted to respond to this challenge through a design-based implementation research project to develop, test, and refine a teacher professional learning framework and a pedagogical model for supporting teachers in the dual shifts of implementing new state science standards in classrooms with increasing numbers of multilingual learners. The project professional learning framework engaged teachers in five distinct professional learning contexts, each of which focused attention on different aspects of how to meet the needs and build on the assets of multilingual learners in ways that were specific to supporting meaning making in science. As elaborated in more detail elsewhere (Authors, 2016a), these professional learning contexts were: (1) An annual week-long Teacher Professional Learning Institute focused on negotiating common understandings of science and language using our initial pedagogical model; (2) An annual two-week-long Student Summer Enrichment Academy for multilingual learners focused on dynamically experimenting with the pedagogical model learned during the teacher institute; (3) A series of annual Steps to College through Science Bilingual Family Workshops focused on teachers, students, and families doing science together and learning about STEM careers; (4) A series of annual Teachers Exploring Student Writing Workshops focused on understanding students’ emergent science meaning making through analyzing their writing; and (5) On-demand classroom support when requested by teachers, focused on co-planning and co-teaching by project staff to integrate the project pedagogical model into secondary science classrooms. The components of the professional learning framework were co-developed with teachers during a prior iteration of this project with attention to the broader literature on the components of effective teacher professional learning (e.g., Desimone, 2011). Central to our pedagogical approach was a rejection of traditional views of fidelity of implementation, in favor of encouraging teachers to make their own professional leaning choices and then studying the implications of those choices through a framework we referred to as multiplicities of enactments (Authors, 2015). The project pedagogical model included six instructional practices that likewise emerged from a combination of our prior research with science teachers of multilingual learners (Authors, 2013) and the broader literature focused on developing the meaning making and communication skills needed to attain academic success in science (e.g., Kuhn, 2005). Brief descriptions of these initial practices that were developed prior to the NGSS are listed in Table 1, with more details on the development of these practices described elsewhere (Authors, 2016a). [Insert table 1 about here] Here, we briefly note three things about the six initial instructional practices in our pedagogical model that are relevant to the present study. First, as described above, a central aspect of our broader research on teacher professional development was to explore teacher agency by empowering teachers to make their own choices about which professional learning contexts to engage in, as well as when and how to enact the six initial instructional practices. Teachers then documented those choices and discussed them with the research team as part of the study. Second, while there is substantive overlap between the six initial instructional practices and the science and engineering practices of the NGSS, these instructional practices did not align seamlessly, as our model predated the release of the Framework for K-12 Science Education (National Research Council, 2012). Because the state where this work occurred did not adopt the NGSS, teachers who advised us during the planning year of the [XX] Project wished us to maintain our existing framework of initial instructional practices. Given our focus on teacher agency, we sought to follow their lead when possible and then study how the model developed over time through participatory collaboration. Finally, while our goal was always to help teachers to more fully integrate their thinking about science and language in their classrooms, and while we initially viewed each of our six instructional practices as supporting this intersection, we came to see over time that our instructional practices did not, in fact, fully integrate science and language. Teachers in the project came to view the first four instructional practices as highlighting aspects of scientific thinking (with language goals embedded) and the last two instructional practices as centering aspects of language (with science goals embedded). Thus, teachers often referred to the pedagogical model as having four science instructional practices and two language instructional practices, even though this was not our intention. It was this emerging realization that led us to pursue the present study. Sheltered Instruction and Secondary Science Teaching Thus far, we have argued that the new wave of NGSS-inspired state science standards may promote more robust use of the language of science through the three-dimensional emphasis on disciplinary core ideas, science and engineering practices, and cross cutting concepts, but that this outcome is not inevitable (or even likely) for multilingual learners without explicit classroom attention to how science meaning is made and supported through language. Sheltered instruction, which has the goal of simultaneously supporting content and language learning for multilingual learners, remains the most common model of ESOL support used to prepare secondary grades content teachers (Schleppegrell & O'Hallaron, 2011). Sheltered instruction builds on Krashen’s (1985) theory of comprehensible input, highlighting the ability of multilingual learners to engage in grade appropriate academic content when the language of instruction is modified to make it more comprehensible, rather than assuming that students must master English prior to engaging in grade-level content area learning. Sheltered instruction is best known to many educators as the basis for the widely adopted Sheltered Instruction Observation Protocol (SIOP) model (Echevarria, Vogt, & Short, 2016) for supporting multilingual learners in content area classrooms. The SIOP model combines instructional strategies that could be considered best practices for all learners in all subjects (such as cooperative learning and differentiation) with strategies that have been identified as particularly useful for supporting multilingual learners (such as attending to usage of the various language modes [speaking, writing, drawing, gesturing, etc.]). As the SIOP model has gained popularity, proponents have attempted to further differentiate it for disciplinary content while also arguing for its broad applicability across contexts (Kareva & Echevarria, 2013). Thus, when applied to science teaching, SIOP training has emphasized strategies such as: setting language objectives as well as science content objectives for every lesson; highlighting a limited number of key disciplinary vocabulary terms for every lesson; and providing authentic opportunities for students to use language to make and share meaning of science concepts (Short, Vogt & Echevarria, 2010). While research on SIOP has yielded a number of positive findings (Echevarria, Vogt, & Short, 2016), some of the underlying assumptions of this model, such as the binary distinction between basic interpersonal communication skills (BICS) and cognitive academic language proficiency (CALP) (Cummins, 2008), as well as the assumption that the language of instruction can be simplified without affecting the underlying meaning, have been broadly critiqued as not representing current thinking about content area language acquisition (e.g., Crawford & Reyes, 2015; Krashen, 2013). Thus, as we worked with secondary science teachers over multiple years, we came to see that one of the contributions our research could make would be to rethink and update the notion of sheltered instruction for science to bring it in line with more current theoretical understandings of the relationship between science and language in the age of NGSS. By examining the instructional choices of teachers in the XX Project, their rationales for those choices, and our own evolving understandings of the language of science, we gradually came to reconceptualize and replace our initial pedagogical model of six instructional practices with a new exploratory set of instructional practices we wish to propose as Next Generation Sheltered Instruction (NGSI) for science. Method Participants All of the participating teachers in this study taught STEM courses at the middle or high school level in one of two districts in the southeastern United States. The number of participating teachers in the project varied by year, since teachers were given the choice of continuing to participate or not at the beginning of each school year. Year zero was a planning and development year, followed by three years of project implementation with corresponding research activities. The result was that teachers participated in one, two or three years of the project. During Year 1, 23 teachers participated in the project. In Year 2, 11 teachers continued from Year 1 and 9 new teachers began participating, for a total of 20 teachers. In Year 3, 7 of the Year 1 teachers and 3 of the Year 2 teachers continued to participate and 4 new teachers began participating, for a total of 14 teachers. All teachers completed teacher logs throughout the years of their participation and a subset took part in other research activities, such as the focus groups and follow-up interviews, as described below. Data Sources and Data Collection This paper is based on data from three sources, collected at different points during the three-year implementation of the [XX] Project: (1) teacher log data collected from all participating teachers in each of the three years; (2) focus group interviews conducted each year with teachers who attended the annual teacher institute; and (3) follow-up interviews with a subset of teachers at the end of the project. We note that while these three data sources all involve teachers’ self-reported data, the larger project included student assessment data as well as observations from the Student Summer Enrichment Academy. While our prior exploratory project included classroom observations during the regular school year, we abandoned those observations in the [XX] Project in favor of the teacher logs, both due to the larger number of teachers in the project, and at the request of the teachers, who were feeling a great deal of stress from high-stakes district-level observations that were simultaneously being implemented. Teacher log data. The teacher log was developed specifically for the [XX] Project to capture teachers’ use of the six initial instructional practices from the pedagogical model in their science classrooms for the three years of the project, from 2014-2017 (Authors, 2016b). Although other methods of studying teachers’ use of the initial instructional practices were considered, such as classroom observations, the team decided to implement teacher logs because of the ability to capture complexity and variability across a school year, via near-real-time recording. Another benefit of logs was the ease with which they could be implemented for large numbers of teachers. The development of the teacher log was iterative and collaborative, as we piloted successive versions and incorporated feedback from pilot teachers into the final version to increase usability, maximize response rates, and ensure that both teachers and program developers felt that the log was adequately capturing classroom practice. Although the teacher log has four parts, The other parts of the log asked more specific questions about teachers’ use of scaffolding resources and technologies while implementing the instructional practices, and whether students wrote about, read about or talked about the science that they were engaging with when using the practices. (See Authors, 2016a and Authors, 2016b for detailed descriptions of these parts of the log.) this paper uses data from Part A, which asked teachers about the six initial instructional practices of our pedagogical model. As seen in Figure 1 below, the log asked teachers specifically about whether their students had the opportunity to participate in from two to three exemplary strategies within each of the six instructional practices during the prior week (See Authors, 2016a and Authors, 2016b for more detailed descriptions of exemplary strategies within each of the six initial instructional practices). [Insert figure 1 about here] Based on the goal of obtaining as many logs from teachers as possible over the school year, balanced against what pilot teachers reported was feasible, teachers were asked to complete one log, covering one week’s worth of classes, on alternating weeks of the school year using an online platform. If all logs were filled out, this would amount to a total of eighteen logs per year, which would leave some flexibility for missing data while still ensuring we collected a sufficient number of logs to provide a reasonable picture of teachers’ classroom practice. It has been estimated that collecting 10 - 20 logs evenly spaced across an intervention period such as a school year is necessary to reliably discriminate among individual instructional behaviors (Camburn & Barnes, 2004; Correnti & Rowan, 2007). Teacher response rates across the three years were fairly consistent, ranging from 71 percent in years one and two to 75 percent in year three. The average number of logs filled out in years one to three of the project was also fairly consistent – 12, 12 and 14, respectively. In this paper, the teacher log data from Part A are used to address our first research question: What choices around our project’s initial instructional practices did secondary grades science teachers enact to foster science meaning making with their multilingual learners and all students? Focus group interviews. Focus group interviews were conducted annually as part of the Teacher Institute. All teachers in attendance participated in the focus groups, which were typically 90 minutes in duration. During the first and second years, the focus groups were organized by school teams, with teams ranging from two to seven participants. During the third and final year of the project, we organized the focus groups by science content area to explore whether this perspective would provide any new insights. A total of 18 focus group interviews were conducted over the three years. The focus group interview protocol consisted of 20 questions, divided into four categories on the topics of: scientific investigation, the language of science, teaching multilingual learners, and assessing science learning. The focus group interviews were facilitated by members of the research team. Interviews were audio-recorded and each interviewer took notes regarding context and non-verbal communication. Follow-up interviews. At the end of the final year of the project, we conducted a supplemental interview study with a small, purposeful sample of teachers (n = 6) who had been engaged in the project for at least two years and who had high percentages of multilingual learners in their classrooms. The goal of these follow-up interviews was to explore in greater depth whether teachers were adapting or modifying their use of the pedagogical model in intentional ways based on their knowledge about their specific multilingual learners in their classes. We developed a semi-structured interview protocol based on the topics addressed in the teacher log, and we asked teachers if they had modified their instruction for their specific multilingual learners, and if so, in what ways. We used each teacher’s individual log data to determine their most commonly reported instructional practices, and we focused our questions on if and how they differentiated these instructional practices for their multilingual learners. Prior to each interview, we sent the teacher a graphic representation of their log data so that these graphs could be referred to during the interview. Individual interviews were conducted by one of four interviewers and lasted about one hour each. Interviews were audio-recorded and contextual notes were taken. In this paper, the teacher focus group interviews and the follow-up interviews were used to address our second research question: What new instructional practices for integrating science and language for multilingual learners in secondary science classrooms can be proposed based on our project teachers’ instructional adaptations in response both to new standards and to their own professional learning? Data Analysis Teacher Log Data. Part A of the teacher log provides detailed descriptive information about teachers’ use of the six initial instructional practices in our pedagogical model. As shown in Figure 1 above, the log listed three exemplary strategies for two of the six initial [XX] instructional practices: 1) Coordinate hypotheses, observations and evidence, and 2) Own the language of science; and two exemplary strategies for the four remaining initial [XX] instructional practices: 1) Learn about controlling variables to design a fair test; 2) Explain cause & effect relationships; 3) Use models to construct scientific explanations and test engineering designs; and 4) Develop academic vocabulary in context. For each year, we summarized teachers’ level of implementation for each of the six instructional practices using the following process. First, we coded each of the exemplary strategies as 1, 0 (yes, no) within each of the six instructional practices for each of a teacher’s logs. We then assigned a value to each of the six instructional strategies for each of a teacher’s logs as shown in Table 2 below. [Insert table 2 about here] Next, we calculated the average score for each instructional practice across all of a teacher’s logs. This average score ranged from 0 to 2 or 0 to 3, depending on the number of exemplary strategies within the instructional practice. We then categorized teachers based on their average score high, medium, or low implementers for the instructional practices made up of three exemplary strategies (COORD and OWNSC) and as high or low implementers for the instructional practices with only two exemplary strategies (CONTV, CAUEFF, MODEL, and ACADVOC). We developed the implementation levels by examining the distributions of the sample in the first year. For 3-level categories, low is less than 1; medium ranges from 1 to 2; and high is greater than 2. For 2-level categories, low is less than 1 and high is 1 or above. Thus, each teacher received an implementation level for each of the six initial [XX] instructional practices for the year. We used the same approach for the three years of teacher log data. We used the results of these analyses for two purposes. First, we used the results to track teachers’ use of the six instructional practices over the course of each year of project participation. Second, we produced individual graphs for each teacher to provide them with an efficient and easily understandable summary of their self-reported enactment of the project instructional practices. Teachers reviewed their results during the annual teacher institute and this review led to productive professional learning conversations about the reasons why teachers were making these choices, their interpretations of the project instructional practices, and possible modifications to the pedagogical model. The six teachers who participated in the follow-up interviews also used their log results to respond to interview prompts. Focus Group Interviews. Each teacher focus group interview was transcribed by the same member of the research team who conducted the interview. Interview transcripts were then analyzed through a hybrid process of inductive (data-driven) and deductive (theory-driven) thematic coding (Fereday & Muir-Cochrane, 2006). Coding of the focus groups was done by two members of the research team, who met repeatedly to discuss codes and reconcile differences. The types of coding included initial coding, focused coding, and theoretical coding. As a first step, we performed initial incident coding, labeling segments of data by incident (coding events). Examples of such initial coding included: think-pair-share time; bell-ringer activities; family workshop conversations; word wall review; two-way re-writing; direct translation; assigning lab roles; lunch bunch; class science talk; concept card review; teacher institute planning; language boosters; etc. As the second step, we used focused coding to refine our initial codes as we synthesized and analyzed the initially coded segments. We merged our overlapping codes to generate a set of mutually exclusive focused codes. Examples of these focused codes included: lab group strategies; professional learning insights; talk moves; translanguaging; semantic waving; register shifting; co-planning; funds of knowledge; etc. Third, we grouped our focused codes to develop categories (code groups) related to our phenomenon of interest. Eleven categories were initially generated as a result of this process, and these categories provided us with the starting point for conceptualizing our new model of NGSI practices, as discussed in the findings section below. Finally, we examined how our inductive analytic categories aligned with the theoretical perspectives discussed earlier in this paper, and this deductive analysis allowed us to further revise our categories, and in three cases, collapse categories as we compared our inductive categories with the relevant research literature. We checked that our interpretations captured participants’ viewpoints by sharing an analytic summary for feedback with one participant from each focus group after coding was complete. In one case, a participant raised questions about our interpretations, leading to a follow-up conversation to clarify and reach agreement about our interpretations. This analysis of the focus group interviews was then combined with the parallel analysis of the follow-up interviews. Follow-up Interviews. Similar to the focus group interviews, the follow-up interviews were analyzed using a combination of inductive and deductive coding (Elo et al., 2014; Fereday & Muir-Cochrane, 2006). In the first stage of analysis, the transcripts were coded in terms of the types of adaptations or modifications that teachers reported using for the multilingual learners in their classrooms. The four interviewers coded the individual teacher interviews in terms of how the teachers described the project’s instructional practices. In analysis meetings, the four interviewers and the note taker, who was familiar with all of the interviews, mapped the individual teacher results into tables to facilitate seeing patterns within and across teachers. The team brought evidence to bear from each teacher’s interview to develop common themes and to highlight relevant findings. Individual interviewers wrote up different sections, which were then synthesized by the team lead into a cohesive set of findings. One coder from the follow-up interviews then worked with one coder from the focus group interviews to conduct a second level of analysis for this paper. We categorized the findings from the first analysis of the follow-up interviews (i.e., types of modifications that teachers made for the multilingual learners) into the categories of instructional practices that came from the analysis of the focus group interviews, which finally led to the current set of NGSI practices described in the findings. Findings We begin by reporting on the findings from the teacher logs to answer our first research question: What choices around our project’s initial instructional practices did secondary grades science teachers enact to foster science meaning making with their multilingual learners and all students? Teachers were consistent implementers of project initial instructional practices As described above, the project teacher log asked teachers to track how often they engaged their students in tasks that they felt were aligned with the goals and instructional practices of the [XX] Project. In this section, we present the findings for the three years of teachers’ reported use of these practices to set the context for the remainder of the findings and the discussion. Figure 2 presents the percentage of teachers who had low, medium, and high implementation levels each year for each of the six project practices. The overall patterns of implementation are similar across all years, with the most differences in Year 2 compared to the other two years. Overall, the findings indicate clear patterns relevant to our first research question. First, for two of the four science investigation practices (Explain cause and effect relationships; Use models to construct scientific explanations and test engineering designs), the majority of teachers were high implementers based on the thresholds we set (see above). For the practice Coordinate hypotheses, observations and evidence, the majority of teachers were medium or high implementers. The majority of teachers were low implementers for the practice, Learn about controlling variables to design a fair test. For the two language of science practices, almost all teachers were high implementers of Develop academic vocabulary in context, while for the practice, Own the language of science, the majority of teachers were medium or high implementers. Thus, with the exception of the Learn about controlling variables practice, the majority of teachers implemented the initial [XX] instructional practices at medium to high levels, consistently across their years of project participation. We note that while we also examined teachers’ log data longitudinally to test for changes over time, we did not find statistically significant changes (Authors, 2016b). [insert figure 2 about here] While the teacher logs were fruitful for seeing patterns in teachers’ implementation of the project practices, and for helping the teachers themselves to reflect on their implementation over time and compared to their peers, the log data alone did not allow us to explore how or under what circumstances the teachers implemented these practices. The logs also could not tell us how the teachers attempted to integrate the science and language practices, nor how the teachers adapted the practices to the specific needs of their multilingual learners. This interpretive work was the goal of our subsequent analyses of the teacher focus group interviews and follow-up interview data. These data allowed us to answer our second research question: What new instructional practices for integrating science and language for multilingual learners in secondary science classrooms can be proposed based on our project teachers’ instructional adaptations in response both to new standards and to their own professional learning? Teachers reimaged sheltered instruction for the NGSS era As we analyzed the teacher focus groups and follow-up interviews, we began to distinguish between those instructional practices the teachers described that were informed by older theories of support for multilingual learners engaging in science learning, and those practices that seemed to be informed by newer theories of disciplinary language development to support science meaning making. We refer to the former set of practices as “first generation sheltered instruction” and the later as “next generation sheltered instruction.” We came to view first generation sheltered instruction as being informed by the earlier work in second language acquisition (e.g., Cummins, 1979) and in content area literacy (e.g., Chamot & O’Malley, 1994) that highlighted strategies for making abstract academic concepts more concrete so as to make these concepts accessible to multilingual learners. Practices that are representative of these older perspectives on supporting multilingual learners in science instruction are listed in the left column of figure 3. In contrast, the evolving practices that teachers described implementing during the interviews were often quite distinct from these first generation practices. These practices are listed in the right column of figure 3 in descending order of the frequency with which they were mentioned during the interviews. Those seven practices we identified as next generation sheltered instruction seem to connect to the more recent thinking about disciplinary discourse as register shifting (e.g., Gibbons & Hammond, 2005), that we described in the earlier section on rethinking the relationship between science and language. These practices tended to highlight intentional shifting across language registers (defined in terms of topic, purpose, and audience) throughout a lesson, such as by using both abstract and concrete language at different points, as needed for disciplinary meaning making. While not perfectly paired, we were struck by the matches we were able to make across the two columns in the figure. That is, in most cases we could see clear links between practices that teachers were using at the start of the project that were common in first generation sheltered instruction, and practices that were being reimagined in the [XX] Project into examples of next generation sheltered instruction. This is not to say that teachers abandoned using practices that we categorized as first generation, but rather that they offered many more examples of next generation practices. [Insert figure 3 about here] In the following sections we provide examples of the next generation sheltered instruction (NGSI) practices that were discussed by teachers during the interviews in decreasing order of frequency. Together, these examples demonstrate how teachers learned to integrate science and language practices in new ways, in their efforts to support all students, and particularly multilingual learners, in making grade appropriate science meaning. Content meaning expressed through linguistic register choices based on topic, purpose and audience The most common NGSI practice expressed by teachers was a more intentional focus on how content meaning is expressed through linguistic register choices based on topic, purpose and audience (discussed in 17 of 18 focus group interviews), with correspondingly less emphasis on pre-teaching key vocabulary and concepts through separate language learning activities. Teachers expressed this practice in various ways, but all related to learning to intentionally shift register based on communicative purpose, topic and audience as key to helping multilingual learners express their scientific ideas. In some cases, teachers discussed communicative purpose in terms of how it varied across content areas. Claudia: We work a lot on how the language we use needs to match the purpose we have. So, like we do an evidence and evaluation model as a school wide initiative, and of course that’s relevant in science. But like we’ve talked about, it’s actually a little different in each subject. You can’t just use the same exact model. How you support a thesis in language arts is like how you support a hypothesis in science, but it’s not exactly the same. In language arts you look for textual evidence, but in science you need to make observations and think about which of your observations can be evidence for your hypothesis. So, my point is that we’re getting more sophisticated in our own thinking about how to help our students use language differently for different purposes. In other cases, teachers considered the context in which they were providing language and ideas to the students, for example how high interest readings could simultaneously support students’ scientific thinking and language acquisition more effectively than traditional deconstruction of disciplinary vocabulary or learning definitions. Sami: For me, I want to break down key words so my kids can express their ideas better, but also, I like to use a reading as an opener for each new unit, something that is catchy for the students but also something that’ll allow them to think about key words and about key concepts to introduce the big ideas they need to learn. Vocabulary can’t just be definitions to memorize like we used to do. I need to give it to the students in a meaningful context that connects to the big idea of the unit. I think the language boosters Language boosters were our project name for short, high-interest readings that teachers developed as part of each [XX] Project lesson. Language boosters had the goals of: (a) unpacking students’ funds of knowledge related to the focal concept of the lesson; (b) introducing key concepts and language in a contextualized way; and (c) providing opportunities for students to read, think, talk and write about the topic in pairs. we developed in [XX] are good for that. In still other examples, teachers expressed an increased interest in getting their students to write more as a way to both express their emergent understandings and to clarify and further develop their disciplinary thinking through writing, using scaffolded supports aligned with a given audience. Jake: Well, I’ve known for a long time that most of my students don’t write well, in general, and specifically in terms of science. And I’m not talking just about my language learners, but all the kids. But I honestly never did a whole lot about it. I figured that was for the language arts teacher to figure out. Now you’ve got me thinking more about how important writing is for getting your ideas clear. If you can write it so that it makes sense to me then it must make sense to you, right? So, I’m making a very explicit, determined effort to set up scaffolding to teach my kids frameworks to write, where they think about the audience and the purpose and the content of what they are writing. I can already see that this is a valuable approach. Thus, one of the key findings that emerged from our analysis of the interviews was that project participation helped science teachers to rethink traditional perspectives on academic language as vocabulary development to instead focus on linguistic register choices based on topic, purpose and audience. Shared experiences with natural or designed phenomena as the basis for meaning making through language The second most common NGSI practice expressed by teachers involved the need to provide shared experiences with natural or designed phenomena as the basis for meaning making through language (discussed in 15 of 18 focus group interviews), as opposed to more traditional instructional practices of teacher demonstrations or showing representations to make concepts concrete. The presentation of “realia” (real objects) has long been a foundation of sheltered instruction but is not well aligned with the goals of the NGSS, which focus on direct engagement with, and subsequent communication about phenomena. Project teachers often reflected on the shared science experiences that we created together in our various professional learning contexts. Experiences in our bilingual family workshops were unique in that the teachers played the role of co-learners along with their students and the students’ families. Lawrence: We can do more to enhance the learning for our English learners by focusing on how we communicate about the things we do together. Like in the family workshops, we share these cool science experiences and then we can talk about what we learned together. Like when we went to the fish hatchery and we learned the science together about breeding the sturgeons for caviar. I used to have a fish tank in my classroom and I would talk about how fish reproduce, but that’s very different from being together with my students in a lab experience at a fish hatchery. That’s been the best part of the whole program for me. Another way teachers talked about the role of shared science phenomena in language learning was in the context of their work with students during the summer academy, where teachers got to practice teaching with multilingual learners the investigations that they co-developed with the research team during the preceding teacher institute. Jake describes building on our work with the science of soccer that emerged from our summer academy, where soccer was the dominant free time activity. Jake: Well, soccer obviously is huge for a lot of our students and over the years I’ve tried to bring in more and more examples of soccer into my teaching for that reason. I saw it as good for motivation, but now I also see it as a way to start with a shared experience that everyone participates in. So rather than just talking about a soccer example, we actually went outside and took some soccer balls and tried out how you put spin on the ball and how that can curve the flight of the ball or how energy transfers from your foot to the ball. I’ve used videoclips of professional players too, but that wasn’t the same as everyone kicking balls and laughing together and then talking about it. Teachers also described changes in how they integrated language into shared experiences with phenomena in their classes during the academic year, based on the planning they did in the summer teacher institute. Paula: The bird seed investigation is a good example for how we used language as we built on a shared experience. The first day I just had them kind of see the materials, pick them up, talk about them and try to determine what was the best tool to pick up each food. The second day we added the amount of time as another variable. And it got competitive between the groups and they were really focusing. Some of my English learners who almost never talk, they were really engaged in the argument about which beak was best. So, we extended the activity for a third day, which I never do, but the lab was just creating such rich learning experience for the students. I used to just show pictures of the different Galapagos finches and their different beaks and the students were interested for like five minutes and then they were bored. Thus, a second key finding was that teachers’ participation in the varied professional learning contexts provided a range of experiences exploring natural and designed phenomena as the basis for science meaning making through language. This allowed teachers to rethink some of their traditional approaches, such as using demonstrations and pictures that were neither language-rich nor interest-rich for students, and how engaging all of their students more frequently with shared phenomena could better support both making and communicating grade-appropriate science meaning. Translanguaging that leverages and values all available linguistic resources to make and share meaning The third most common NGSI practice that teachers expressed was the move toward a practice of translanguaging that leverages and values all available linguistic resources to make and share meaning (discussed in 14 of 18 focus group interviews), rather than emphasizing the unidirectional movement of students’ communication from home language to English as the communicative priority. Translanguaging (García & Wei, 2014) is a theory of multilingualism that validates and supports multilingual learners in constructing meaning through dynamic borrowing among all available linguistic repertoires, while also normalizing the experiences of being and becoming multilingual. Many teachers in the project developed an increased appreciation for the academic benefit of speaking a second language, and particularly Spanish, as a resource that was useful for interpreting the language of science. Some teachers further saw how their multilingual learners gained socio-emotional benefits when their monolingual classmates came to recognize that they could learn unexpected things from their multilingual peers. Claudia: I feel like especially for somebody who’s speaking Spanish or any Latin language, since most of the scientific language is based on Latin, they have more connections they can make to understand the meaning of a science term or concept. I tell them all the time that their Spanish is an advantage for learning scientific language and they should use their Spanish to try to figure out scientific terms they don’t know. Does it sound like a word you know in Spanish? So, I see them doing that more with other Spanish speakers, but I also see them starting to relate that to other students. It’s a resource they have that they can share. And if our other students start to see the Spanish speakers as having some useful knowledge that they don’t have, that can really change the classroom culture about language and who’s good at science. Another way teachers expressed the value of translanguaging was through leveraging multilingual resources to build more flexible ways of thinking to support more successful science learning. Ashley: I think having that dual language ability is an asset for English learners because when they come across new vocabulary they have a double set of root words to draw from that helps them interpret new vocabulary. But more than that, the mental exercise of transitioning between one language and another, thinking from different perspectives, I think that’s also really beneficial in science because it helps you to be able to analyze and synthesize new information in multiple ways. Learning to function in two languages teaches mental flexibility and that flexible thinking is really helpful in science. Further, several teachers mentioned that they, too, personally valued the workshop opportunities to engage bilingually in science learning experiences and how this subsequently led to their increased comfort with encouraging translanguaging approaches in their classrooms. Trish: The bilingual family workshops have really helped me to get better at teaching and interacting with Spanish speakers. I want to increase my proficiency helping my students to use Spanish or whatever language they have in ways that can help them learn, and so the family workshops have been a great opportunity to learn to do that, and also to practice my little bit of Spanish as well. The activity we do in the workshops where we translate back and forth between English and Spanish using everyday and scientific language helped me see that all of these are helpful for learning science. I used to think that our goal was to get the kids out of Spanish and into English as fast as possible, but now I see it as, here’s all the language resources you have; how can we bring them all to the table to make sure that you learn the science? Thus, a third key finding was that teachers came to question the supposed relationship between learning English and learning science. Our collaborative work exploring translanguaging pedagogy helped to normalize multilingual and multicultural perspectives for our mostly monolingual teachers such that they came to more fully value and utilize the range of linguistic and cultural resources that their students bring to the classroom. Purposeful pairing of students based on specific science and language learning goals of the lesson During the follow-up interviews, we considered each teacher’s log data to differentiate our interview questions, highlighting the three most frequent of our initial instructional practices that each teacher reported using. For this reason, the follow-up interviews did not provide the teachers with opportunities to discuss each of the NGSI practices. Our goal was to explore how each teacher modified their most commonly used practices based on their perceptions about the specific multilingual students in their classrooms. Overall, five of the six teachers in the follow-up interviews reported that they made substantive modifications to each of the initial practices that we discussed with them. Our analysis of the modifications that the teachers described allowed us to further clarify and refine our list of emergent NGSI practices. We note that these practices were also identified within the focus group interviews, and we continue to note the number of focus groups in which each practice was discussed. Many teachers modified our initial instructional practices based on a growing awareness of the importance of purposeful pairing of students based on specific science and language learning goals of the lesson rather that generalized pairing of emergent English speakers with more fully bilingual peers for translation (discussed in 12 of 18 focus group interviews). Because encouraging paired student talk and writing was often highlighted in our professional learning contexts, teachers utilized this approach frequently, but they did so for varying purposes based on specific needs of the lesson and the individual students, rather than because this was a generally endorsed sheltered instruction strategy. For example, one teacher discussed specifically pairing up multilingual students when she wanted them to practice expressing their ideas without the pressure or influence of their native English-speaking peers stating, “Sometimes I group my English speaking students with equivalent peers and then I group my multilingual students together so they can practice with just their brains. This challenges them to write more and use academic language.” Another teacher specifically paired up multilingual students with students at higher reading levels when her goal was to try to keep her multilingual students motivated during the challenging task of reading grade level science content in a second language. I started experimenting with different pairings…sometimes I pair up highly motivated students with lower motivation students. My EL students are often lower motivation because they have seen such little success. While this statement points to a continuing deficit perspective of multilingual learners as having “lower motivation” on the one hand, it also points to an evolving understanding of how purposeful pairing of students for specific tasks can address socioemotional as well as academic needs of multilingual learners. Similarly, a teacher described pairing particular multilingual students while using technology, because some students had stronger technology skills even if they had more limited English proficiency. Another said that she paired bilingual students with some of her Spanish-dominant students as an effective way to promote questioning. I am really fortunate because my entire time here I have always had a good mix of bilingual students and I always pair them up with my EL students that tend to struggle some with the translation or don’t feel comfortable or don’t want to appear that they may not understand as much. So sometimes that keeps them from asking questions, so I would pair them up with a bilingual student. Thus, one of the key findings from our analysis of the follow-up interviews was that teachers in the project began thinking more purposefully about how to pair their multilingual learners with other students depending on particular pedagogical goals and based on their knowledge of specific needs and strengths of students as individuals, rather than generic assumptions about the needs of multilingual learners. Two directional register shifting along the continuum between colloquial and disciplinary language Another common way that teachers modified the initial instructional practices to support their multilingual learners was through the practice of two directional register shifting along the continuum between colloquial and disciplinary language (discussed in 12 of 18 focus group interviews). Within our initial instructional practice of Owning the Language of Science, one sub-practice that we often highlighted in our workshops was two-directional translating back and forth between scientific language and colloquial language. This was meant to show that teaching the language of science is not about replacing students’ colloquial language with scientific language, but rather about developing a broader linguistic repertoire that allows students to make nuanced linguistic choices based on their communicative purposes while also developing grade-appropriate meanings. Several teachers in the follow up interviews discussed ways in which they adapted the initial instructional practice of Using Academic Vocabulary in Context and combined it in new ways with two-way register shifting: I would say academic vocabulary and translating scientific language are important ones. My EL students have to think about [a scientific concept] in Spanish, translate it to English and then think about it in scientific way. Translating scientific language is the one [practice] I think has helped my students a lot. I try to use it as much as I can; I like to push my students that way. Another teacher claimed, I think for me, breaking down, translating, and using academic vocab are ones that I focus on constantly…And what I really love about [the XX Project] is telling how to use language in a scientific versus a social or formal or informal way. And doing that process with the kids is really beneficial. I have appreciated the emphasis on that. Thus, a key finding from the follow-up interviews was that teachers had come to understand the benefits of multilingual learners practicing how to move back and forth between the disciplinary language of science and less formal, more colloquial language, particularly when that colloquial language was in Spanish rather that seeing the goal as simply replacing students’ colloquial language with scientific language as much as possible. Engagement in multimodal and intertextual approaches to help students gain access to key ideas Another common way that teachers modified the initial instructional practices was aligned with our emergent NGSI practice of engagement in multimodal and intertextual approaches to help students gain access to key ideas (discussed in 10 of 18 focus group interviews). For example, teachers in the follow-up interviews mentioned that to successfully engage their multilingual students in the initial instructional practices such as Explaining Cause and Effect Relationships, they frequently gave their multilingual learners additional visuals and models (e.g., providing sequential pictures) to scaffold student meaning making and writing. One teacher mentioned that when it became apparent that certain multilingual learners in her class were not performing well on a test, she concluded that this was due to the confusing way that she was using key terms in her questions. Instead of removing or simplifying grade appropriate terms such as glacial striations, she added pictures and definitions of the terms to the assessment. She then retested those multilingual students and found that they performed much better, leading this teacher to conclude that it was “the lack of context clues rather than the concept” that her multilingual learners were missing. While a first generation sheltered instruction strategy would have been to either simplify or pre-teach the vocabulary, this teacher added more multi-modal resources to make the language and ideas more accessible. Another teacher reported that she modified graphic organizers in different ways for different multilingual students, such as by adding pictures and cloze sentences, another example of a multi-modal approach. Finally, one teacher reported using graphic organizers more frequently with her multilingual learners than with her other students, as she found this a more accessible means of expression: I use the graphic organizers more with my EL kids, with half of it blank and the half with lines, so if they can’t express it in writing, they can draw it. That gives them a way to communicate with me. And then I will help them with writing it….Whenever they do summaries, I give them the option to draw and write. In these examples we can see that teachers were modifying our initial instructional practices through the use of multi-modal approaches, considering the specific circumstances of individual students and the specific science learning goals of the lesson as they learned to avoid thinking of multilingual learners as a single, homogeneous demographic group with the same needs. Leveraging what the teacher knows about specific students’ lived experiences to help students make personal connections to content One final way that teachers modified our initial instructional practices was through their use of examples that leverage what they know about specific students’ lived experiences to help students make personal connections to content (discussed in 9 of 18 focus group interviews). In the follow up interviews, two teachers described the importance of talking to their multilingual students about their life experiences and helping students to relate those experiences to school, and particularly to science. One teacher commented that, “Science is hard if you don’t… have a lot of life experience to relate it to, it’s frustrating. I try to have conversations with students and learn about them and give them real examples.” While this comment, like a few other teacher comments we have shared, continue to raise some problematic deficit thinking about multilingual learners’ life experiences, it also highlights the evolution of this teacher’s thinking as she gained more experience working with multilingual students. In a somewhat more asset-oriented framing of a similar idea, another teacher reflected: Building a relationship is absolutely the most important thing you can do to help an English Learner student. You need to have an understanding and relationship to educate them. And showing an interest in their background is important and letting them teach you and explain it to you. This goes hand in hand with when you are talking to them later, you can use [what you know about their background] as a tool to help them understand. Thus, in our findings, we have highlighted multiple ways in which teachers in the [XX] Project were able to articulate nuanced and strategic adaptations to our initial instructional practices to better meet the needs of the multilingual learners in their classrooms. Overall, we found that teachers were moving away from implementing standardized sheltered instruction practices based on students’ formal classifications, such as ESOL level or past standardized test scores, as had often been recommended to them in prior district professional development workshops they had received. Instead, participating in the [XX] Project supported these teachers in taking a more nuanced approach to adopting, adapting and rejecting practices based on perceived needs, assets, and personal knowledge about individual multilingual students, such that these teachers more flexibly enacted language scaffolding based on science learning goals and students’ evolving ability to communicate meaning. Thus, our analysis of the teacher logs, focus groups, and follow up interviews helped us to conceptualize and refine what we now see as a new and exploratory set of Next Generation Sheltered Instruction practices. Discussion & Recommendations In the opening sections of this paper, we argued that the relationship between science and language that is presented in secondary school science classrooms must be fundamentally reconceptualized as our understandings of both science teaching and language teaching continue to evolve. Science is now conceptualized as a three-dimensional framework of disciplinary core ideas, science and engineering practices, and crosscutting concepts. Our understanding of the language of schooling has shifted as well, from a focus on the need to develop general academic language plus discipline-specific vocabulary all in English, to a focus on taking ownership of the disciplinary discourses unique to each content area framed through new understandings of language registers and modalities that value and utilize the full range of linguistic and cultural resources that students bring to the science classroom (NAS, 2018). We have argued that while these new approaches to both science teaching and language teaching have the potential to be mutually reinforcing in exciting ways, such mutualism is not an automatic outcome of a teacher holding these reform-based conceptions. Rather, science teachers need to explicitly consider how the learning environments they construct for students are structured to support these intersections in both powerful and accessible ways (Lyon et al., 2016). Further, we have argued that the changing demographics in U.S. schools, particularly the rapid increases of multilingual learners in science classrooms where these students were previously rare, provides a unique opportunity as well as a challenge. As teachers grapple with how to make science meaningful and engaging for students who are learning these ideas in a new language, teachers may well learn to reach all of their science students in new and powerful ways (Authors, 2018). One of our teachers perfectly expressed the tensions we have noticed with many teachers, hinting at past and perhaps residual deficit views of multilingual learners, while also attempting to rise to the challenge of meeting the learning needs of a changing student body within a changing society within a changing world, “Our English Learners are a blessing in disguise. What I’ve tried to learn to be a better teacher for them has made me a better teacher, period.” Like this teacher, our work on the [XX] Project continuously pushed our thinking about new ways to more fully integrate science and language. Looking back to the start of the project, we can see that we were still conceptualizing science practices and language practices as fairly distinct. For this reason, our initial pedagogical model that guided our professional learning framework consisted of four practices that focused primarily on conducting science investigations (with language affordances embedded) and two other practices that focused explicitly on developing new language abilities to support the science investigation practices. As conceptualized and codified in our teacher log, we were presenting these initial practices to teachers as related, but not as fully integrated. Now at the end of our project, our analysis of how teachers worked with us to reimagine and adapt these practices leads us to propose a new and exploratory set of Next Generation Sheltered Instruction practices (as presented in figure 3), in the hope that these can be further iterated and refined through additional classroom testing. Our first research question asked what instructional choices secondary grades science teachers in the [XX] Project made to foster science meaning making with their multilingual learners and all students. This question was important to our thinking about the integration of science and language because a fundamental principle of our professional learning framework is that teachers should have the agency to make their own decisions about how to enact project practices in ways that they believe best meet the needs of their students. We refer to this approach as embracing a multiplicity of enactments rather than seeking or expecting fidelity of implementation of project practices (Authors, 2015). Given this professional trust and freedom, it was an open question as to which project practices teachers would enact and how. As we found from the teacher log data, there was a fair degree of consistency across teachers and years as to which practices teachers enacted and which what frequency. Over time, the teacher log data helped point us toward new instructional practices that highlighted the importance of more flexibly scaffolding language based on the science learning goals of the lesson and specific students’ evolving ability to communicate meaning (MacDonald, Miller & Lord, 2017). Our second research question asked specifically about what new instructional practices for integrating science and language with multilingual learners could be proposed based on the project teachers’ instructional choices. While the teacher logs allowed us to track the degree of implementation of the initial project practices for each teacher, the focus group and follow-up interviews allowed us to reimagine those practices based on teachers’ instructional adaptations when teaching reform-based science to multilingual learners while also engaging with the research team through a range of different professional learning opportunities. Through our collaborative work as researchers and teachers, we all came to see that we needed to do more to help students fully integrate science practices and language practices. In particular, our increased exploration of and reliance on ideas from systemic functional linguistics (Halliday & Hasan, 2006) helped us to better articulate how language is not just the tool through which we express our understanding of science, but also, that is it through language that we make meaning of our experience engaging in science. This conceptual shift had clear instructional implications for our work. For example, we came to see, much like Harman and Khote (2018), that disciplinary meaning making was often enhanced by a flexible combination of colloquial language and discipline-specific academic language, shifting as needed in response to the topics, purposes, and audiences. Our analysis of teacher logs, focus groups, and follow up interviews supported our growing awareness of the need to represent the integration of science and language practices in ways that build on newer understandings from applied linguistics and led us, in turn, to develop our exploratory set of NGSI principles. Our collaborative work with teachers gave us much of the guidance for how we could move beyond older models of practice that keep language activities conceptually separate from content learning. Such practices, which emerged from the work on second language acquisition and ESOL instruction that began in the 1980s, still provide much of the conceptual underpinning for current teacher professional development around sheltered instructional practices such as the SIOP model (Echevarria et al., 2016) and the separation of basic interpersonal communication skills (BICS) and cognitive academic language proficiency (CALP) (Cummins, 2008). These first generation approaches to sheltered instruction are largely agnostic about disciplinary content, advocating much the same general methods across all content areas and are overly binary around distinctions such as academic language/ everyday language and home language/ second language. There is a small but growing research base of new models and approaches for more explicitly connecting support for learning science concepts with support for learning and applying the language of science with multilingual learners (e.g., Johnson, Bolshakova & Waldron, 2016; Lyon, et at., 2016), to which we add the present study. As Lee (2018) has articulated, however, there is still much work to do in order to align the English language proficiency (ELP) standards that guide content area instruction of multilingual learners (e.g., WIDA, ELPA21) with current frameworks for disciplinary learning (e.g., NGSS). Multiple challenges arise when attempting to create an agreed upon set of ELP standards that simultaneously address grade appropriate disciplinary content, differences across disciplinary standards, and differences across multilingual learners' English language proficiency levels. Indeed, such a grand unified framework for English language proficiency standards may not be possible due to the differences in disciplinary discourses necessary for disciplinary meaning making. Thus, discipline-specific ELP standards may be a more appropriate and feasible goal. The exploratory set of NGSI practices that we propose here serve as a possible contribution to address this challenge of aligning English language proficiency standards for multilingual learners with the goals of the NGSS. As a next step, our research team plans to revise our professional learning framework to prepare a new group of teachers in how to apply the NGSI practices. Using a modified teacher log and interviews, we will study how these teachers implement and adapt the new practices as we continue to revise our model. In summary, a new model of Next Generation Sheltered Instruction for science will require the integration of three-dimensional science learning with explicit attention to how language works to make specific kinds of meaning in science. Unlike first generation sheltered instruction, next generation sheltered instruction must be discipline-specific. Further, unlike some proponents of the NGSS, we do not assume that simply providing multilingual learners (or any students) access to robust science phenomena will implicitly provide the necessary environment for language development nor the skills for shifting language registers to align with students’ varying communicative topics, purposes and audiences. Through our work with teachers on the [XX] Project, we have nominated a set of specific practices that we believe have the potential for supporting the meaningful integration of science and language for multilingual learners and all students. We hope that others who share this interest will work to test and refine these and other potential next generation sheltered instruction practices so that all students can fully participate in the excitement that comes from engaging in scientific exploration.  References Authors. (2013). Authors. (2015). Authors. (2016a). Authors. (2016b). Authors. (2018). Ball, D. L., & Cohen, D. K. (1999). Developing practice, developing practitioners: Toward a practice-based theory of professional education. Teaching as the Learning Profession: Handbook of Policy and Practice, 1, 3-22. Borko, H. (2004). Professional development and teacher learning: Mapping the terrain. Educational Researcher, 33(8), 3-15. Bunch, G. C., Walqui, A., & Pearson, D. P. (2014). Complex text and new common standards in the United States: Pedagogical implications for English learners. TESOL Quarterly, 48(3), 533-559. Bunch, G. C. (2013). Pedagogical language knowledge: Preparing mainstream teachers of English learners in the New Standards Era. Review of Research in Education, 37, 298–341. Camburn, E., & Barnes, C. A. (2004). Assessing the validity of a language arts instruction log through triangulation. The Elementary School Journal, 105(1), 49–73. Carnegie Corporation of New York. (2017). Instructional materials and implementation of NGSS: Demand, supply, and strategic opportunities. New York, NY: Author. Chamot, A. U., & O'Malley, J. M. (1994). The CALLA handbook: Implementing the cognitive academic language learning approach. Massachusetts: Addison-Wesley. Cheung, A. C., & Slavin, R. E. (2012). Effective reading programs for Spanish-dominant English language learners (ELLs) in the elementary grades: A synthesis of research. Review of Educational Research, 82(4), 351-395. Cook, V. (2013). Second language learning and language teaching. New York: Routledge. Correnti, R., & Rowan, B. (2007). Opening up the black box: Literacy instruction in schools participating in three comprehensive school reform programs. American Educational Research Journal, 44(2), 298–339. Coxhead, A. (2000). A new academic word list. TESOL quarterly, 34(2), 213-238. Crawford, J. & Reyes, S. (2015). The trouble with SIOP. Portland, OR: Institute for Language & Education Policy. Cummins, J. (2008). BICS and CALP: Empirical and theoretical status of the distinction. In Encyclopedia of Language and Education (pp. 487-499). New York: Springer. Cummins, J. (2000). Putting language proficiency in its place: Responding to critiques of the conversational/academic language distinction. English in Europe: The acquisition of a third language, 54-83. Cummins, J. (1979). Linguistic interdependence and the educational development of bilingual children. Review of Educational Research, 49(2), 222-251. Desimone, L. M. (2011). A primer on effective professional development. Phi Delta Kappan, 92(6), 68-71. Duff, P. A. (2010). Language socialization into academic discourse communities. Annual Review of Applied Linguistics, 30, 169-192. Echevarria, J., Vogt, M., & Short, D. (2016). Making content comprehensible for English learners: The SIOP model. Pearson. Elo, S., Kääriäinen, M., Kanste, O., Pölkki, T., Utriainen, K., & Kyngäs, H. (2014). Qualitative content analysis: A focus on trustworthiness. Sage Open, 4(1), 2158244014522633. Fang, Z. & Schleppegrell, M. J. (2008). Reading in secondary content areas: A language-based pedagogy. University of Michigan Press. Fereday, J., & Muir-Cochrane, E. (2006). Demonstrating rigor using thematic analysis: A hybrid approach of inductive and deductive coding and theme development. International Journal of Qualitative Methods, 5(1), 80-92. Ferrini-Mundy, J. (2013). Driven by diversity. Science, 340(6130), 278-278. Flores, N., & Rosa, J. (2015). Undoing appropriateness: Raciolinguistic ideologies and language diversity in education. Harvard Educational Review, 85(2), 149-171. García, O., & Wei, L. (2014). Translanguaging and education. In Translanguaging: Language, bilingualism and education (pp. 63-77). London: Palgrave Macmillan. Gibbons, P. & Hammond, J. (2005). Putting scaffolding to work: The contribution of scaffolding in articulating ESL education. Prospect: an Australian Journal of TESOL, 20(1), 6. Gilbert, J. K. (2004). Models and modelling: Routes to more authentic science education. International Journal of Science and Mathematics Education, 2(2), 115-130. Goodman, K. S. (2014). Didn’t Found Whole Language. In Making Sense of Learners Making Sense of Written Language. (pp. 16-29). Routledge. Halliday, M. A. K., & Hasan, R. (2006). Retrospective on SFL and literacy. In R. Whittaker, A. McCabe, & M. O'donnell (Eds.). Language and literacy: Functional approaches. New York: Bloomsbury Publishing, p. 15-44. Harman, R. M., & Khote, N. (2018). Critical SFL praxis with bilingual youth: Disciplinary instruction in a third space. Critical Inquiry in Language Studies, 15(1), 63-83. Johnson, C. C., Bolshakova, V. L. J., & Waldron, T. (2016). When good intentions and reality meet: Large-scale reform of science teaching in urban schools with predominantly Latino ELL students. Urban Education, 51(5), 476-513. Kareva, V., & Echevarria, J. (2013). Using the SIOP model for effective content teaching with second and foreign language learners. Journal of Education and Training Studies, 1(2), 239-248. Klahr, D. (2002). Exploring science: The cognition and development of discovery processes. MIT press. Krashen, S. D. (1985). The input hypothesis: Issues and implications. New York: Longman. Krashen, S. D. (2013) Does SIOP research support SIOP claims? International Journal of Foreign Language Teaching, 8(1), 11-24. Krashen, S. D., & Mason, B. (2015). Can second language acquirers reach high levels of proficiency through self-selected reading? An attempt to confirm Nation's 2014 results. International Journal of Foreign Language Teaching, 10(2), 10-19. Kuhn, D. (2005). Education for thinking. Harvard University Press. Lee, O. (2018). English language proficiency standards aligned with content standards. Educational Researcher, 47(5), 317-327. Lee, O. (2017). Common Core State Standards for ELA/Literacy and Next Generation Science Standards: Convergences and discrepancies using argument as an example. Educational Researcher, 46(2), 90-102. Lee, O., Quinn, H., & Valdes, G. (2013). Science and language for English language learners in relation to next generation science standards and with implications for common core state standards for English language arts and mathematics. Educational Researcher, 42(4), 223–233. Lemke, J. L. (2001). Articulating communities: Sociocultural perspectives on science education. Journal of Research in Science Teaching, 38, 296-316. Lucas, T., & Villegas, A. M. (2013). Preparing linguistically responsive teachers: Laying the foundation in preservice teacher education. Theory Into Practice, 52(2), 98-109. Lyon, E. G., Tolbert, S., Stoddart, P., Solis, J., & Bunch, G. C. (2016). Secondary science teaching for English learners: Developing supportive and responsive learning contexts for sense-making and language development. New York, NY: Rowman & Littlefield.  MacDonald, R., Miller, E., & Lord, S. (2017). Doing and talking science: Engaging ELs in the discourse of the science and engineering practices. In A. Oliveira & M. Weinburgh (Eds.), Science teacher preparation in content-based second language acquisition (pp. 179-197). Dordrecht, the Netherlands: Springer. National Academies of Sciences, Engineering, and Medicine. (2018). English Learners in STEM Subjects: Transforming Classrooms, Schools, and Lives. Washington, DC: The National Academies Press. National Research Council. (2012). A framework for K-12 science education: Practices, crosscutting concepts, and core ideas. National Academies Press. NGSS Lead States. (2013). Next generation science standards: For states, by states. Washington, DC: The National Academies.
 No Child Left Behind Act of 2001, P.L. 107-110, 20 U.S.C. § 6319 (2002). Schauble, L. (1996). The development of scientific reasoning in knowledge-rich contexts. Developmental Psychology, 32(1), 102-119. Schleppegrell, M. J., & O'Hallaron, C. L. (2011). Teaching academic language in L2 secondary settings. Annual Review of Applied Linguistics, 31, 3-18. Schwarz, C. V., Reiser, B. J., Davis, E. A., Kenyon, L., Achér, A., Fortus, D. & Krajcik, J. (2009). Developing a learning progression for scientific modeling: Making scientific modeling accessible and meaningful for learners. Journal of Research in Science Teaching, 46(6), 632-654. Short, D., Vogt, M. E., & Echevarria, J. (2010). The SIOP Model for teaching science to English learners. Boston: Allyn & Bacon. Snow, C. E., Lawrence, J., & White, C. (2009). Generating knowledge of academic language among urban middle school students. Journal of Research on Educational Effectiveness, 2(4), 325-344. Stage, E. K., Asturias, H., Cheuk, T., Daro, P. A., & Hampton, S. B. (2013). Opportunities and challenges in next generation standards. Science, 340(6130), 276-277. Tong, F., Irby, B. J., Lara-Alecio, R., Guerrero, C., Tang, S., & Sutton-Jones, K. L. (2019). The impact of professional learning on in-service teachers’ pedagogical delivery of literacy-infused science with middle school English learners: a randomized controlled trial study in the US. Educational Studies, 45(5), 533-553. Toth, E., Klahr, D., & Chen, Z. (2000). Bridging research and practice: A cognitively based classroom intervention for teaching experimentation skills to elementary school children. Cognition and Instruction, 18(4), 423-459. Turkan, S., De Oliveira, L. C., Lee, O., & Phelps, G. (2014). Proposing a knowledge base for teaching academic content to English language learners: Disciplinary linguistic knowledge. Teachers College Record, 116(3). Walldén, R. (2019). Scaffolding or side-tracking? The role of knowledge about language in content instruction. Linguistics and Education, 54. Acknowledgement: Funding for this research was provided by the National Science Foundation under award number [XXX].   Table 1. Six Initial Instructional Practices in the [XX] Project Pedagogical Model Initial [XX] Instructional Practice Basis in the Research Description Coordinate hypothesis, observation & evidence Schauble, 1996; Kuhn, 2005 Learn how to select observations that serve as evidence to test hypotheses, how to discount other observations as not pertinent, and how to use specific language to express these relationships. Learn about controlling variables to design a fair test Klahr, 2000; Kuhn, 2005 Learn how to manipulate and control variables during investigations, to distinguish between dependent and independent variables, and to use the language of variables to evaluate a fair test. Explain cause and effect relationships Toth, Klahr, & Chen, 2000; Kuhn, 2005 Learn how to distinguish co-occurrence from causality when relating actions, events, and conditions to specific consequences, and to communicate those relationships using the language of cause and effect. Use models to construct scientific explanations and test engineering designs Schwarz et al., 2009; Gilbert, 2004 Learn how to use and construct different types of models (physical, conceptual, mathematical) to visualize, understand, and represent a phenomenon under investigation. Use general academic vocabulary in context Coxhead, 2000; Snow et al., 2009 Integrate the usage of vocabulary (e.g., indicate, feature, benefit) that is common in written academic texts, assessments, standards and teacher talk, but rare in students’ oral, conversational language. Own the language of science Fang & Schleppegrell, 2008; Lemke, 2001 Learn how to explicitly deconstruct and reconstruct language that supports the accurate and concise communication of scientific thinking  Table 2. Creating Summary Scores for the Six Initial Instructional Practices in the [XX] Project Pedagogical Model Initial [XX] Instructional Practice (Variable Name) Number of Exemplary Strategies (ES) Scoring for the Instructional Practice Coordinate hypothesis, observation & evidence (COORD) 3 ESs If ESs 1, 2, and 3 = 1, then COORD = 3; if any two of these ESs =1, then COORD = 2; if only one ES = 1, then COORD = 1; otherwise =0. Learn about controlling variables to design a fair test (CONTV) 2 ESs If ESs 1 and 2 = 1; then CONTV= 2; if any one of these ESs =1, then CONTV= 1; otherwise = 0. Explain cause and effect relationships (CAUEFF) 2 ESs If ESs 1 and 2 = 1; then CAUEFF = 2; if any one of these ESs = 1, then CAUEFF = 1; otherwise = 0. Use models to construct scientific explanations and test engineering designs (MODEL) 2 ESs If ESs 1 and 2 = 1; then MODEL = 2; if any one of these ESs = 1, then MODEL = 1; otherwise = 0. Use general academic vocabulary in context (ACADVOC) 2 ESs If ESs 1 and 2 = 1; then ACADVOC = 1; if any one of these ESs = 1, then ACADVOC = 1; otherwise = 0. Own the language of science (OWNSC) 3 ESs If ESs 1, 2, and 3 = 1; then OWNSC = 3; if any two of these ESs = 1 then OWNSC = 2; if only one ES = 1 then OWNSC = 1; otherwise = 0.  Figure 1. Six [XX] Project Initial Instructional Practices and Exemplary Strategies within Each from [XX] Project Teacher Log, Part A Part A: Thinking about the previous week, did any of the students in your science classes have opportunities to… Coordinate hypotheses, observations and evidence by… □ Yes □ No Stating their expectations based on prior experiences and knowledge (hypotheses)? □ Yes □ No Using their senses and tools to make targeted observations and collect data? □ Yes □ No Selecting appropriate observations to serve as evidence to evaluate their hypotheses? Learn about controlling variables to design a fair test by… □ Yes □ No Identifying variables in science (anything that can change during an observation or experiment)? □ Yes □ No Distinguishing between independent variables, dependent variables and controlled variables (constants) when designing an investigation or discussing the investigations of others? Explain cause and effect relationships by… □ Yes □ No Identifying cause and effect relationships (where one event, the cause, brings about another event, the effect, through some mechanism or process)? □ Yes □ No Describing key mechanisms or processes that relate to the cause and effect relationship? Use models to construct scientific explanations and test engineering designs by… □ Yes □ No Using one or more types of models (e.g., physical, drawn, simulation, mathematical) to explain scientific concepts? □ Yes □ No Using one or more types of models (e.g., physical, drawn, simulation, mathematical) to test and improve designs? Develop academic vocabulary in context by… □ Yes □ No Using general academic vocabulary (non-science-specific vocabulary common across academic content areas) orally to support meaningful explanation of science concepts or practices? □ Yes □ No Using general academic vocabulary (non-science-specific vocabulary common across academic content areas) in writing to support meaningful explanation of science concepts or practices? Own the language of science by… □ Yes □ No Translating scientific language into everyday language or vice versa? □ Yes □ No Breaking down the technical nature of scientific vocabulary through roots, prefixes and suffixes? □ Yes □ No Translating dense, abstract, or depersonalized text into more active personalized text, or vice versa?  Figure 2. Percentage of Teachers Enacting the Six Initial [XX] Project Instructional Practices, by Implementation Level and Project Year NOTE: The log listed three exemplary strategies for two of the six initial instructional practices: 1) Coordinate Hypotheses, Observations and Evidence, and 2) Own the Language of Science. The log listed only two listed two exemplary strategies for the remaining four initial project instructional practices: 1) Learn About Controlling Variables; 2) Explain Cause & Effect Relationships; 3) Use Models; and 4) Develop Academic Vocabulary in Context. For each teacher, we totaled the number of strategies used in a given week within each instructional practice and averaged the weekly totals across the number of logs that teacher completed during the year. We then categorized teachers based on these averages into "high" and "low" implementers for instructional practices with only two exemplary strategies, and "high," "medium," and "low" implementers for instructional practices with three exemplary strategies. The figure shows the percentage of teachers at each implementation level for each instructional practice. For instructional practices with three exemplary strategies, low =<1; middle = 1-2; high = >2. For instructional practices with two exemplary strategies, low = <1, high = 1-2.   Figure 3. Representative practices of first generation and next generation sheltered instruction in the secondary science classroom