“But Who Eats the Mosquitos?”: Deaf Learners’ Language Use and Translanguaging During STEAM Discussions

Abstract

Science, technology, engineering, arts, and mathematics (STEAM) education represents an array of fields that have significant promise for the future careers of students. However, in deaf education, little research has been conducted to understand how best to provide access to STEAM learning opportunities for deaf students. This manuscript explores STEAM learning and Deaf Education through the lens of translanguaging. Translanguaging is the use of multiple linguistic resources by multilingual individuals. The authors recorded deaf teens attending a STEAM camp as they engaged in a collaborative problem-solving activity to explore the language resources they used to make sense of and communicate their understanding of the problem during various stages of the activity (gathering information, generating ideas, and evaluating ideas). We viewed their interactions through a translanguaging lens. We found that the campers used an array of both language-based (ASL, spoken English, gesture, and fingerspelling) and tool-based (writing on a whiteboard, engaging with informational papers, using computers or phones) translanguaging activities to gather information and communicate with one another. While selection of language resources did not differ by activity stage, they did differ by group, suggesting that campers were sensitive to the communication needs of their group mates.

Science, technology, engineering, arts, and mathematics (STEAM) as an area of focus in educational contexts emerged in the first decade of the 21st century in response to social, political, and financial pressures giving more value to STEM fields while diminishing the value given to arts and humanities (Bequette & Bequette, 2015). In light of these various pressures, conceptualizations and visions of what STEAM entails are broad and lack strong theorizing, with great variation in the positional privileging of Arts and the component STEM disciplines in relation to each other (Mejias et al., 2021). However, scholars have recognized potential connections among the epistemological work of these disciplines that can provide helpful visions to understand potential ways they can be synthesized in research and practice (Bevan et al., 2019). Importantly, STEAM as an interdisciplinary practice incorporates creative thinking and problem solving into the work of science, technology, engineering, and mathematics (Belbase et al., 2021).

Mejias et al. (2021) address this synthesis challenge by creating a four quadrant typology that can be used to locate particular visions and applications of STEAM across practice, policy and research. This typology entails two primary axes for consideration, developed by reviewing and synthesizing rhetorical positions described or represented in the relevant literature related to STEAM (see Figure 1).

One axis employs instrumentalism for theoretical grounding exploring how the disciplines are operationalized with respect to each other. The spectrum reaches from visions of STEAM where aspects of one or several disciplines are operationalized in the service of a primary discipline (one-sided) to a transdisciplinary vision where all disciplines are mutually intertwined. The other axis focuses on pedagogical considerations and manifestations in various STEAM visions that range from those wholly focused on learning to those that do not address learning at all.

In the current project, we examined the ways that a group of deaf learners in a STEM camp navigated the linguistic and cognitive demands of STEAM-oriented activities. Applying this STEAM typology to the study described in this article, we acknowledge that the sensemaking work by deaf learners that served as the focal context resides on the one-sided end of the instrumental axis, as science is the primary discipline where the work occurs, leveraging elements of representational design from engineering and art. On the other axis, we locate this study much closer to the pedagogical end of the spectrum as the focus is on a sensemaking activity, while occurring in an informal learning context, that is grounded in pedagogical models and practices implemented and emphasized in science classrooms.

Before we continue with this study, it is worthwhile to recognize the different usage of deaf/Deaf. Using a lowercase “d” in the word “deaf” includes individuals with hearing loss ranging from mild to profound hearing levels with presence or absence of assistive technology. These individuals also have a range of communication preferences and abilities in signed and/or spoken languages. Sometimes, there will be capitalized “D” emphasizing a Deaf person as a member of a culturally and linguistically minoritized group.

1. Focus of the Study

Science and STEM as areas of focus in research receive significant attention in the educational lives of hearing learners (e.g., Bowman, 2012; Ramsey, 2022). However, deaf individuals have historically not received the same type of access to STEM educational opportunities (Nagle et al., 2016; NCES, 2023; Solomon et al., 2012; University of Delaware, 2024). As a result, the research base on teaching STEM content and concepts to deaf learners is not as robust as it could or should be. Despite this, there is some research on STEM education with deaf learners. For instance, there is evidence that deaf students learn science content more effectively through direct instruction with a signing teacher than when this knowledge is mediated by even a very skilled interpreter (Kurz et al., 2015). Prior research has also uncovered that deaf students from non-white backgrounds prefer to have mentors in STEM that share both their racial background as well as their deafness (Renken et al., 2021). Finally, there is evidence that language and reading level of deaf learners may relate to their STEM learning outcomes (Vosganoff et al., 2011). However, to our knowledge there is no research that focuses on how deaf learners engage with their peers during active learning opportunities in STEM.

It is worth considering the potential role that interactive learning plays in STEM education, especially considering the recent theoretical research put forth regarding translanguaging and deaf students (e.g., Wolbers et al., 2023). According to García and Wei (2015), the concept of translanguaging is the flexible use of multiple linguistic resources—for multilingual, hearing learners, this may be simultaneously drawing upon their knowledge in English and Spanish, as an example, as they make sense of new content. There is prior research that suggests hearing multilingual students may engage in translanguaging practices in science learning (e.g., Hou et al., 2024; Karlsson et al., 2017). In the only research exploring the use of translanguaging by deaf learners while engaging with disciplinary content, such as science or mathematics, the authors argue for the use of signed language, written text, gesture, images, and other forms of communication for both teaching and learning (Scott & Cohen, 2023). Though this recommendation was based on prior research with both deaf and hearing learners, this particular approach has not been examined in terms of how it might play out in a STEM learning context. Our goal in this study was to gather data regarding how deaf learners employ their linguistic resources via translanguaging when engaged with peers in interactive activities in an informal STEM learning environment.

2. Literature Review

2.1. STEM Sensemaking and Discourse

Current visions for science learning recognize a collection of activities comprising an overall view of the epistemological work of these disciplines (NRC, 2012). These activities can be conceptualized into three broad spheres of activity, which include the investigative, explanatory, and evaluative spheres (Osborne, 2014). The investigative sphere, focused on the exploration of phenomena and collection of empirical information related to questions about them, is perhaps the most commonly identified set of activities that distinguish science and related STEM disciplines (Schwartz et al., 2023). The explanatory sphere entails the development of causal, mechanistic accounts for phenomena, which can include the development of models, theories, and laws (Osborne, 2014). Deliberation and scientific argumentation exist as the central sphere of scientific activity as a sensemaking and refinement endeavor that connects the investigative and explanatory spheres (Schwartz et al., 2023). Artifacts created during investigative and explanatory activities, including methodological designs or functional, predictive models, emerge from intertwined generative and critiquing processes that constitute the overall evaluative sphere (Osborne, 2014; González-Howard & McNeill, 2020; Waldrip & Prain, 2017). The processes of the evaluative sphere are fundamentally grounded in discourse (Schwartz et al., 2023; González-Howard & McNeill, 2020; Grapin et al., 2023). In the current manuscript, we explore the role of these types of sensemaking activities as they play out in the interactions between deaf students as they discuss science content with their peers.

The role of discourse in supporting science sensemaking, in both formal and out-of-school contexts, continues to emerge as a critical area of inquiry (Schwartz et al., 2023; Kelly et al., 2023). We conceptualize science sensemaking discourse as embodying an array of linguistically rich written, drawn, and verbal products, as well as the use of myriad linguistic resources in social interactions, all aimed at constructing knowledge in response to pressing epistemological needs (Kelly et al., 2023). We also take an expansive view of productive science discourse to include and value linguistic resources students bring with them to these learning experiences from their lives outside, moving beyond more rigid academic conceptualizations that can act in oppressive ways for learners (Scott & Cohen, 2023; Licona & Kelly, 2020; Parkinson & Crouch, 2011; Malone, 2023). Scholars have noted how understanding the multiple linguistic resources students use to engage in science and STEAM learning can lead to a multitude of benefits (Grapin et al., 2023; Licona & Kelly, 2020). Sensemaking discourse occurring among groups of students and between students and teachers when engaged in generative and evaluative work provides learners with opportunities to take up more epistemic agency, enhancing their ability to author the production of scientific knowledge individually and collaboratively (Schwartz et al., 2023; Meyer & Crawford, 2015; Stroupe, 2014). Beyond increasing their conceptual and epistemic understandings of science, positioning learners and their linguistic resources during sensemaking discourse as assets supports the enhancement of their affinity and affect toward science (Dini et al., 2021; Kelly et al., 2023; Starr et al., 2020).

Although a myriad of benefits are attributed to fostering and sustaining productive sensemaking discourse in science learning, several challenges emerge that warrant attention. A multitude of studies demonstrate that a significant amount of time is dedicated to generative work, including the production of investigations, evidence, models, arguments, and explanations (Christodoulou & Osborne, 2014; Duncan et al., 2018; Manz & Renga, 2017; Osborne et al., 2016; Osborne, 2014; Schwartz et al., 2023). However, similar attention and time are not provided for students to engage in the evaluative work of science, where learners engage in constructive critique of each others’ work (Dini et al., 2021; Osborne et al., 2016; Starr et al., 2020). By limiting discourse to primary considerations of generating knowledge, learners are not afforded opportunities to understand the full epistemic function of the sensemaking practices they are engaged in (Osborne, 2014; González-Howard & McNeill, 2020; Waldrip & Prain, 2017).

Teachers’ framing and use of varied talk moves comprise another area for critical consideration as they facilitate science sensemaking with their discursive contributions framing the overall goals of the work as well as supporting attitudinal and identity-related connections to the learning occurring (Benedict-Chambers et al., 2017; González-Howard & McNeill, 2020; Sandoval et al., 2019). However, research involving multilingual learners demonstrates that teachers may focus more on structural aspects of scientific knowledge products generated by students rather than the conceptual and epistemic quality while their language instruction focused on appropriate use of academic vocabulary (González-Howard & McNeill, 2019). To address these challenges, several scholars call for expanding views of what counts as acceptable scientific discourse, emphasizing translanguaging as a valuable ideological and practical perspective for supporting learning across a broader range of learners (Grapin et al., 2023; Jakobsson et al., 2021; Kelly et al., 2023).

2.2. Science Education, STEAM, and Translanguaging

Translanguaging as a framework was originally conceptualized based on the language use of multilingual individuals (García, 2009; Schwartz et al., 2023; García & Wei, 2014). Overall, the theory of translanguaging rejects rigid language boundaries, arguing instead that multilingual individuals use all the linguistic resources available to them, not separately as in code switching between one language and another, but simultaneously (García et al., 2021). The practical application of this theory is that when teaching multilingual learners, we should also provide opportunities for students to receive and express information using all of their language resources.

Translanguaging in science and STEAM learning contexts allows multilingual students to use all of their available semiotic resources (e.g., first languages, diagrams, gestures, simulations, etc.) during learning experiences to generate and evaluate knowledge products and consensus understandings reached through scientific discourse. For multilingual students, using translanguaging supports the development of conceptual understanding and enhances their proficiency with science and engineering practices while also broadening their perceptions of how their culturally grounded identities connect to STEM identities (González-Howard & McNeill, 2019; González-Howard et al., 2023; Malone, 2023; Pierson et al., 2021). Pierson et al. (2021) noted how multilingual learners in a science classroom engaged in modeling during an ecology unit used not only formal language resources to communicate their ideas but also other available semiotic resources, particularly embodied actions and gestures. The researchers argued that translanguaging can be explicitly implemented as a collection of meaning-making practices analogous to the practice of scientific modeling. Similarly, Suárez and Otero (2024) explored how multilingual learners in an elementary classroom create scientific models explaining the production of sound by a guitar using a variety of semiotic resources, with some becoming shared among groups of learners while others remained distinct. Although learners varied in their use of these resources, evidence emerged that all learners made progress toward a consensus understanding of the phenomenon, seen as one of the primary goals for engaging in sensemaking practices and discourse (McNeill & Berland, 2017).

When considering teachers’ leveraging the full range of learners’ linguistic resources, translanguaging research in science learning contexts makes clear that teachers must be intentional and explicit in their guidance to learners. Suárez (2020) notes that teachers’ modeling of translanguaging practices implicitly signal to students how and when to use those same practices. However, other studies document how when teachers intentionally positioned the use of named languages other than English, they made translanguaging accessible to both multilingual and monolingual learners (Grapin et al., 2023). Licona and Kelly (2020) highlighted how a teacher, Ms. Ramirez, intentionally used both English and Spanish while engaging multilingual learners in the generation of scientific arguments. While doing so, the teacher also explicitly described the epistemic and discursive function of various components of argumentation, linking the epistemic work with multiple linguistic resources for sensemaking. Further, work in out-of-school contexts showed how honoring learners’ translanguaging practices expanded their understanding of STEM and familiarized them with cultural practices prevalent in formal school contexts (Grapin et al., 2023). Taken together, these studies and others highlight the fruitfulness of employing translanguaging in science and STEAM learning spaces, particularly when working with multilingual learners (González-Howard & McNeill, 2019; González-Howard et al., 2023).

While translanguaging has gained traction in bilingual education, there have also been studies that observed deaf learners communicating in STEM learning spaces. For example, Classroom of the Sea (COS) was a partnership between Rochester Institute of Technology (RIT), University of Connecticut, and the American School for the Deaf (ASD) to bring deaf students on a research vessel (Zheng & Goldin-Meadow, 2002). Professors from the two universities worked with the science teachers at the school to develop a curriculum investigating how the acoustics produced by human activities affect marine wildlife in the water. The project used problem-based learning to conduct real-world investigations of marine science. ASD teachers worked with the RIT professor to prepare the deaf students to communicate concepts and work with the equipment aboard the research vessel. The study found that having deaf students in an authentic scientific research environment afforded them the experience of doing STEAM work while reciprocally benefiting the researchers’ perspectives.

While translanguaging was not the focus of this project, participants in COS used forms of translanguaging to communicate different concepts. The COS team investigated a signed lexicon for their lessons on marine science and identified the shortfalls in developing communication tools to prepare students for the trip. Deaf students used sign language interpreters to converse with the bioacoustic professor, while the ASD teacher and NTID professor could communicate using a signed language with the students directly. They also used models, figures, images, and other visual and language representations to discuss the concept of acoustic and how it moves underwater to prepare the students to work with the bioacoustic professor on the research vessel.

Molander et al. (2010) investigated ways of reasoning produced by deaf and hearing students about ecological phenomena and found that uncertainty between signed language and written language caused unproductive reasoning. This occurred especially among terms that had colloquial meanings distinct from the scientific concepts, such as producer, consumer, and nutrients. Learning science is not only content learning but also language learning to describe, explain, and expand the conceptual understanding of core ideas in the service of explaining real world phenomena (Molander et al., 2010). Molander et al. (2010) recommended an expansive bilingual approach to support deaf students in developing a deeper conceptual understanding, thus, translanguaging could play a key role in helping students navigate multiple language resources to learn the content.

2.3. Deaf Education and Translanguaging

Several researchers have explored the translanguaging framework within the deaf education context (e.g., De Meulder et al., 2019; Hoffman et al., 2017; Swanwick, 2017). It is possible that the first scholarship explicitly connecting translanguaging and the language practices of deaf individuals was Hoffman’s (2014) dissertation, which explored the reading of deaf students via a translanguaging lens. Shortly afterward, researchers continued to explore the possible utility of applying a translanguaging lens to deaf people’s language use and communication, some arguing that translanguaging is an appropriate framework for exploring the use of signed, spoken, and written languages among deaf individuals (Hoffman, 2014). Others, however, rightly advocate for caution that “sensory asymmetries” between hearing and deaf individuals and the way these asymmetries are often perceived by those with political and social power (De Meulder et al., 2019). De Meulder et al. (2019) recommend the use of translanguaging as a framework for describing the language uses of deaf individuals, but recommend against using it to prescribe classroom or language practices.

In 2023, Wolbers and colleagues proposed the Translanguaging Framework for Deaf Education (TFDE). This framework includes both receptive and expressive language, and suggests multiple means of engaging in translanguaging in the deaf education classroom, including validation of individual ideolects, building a shared understanding, building metalinguistic knowledge, and communicating with an external audience, all in the social communication context (Wolbers et al., 2023). Scott and Cohen (2023) further argued that within disciplinary contexts, translanguaging among deaf children should include not only signed, spoken, and written languages, but also particular ways of communicating within that discipline, such as images, graphs, or other ways that information may be received or conveyed. In this study, we adopt this expansive view of translanguaging using both Wolbers et al. (2023) and Scott and Cohen’s (2023) frameworks.

We sought to answer the following research questions:

• How frequently do deaf learners in a STEM camp context engage in conversations that are informational, generative, and evaluative when completing a food web building activity grounded in several science and engineering practices?
• How frequently do these same learners employ translanguaging strategies to communicate with one another and engage in scientific sensemaking? Do these strategies differ by type of conversation (informational, generative, evaluative)?

3. Methods

3.1. Participants

The participants in this project were all adolescents who were attending a camp that was designed for deaf teens to engage with STEM1. Although the camp itself explored multiple disciplines within this framework, the particular activity we examine here was specifically within the domain of science. All campers were from Georgia and were between the ages of 16–20. For this camp, teens using any language/modality were eligible to attend, but the vast majority of attendees used American Sign Language (ASL). Nine campers knew ASL, and one camper knew spoken English only. Many campers were seen to communicate with one another using voice as well as signs. Campers came from a range of backgrounds, including a deaf residential school, a day school for the deaf, a self-contained deaf education classroom in a public school, and a primarily mainstream setting. The campers were split into groups to complete this activity by the camp director (author 3) based on observations of their comfort with engaging in science practices and their language use. There were ten campers in these groups; of them, 7 (70%) were boys, and 3 (30%) were girls. All of the participants were Black, Indigenous, or People of Color (BIPOC). Although our analyses in this project do not examine racial or gender dynamics, we include this basic demographic information for contextual purposes. Below, individuals within each group are described, along with the languages/dynamics within the group.

Group 1: Group one was made up of all boys (n = 3). One camper was Black (Michael2), one was Asian (Tuan), and one was Hispanic (Hector). These campers all used ASL as their primary language, and communicated with one another using signed language, without voice/spoken English. Michael (age 16) and Hector (age 18) attended the day school for the deaf, and Tuan (age 17) attended school in a mainstream classroom where a signed language interpreting service was provided.

Group 2: Group two was made up of one boy and two girls (n = 3). The two girls were Black (Kayla, age 17 and Morgan, age 16), and the boy was Asian (Andrew, age 17). Two of the group members (Andrew and Kayla) used ASL exclusively to communicate. Kayla attended a day school for the deaf, and Andrew was enrolled in mainstream classrooms in his county’s deaf education program. The third group member, Morgan, came from a spoken English background and did not know any signs. She attended general education using assistive technology to participate in the classroom. For this reason, this group seemed to make more use of gestures and writing for communication during group activities.

Group 3: Group three was made up of three boys and one girl (n = 4). The three boys were Black (Andre [age 16], Chris [age 18], and Isaiah [age 20]), and the girl was Asian (Ana, age 16). While all group members were able to sign, the three boys also used spoken English, and Ana was not observed to use spoken English. All of the group members attended a school for the deaf that used sign language as their primary instructional language, either a day school (Ana, Andre, and Chris) or a residential school (Isaiah). Because of this dynamic, this group used both signed and spoken language at times in the course of their activities.

3.2. Setting

The data for this study were collected during a summer camp for deaf adolescents that has the explicit goal of exposing these teens to STEM concepts and activities. The camp itself was a co-venture between Georgia State University and the Georgia Center for the Deaf and Hard of Hearing, and was designed and implemented by a Deaf former science teacher and doctoral candidate in science education at the time of the camp. He is also the third author on this publication. He was assisted by primarily deaf camp staff. During the camp, the campers engage in problem solving activities, science experiments, and go off-campus to significant sites of STEM knowledge generation and dissemination in the state of Georgia. The emphasis on interactive learning opportunities made the data from this camp ideally suited for analysis through the TFDE lens (Wolbers et al., 2023). This study was approved by the Georgia State University Institutional Review Board. Parents/guardians of all participants provided permission for their child to participate in the study, and participants themselves provided consent or assent (depending on age) prior to data collection.

The data used in this study came from one particular activity adapted from the Argument-Driven Inquiry (ADI) food web activity (Grooms et al., 2015). The activity, designed using an instructional model that emphasizes providing learners with epistemic agency (Stroupe, 2014; Grooms et al., 2015) by engaging them in scientific argumentation to develop a response to a guiding question grounded in real-world and scientific contexts. Though this activity occurred within the context of a STEM camp, we argue here that because the campers also engaged with design elements that this activity is suited for exploration using the STEAM framework. For this activity, the campers are presented with a scenario involving a marsh ecosystem interacting with a human neighborhood. The focal issue concerns what the possible ecosystem impacts would be of a large scale spraying of mosquito repellent. This context is used to have learners compare what would happen to the ecosystem’s food web if any one particular species was removed. Learners are provided with habitat data for 16 ecosystem species, including their diet, niche dynamics, and reproductive habits. To develop evidence to argue from, groups collaborate to create a food web model using that information, a process which necessitates learners synthesizing graphic artistic abilities with scientific knowledge. The campers next create a visual representation of their web to share with their fellow campers. This creative and conceptual model is then used to evaluate the relative impacts on feeding relationships if a particular species is removed. For clarity, there is not a definitively ‘correct’ argument to be made. Rather, models, claims, and other knowledge products should be evaluated by examining the quality and relevance of any particular group’s evidence and reasoning. In making their final determinations, learners then used their analyses to consider whether a small town near the focal marsh ecosystem should conduct large-scale spraying for mosquitos.

3.3. Methodological Design

To develop a robust understanding of how deaf learners engaged in a science sensemaking activity through translanguaging, we chose to employ a mixed methods exploratory design (Creswell & Clark, 2017). Attempting to understand discourse in action (Hoffman, 2014; Kelly et al., 2023), we used video cameras placed at various angles to capture the discursive moves of the groups. Video data analysis has precedent in deaf education research (Crasborn & Sloetjes, 2010) as well as translanguaging research in science contexts (Karlsson et al., 2017). This video data served as the foundation for coding and content analysis to identify evidence of interactions of scientific sensemaking and translanguaging. The coding analysis also facilitated quantitative analysis, further exploring those interactions. Together, these findings provided a collection of evidence to develop several interpretations discussed later below.

3.4. Sources of Data

The data from this study includes two hours of video recordings of three different groups of campers as they engage with the ADI food web activity. During this activity, the campers had access to printed documents describing various animals within a food chain and what they eat, internet-connected computers displayed on the wall (each group had access to their own computer plus a screen above projecting information gathered by the activity leader), a whiteboard, and their phones. These resources allowed for the application of an expansive view of translanguaging using multiple modalities and languages (Scott & Cohen, 2023) via the TFDE framework (Wolbers et al., 2023).

Initial summaries of the videos were completed in three minute chunks by graduate research assistants, who are also the fourth and fifth authors on this manuscript. These were then expanded upon by authors 1 and 3, both of whom are fluent users of ASL to include specific detail on how and with whom the campers are interacting, and what physical and virtual resources they are using in the course of those interactions. This process resulted in a robust description of the campers, the classroom, the language and communication activities, and their use of various resources for communication and expression.

3.5. Our Positionalities

Author 1: I (Jessica Scott) am a white, hearing woman. I started learning ASL at the age of 17, and have worked in schools for the deaf and in Deaf Education teacher preparation programs for my entire adult life. Prior to my work as an associate professor of Deaf Education, I worked as a high school ELA/social studies teacher for deaf students, and a K-8 reading specialist for deaf students. My work centers around issues of language, specifically signed languages, and their role in the education of deaf students.

Author 2: I (Patrick Enderle) am a white, hearing male. I taught high school and post-secondary level biology courses, along with years of work in an array of middle and high school science classrooms supporting the implementation of instruction emphasizing scientific practices, particularly argumentation. I have extended my research focus to understand how different groups of teachers and students, including those who are deaf or hard of hearing, can be supported in such sensemaking work.

Author 3: I am Scott Cohen, a white man who was diagnosed with deafness shortly after birth. I learned American Sign Language (ASL) and spoken English in school and graduated from a program in deaf education. I initially became a science teacher specializing in deaf education and now serve as a professor in that field. My work focuses on formal and informal science teaching and learning for deaf students.

Author 4: My name is Jasmine Smith and I am a Black hearing woman. I grew up with Deaf grandparents, which shaped my passion for Deaf culture and education. After seven years of interpreting, I’m now graduating with my Master’s in Deaf Education. My focus? Supporting Black Deaf educators-because representation matters. I want to help build a future where every Deaf student has role models who look like them and understand their experiences.

Author 5: I am Reagan Hutchison, and I am a white, hearing woman. I started learning American Sign Language (ASL) at the age of 20, which led me to pursue a degree in ASL Interpreting and shaped the direction of my professional life. Today, I work as both a Teacher of the Deaf for middle and high school students and as an ASL instructor in the public high school system. Previously I’ve served as an Educational Interpreter across all grade levels and also taught undergraduate courses in a Deaf Studies college program. As a hearing professional in Deaf educational spaces, I continually reflect on my role and responsibility. I strive to approach my work with humility, cultural responsiveness, and a deep respect for Deaf voices and experiences, always learning from the communities I serve.

3.6. Analytic Approach

In this study, we used a convergent parallel mixed methods study design, in which quantitative and qualitative data were collected and analyzed simultaneously (Creswell & Clark, 2017). This approach allowed us to engage with video data via both numerical counts of types of behaviors as well as rich descriptions of specific interactions, which could then be used to compare findings across analytic approaches. Below we explore the specific ways that we engaged with quantitative and qualitative data in this study.

3.7. Quantitative Data

In order to analyze the data, the authors undertook an interval coding approach (Pesch & Lumeng, 2017), breaking the videos into three minute segments. Each segment was coded for the totality of observable instances, meaning that all activities within that three minute span were represented, but codes were not completed line-by-line of discursive contributions. We coded our data for two major purposes: To understand how the campers were engaging in the process of completing the scientific argumentation task, and to understand how they were using translanguaging strategies during the completion of the task. We focused specifically on an activity that had significant opportunity for interaction and collective meaning-making (Scott & Cohen, 2023; Osborne, 2014). We coded for discourse type (e.g., informational discussion; generative discussion; evaluative discussion) and linguistic resources (e.g., use of ASL; fingerspelled terms; spoken English; gesture; writing on whiteboards; writing on paper; typing on a computer; typing on a phone). See

Table 1. Codebook for Discourse Type and Linguistic Resource Use.

Discourse Type Code	Description
Informational	Discourse involving the process of the activity and/or sharing factual knowledge without using it to generate food web models
Generative	Discourse involving coming up with ideas related to solving the problem, including examining data, bringing ideas from that data to the group, or describing a rationale for an idea
Evaluative	Discourse involving exchange of feedback on generated ideas and products and assessing model quality
Linguistic Resource Code	Description
ASL	Communication using ASL; Grammatical/vocabulary knowledge or accuracy not assessed
Fingerspelling	Communication using fingerspelling to denote specific terms or concepts.
Spoken English	Communication using spoken English; Grammatical/vocabulary knowledge or accuracy not assessed
Gesture	Communication using some form of gesture (e.g., pointing, shrugging) for communication of ideas with others.
Computer	Communication using computer searches for images and information and sharing with peers to enhance food webs.
Whiteboard	Communication using classroom whiteboards as writing and drawing spaces to generate and evaluate food webs.
Hardcopy Paper	Communication using paper resources designed for the activity to make sense of their ideas and share them.
Phone	Communication using typed messages to one another or gathering information about the ecosystem using their phones

for a description of each code.

Authors one and three co-coded 20% of the data to develop familiarity and agreement related to the coding system. After this, each author coded an additional 20% of the data to determine interrater reliability. We achieved a Cohen’s Kappa of 0.637, which demonstrates substantial agreement between raters. Disagreements were discussed by the group and resolved to 100% agreement. The remaining data were divided between authors 1 and 3 for final coding, and each author reviewed the codes from the other to ensure that any areas for potential disagreement were resolved.

3.8. Qualitative Data

To engage with the qualitative data, we used a content analysis approach (Krippendorff, 2003; Neuendorf, 2017; Patton, 2014) employing grounded theory (Green et al., 2007). Authors 1 and 2 reviewed the videos and transcripts to identify instances of productive science sensemaking occurring influenced by and/or using multiple linguistic resources. These were identified by connecting the translanguaging codes (described above) to video sections where multiple translanguaging codes appeared simultaneously. We then expanded our description of these episodes to better understand how translanguaging manifested as the campers engaged in sensemaking activities.

4. Findings

4.1. Quantitative Findings

In this section, we present results of quantitative coding of video segments. Specifically, we report on how often students engaged in the three types of sensemaking activities (informational, generative, and evaluative activities), the frequency of translanguaging activities during each sensemaking activity, correlations between types of translanguaging activities, and finally between-group differences in translanguaging activities.

4.2. Frequency of Codes

First, we examined how frequently different types of sensemaking activities (informational, generative, and evaluative discussions) occurred during the task. Campers engaged in informational tasks 59 times across segments, in generative tasks 61 times across segments, and in evaluative tasks 37 times across segments. A segment could include more than one type of sensemaking activity. See

Table 2. Frequencies and percentages of STEAM sensemaking codes.

Science Code	Frequency	Percent	Cumulative Percent
Informational	59	37.6%	37.6%
Generative	61	38.9%	76.4%
Evaluative	37	23.6%	100%

for information on the frequency and percentages of the three sensemaking codes.

In terms of language-based translanguaging activities, ASL occurred the most frequently. In fact, out of a total of 157 coded segments, 156 included the use of ASL. Gesture occurred 69 times, fingerspelling 33 times, and spoken English 9 times. In terms of tool-based translanguaging activities, the whiteboard was used in 99 segments, hard copy paper was used in 77 segments, the computer in 69 segments, and a phone in 7 segments.

4.3. Translanguaging by the Sensemaking Activity

Next, we explored cross-tabulations between the science sensemaking activity and the presence of specific translanguaging activities. We found that ASL was used most often across informational, generative, and evaluative activities, while spoken English was used least. A whiteboard was the most common communication tool used across segments, and individual campers’ phones were used least often. See

Table 3. Percentages of segments within each sensemaking activity including the language-based and tool-based translanguaging activities.

	Informational (n = 59)	Generative (n = 61)	Evaluative (n = 37)
Language
ASL	59 (100%)	60 (98%)	37 (100%)
Gesture	24 (41%)	31 (51%)	14 (38%)
Spoken English	4 (7%)	5 (8%)	0 (0%)
Fingerspelling	12 (20%)	13 (21%)	8 (22%)
Tools
Whiteboard	34 (58%)	46 (75%)	19 (51%)
Hard copy paper	30 (51%)	32 (52%)	15 (41%)
Phone	3 (5%)	8 (13%)	0 (0%)
Computer	22 (37%)	32 (52%)	15 (41%)

for percentages of segments within each sensemaking activity that included the various language-based and tool-based translanguaging activities.

Language-based translanguaging by group by science sensemaking activity stage. We also explored the use of particular translanguaging activities within each group. First we present this data for translanguaging codes that were explicitly language-based (e.g., use of ASL, gesture, spoken English, and fingerspelling).

For group one, overall we found 56 segments that used ASL, 25 that used gesture, 14 that used fingerspelling, and none that used spoken English. For group one, segments were largely evenly divided between informational, generative, and evaluative discussions. Across these, ASL was most frequently used regardless of sensemaking activity stage, followed by fingerspelling and gesture. Spoken English was not used at any point within this group.

For group two, overall we found 57 segments that used ASL, 31 segments that used gesture, 14 segments that used fingerspelling, and 0 segments that used spoken English. Informational and generative segments appeared at the same rate (n = 19), while there were fewer evaluative segments (n = 7). Again, ASL was used most frequently regardless of sensemaking stage, and informational and generative segments also frequently used gesture and fingerspelling. Like group one, spoken English was not used at any point within this group.

For group three, overall we found 57 segments that used ASL, 31 segments that used gesture, 9 segments that used spoken English, and 6 segments that used fingerspelling. Informational and generative segments appeared at the same rate (n = 22), while there were fewer evaluative segments (n = 13). Again, we found a pattern of ASL appearing most frequently across sensemaking stages along with gesture. This group also used spoken English with relative frequency, and was the only group to do so. See

Table 4. Language-based codes by group and sensemaking activity.

Group	Science Activity	Translanguaging Coding	Frequency
Group 1	Informational (n = 18)	ASL	18 (100%)
		Gesture	9 (50%)
		Spoken English	0 (0%)
		Fingerspelling	4 (22%)
	Generative (n = 20)	ASL	20 (100%)
		Gesture	6 (30%)
		Spoken English	0 (0%)
		Fingerspelling	7 (35%)
	Evaluative (n = 17)	ASL	17 (100%)
		Gesture	9 (53%)
		Spoken English	0 (0%)
		Fingerspelling	3 (18%)
Group 2	Informational (n = 19)	ASL	19 (100%)
		Gesture	8 (42%)
		Spoken English	0 (0%)
		Fingerspelling	6 (32%)
	Generative (n = 19)	ASL	19 (100%)
		Gesture	16 (84%)
		Spoken English	0 (0%)
		Fingerspelling	3 (16%)
	Evaluative (n = 7)	ASL	7 (100%)
		Gesture	0 (0%)
		Spoken English	0 (0%)
		Fingerspelling	4 (57%)
Group 3	Informational (n = 22)	ASL	22 (100%)
		Gesture	7 (32%)
		Spoken English	4 (18%)
		Fingerspelling	2 (9%)
	Generative (n = 22)	ASL	21 (98%)
		Gesture	9 (41%)
		Spoken English	5 (23%)
		Fingerspelling	3 (14%)
	Evaluative (n = 13)	ASL	13 (100%)
		Gesture	5 (38%)
		Spoken English	0 (0%)
		Fingerspelling	1 (8%)

for all language-based codes that appeared by activity and group.

Tool-based translanguaging by group by science sensemaking activity. We also calculated within-group frequencies for the translanguaging activities that employed external tools (e.g., use of a whiteboard, paper, computer, and phone). Overall, group one had 35 segments that used a whiteboard, 34 segments that used hard copy paper, 33 segments that used a computer, and 1 segment that used a phone. For group one, across all sensemaking types, reading and writing on paper, a whiteboard, or computer were used most frequently, though at slightly different rates, and phone use was least frequent.

Overall, group two had 40 segments that used a whiteboard, 27 segments that used hard copy paper, 26 segments that used a computer, and 13 segments that used a phone. The use of a whiteboard was the most frequent tool across all sensemaking activities, while a phone was used most commonly in generative discussions and not at all in evaluative discussions.

Overall, group three had 33 segments that used a whiteboard, 20 segments that used hard copy paper, 17 segments that used a computer, and 0 segments that used a phone. Again, a whiteboard was used most frequently regardless of type of sensemaking discussion followed in all cases by use of a computer and/or written information on paper. This group did not use a phone at any point. See

Table 5. Tool-based codes by group and sensemaking activity.

Group	Science Activity	TL Coding	Frequency
Group 1	Informational (n = 18)	Whiteboard	10 (56%)
		Hard Copy Paper	12 (67%)
		Phone	0 (0%)
		Computer	10 (56%)
	Generative (n = 20)	Whiteboard	16 (80%)
		Hard Copy Paper	13 (65%)
		Phone	1 (5%)
		Computer	15 (75%)
	Evaluative (n = 17)	Whiteboard	8 (47%)
		Hard Copy Paper	8 (47%)
		Phone	0 (0%)
		Computer	8 (47%)
Group 2	Informational (n = 19)	Whiteboard	12 (63%)
		Hard Copy Paper	10 (53%)
		Phone	3 (16%)
		Computer	7 (37%)
	Generative (n = 19)	Whiteboard	16 (84%)
		Hard Copy Paper	10 (53%)
		Phone	7 (37%)
		Computer	9 (47%)
	Evaluative (n = 7)	Whiteboard	4 (57%)
		Hard Copy Paper	4 (57%)
		Phone	0 (0%)
		Computer	3 (43%)
Group 3	Informational (n = 22)	Whiteboard	12 (55%)
		Hard Copy Paper	8 (36%)
		Phone	0 (0%)
		Computer	5 (23%)
	Generative (n = 22)	Whiteboard	14 (64%)
		Hard Copy Paper	9 (41%)
		Phone	0 (0%)
		Computer	8 (36%)
	Evaluative (n = 13)	Whiteboard	7 (53%)
		Hard Copy Paper	3 (23%)
		Phone	0 (0%)
		Computer	4 (31%)

for tool-based codes that appeared by activity and group.

4.4. Correlations Between Translanguaging Events

Next, we examined the data for correlations between different codes. We wanted to determine whether certain types of translanguaging and sensemaking activities appeared to co-occur with one another during this type of STEAM learning experience. See

Table 6. Correlation Matrix.

	ASL	Gesture	Voice	Fingerspelling	Whiteboard	Hardcopy Paper	Phone	Computer
ASL	1
Gesture	0.071	1
Voice	0.020	−0.108	1
Fingerspelling	0.041	−0.068	−0.060	1
Whiteboard	−0.061	0.326 ***	−0.038	−0.221 **	1
Hardcopy Paper	0.079	0.158 *	−0.023	0.182 *	0.091	1
Phone	0.022	0.310 ***	−0.068	0.042	0.210 **	0.130	1
Computer	0.071	−0.005	0.058	0.079	−0.014	−0.022	0.310 ***	1

* p < 0.05; ** p < 0.01; *** p < 0.001.

for the correlation matrix.

In this table, we can see several correlations between various translanguaging activities. Gesture was correlated with the greatest number of other translanguaging activities—there was a correlation between gesture and use of the whiteboard, hard copy papers, and the phone. In contrast, the use of fingerspelling was negatively correlated with the use of a whiteboard, while the use of a phone was positively correlated with the use of a whiteboard. The use of a phone was also correlated with the use of a computer. Finally, the use of fingerspelling was correlated with interacting with hard copy papers. Though we tested for a correlation between translanguaging activities and the sensemaking activities (e.g., informational, generative, and evaluative), there were no significant findings to report.

4.5. Between Group Differences in Translanguaging

Our final quantitative analytic activity was to conduct independent sample t-tests between the groups with regard to their use of translanguaging activities to determine whether group dynamics played a role in what types of linguistic resources campers were drawing upon to complete the activity. We explored these between group differences because of the observed differences among campers for language use (e.g., one group that used only ASL, one group that had a non-signing member, and one group where most group members used both ASL and spoken English). We conducted a post-hoc power analysis using G*Power. The power analysis, based on a large effect size (f = 0.50), alpha level of 0.05, and power of 0.84, finds that the observations between groups one (n = 55) and three (n = 57) are sufficient for detecting between group differences. Similar results (f = 0.50), alpha level of 0.05, and power of 0.80, were achieved between groups one (n = 55) and two (n = 45) as well as (f = 0.50), alpha level of 0.05, and power of 0.80 between groups two (n = 45) and three (n = 57).

Language-based translanguaging between-group differences. We first tested the differences between groups one and two on language-based translanguaging activities, and found no significant differences in terms of the two groups’ use of ASL, spoken English, fingerspelling, or gesture. However, groups one and three differed in several ways. Group three (M = 0.16; SD = 0.37) compared with group one (M = 0.00; SD = 0.00) used spoken language more frequently t(−3.18) = 110, p < 0.001. Conversely, group one (M = 0.25; SD = 0.44) compared with group three (M = 0.10; SD = 0.31) used fingerspelling more frequently t(2.08) = 110, p < 0.001.

Groups two and three also differed significantly in a couple of areas. First, group three (M = 0.16; SD = 0.37) compared with group two (M = 0.00; SD = 0.00) used spoken English more often t(−2.88) = 100, p < 0.001. Second, group two (M = 0.29; SD = 0.46) compared with group three (M = 0.11; SD = 0.31) used fingerspelling more frequently t(2.41) = 100, p < 0.001.

Tool-based translanguaging between-group differences. Next, we tested between group differences of the three groups in their use of tools-based translanguaging strategies (e.g., whiteboard, phone, computer, hard copy papers). All three groups were found to have differences in at least one area. Firstly, group two (M = 0.22; SD = 0.42) compared with group one (M = 0.02; SD = 0.13) used a phone more frequently t(−3.40) = 98, p < 0.001. Additionally, group two (M = 0.71; SD = 0.46) compared with group one (M = 0.62; SD = 0.49) used a whiteboard more frequently t(−0.97) = 98, p = 0.05.

Group one also differed significantly from group three. Group one (M = 0.02; SD = 0.13) compared with group three (M = 0.00; SD = 0.00) used a phone more frequently t(1.02) = 110, p = 0.04. Group one (M = 0.60; SD = 0.49) compared with group three (M = 0.30; SD = 0.46) also used a computer more frequently t(3.34) = 110, p = 0.03.

Groups two and three had the most between-group differences according to t-test results. Group two (M = 0.71; SD = 0.46) compared with group three (M = 0.58; SD = 0.50) used the whiteboard during their group activities more frequently t(1.38) = 100, p = 0.008. Group two (M = 0.22; SD = 0.42) compared with group three (M = 0.00; SD = 0.00) used a phone more frequently t(4.00) = 100, p < 0.001. Group two (M = 0.42; SD = 0.50) compared with group three (M = 0.30; SD = 0.46) also used a computer more frequently t(1.30) = 100, p = 0.021. Finally, although it only approached significance, findings suggest that group two (M = 0.43; SD = 0.50) compared with group three (M = 0.35; SD = 0.48) used hard copy papers more frequently t(1.86) = 100, p = 0.056.

4.6. Qualitative Findings

In this section, we present vignettes of how various groups made use of translanguaging resources to communicate with one another and shape their sensemaking activities. The vignettes were selected via discussion and consensus among the researchers of instances that provide concrete examples of three quantitative patterns. The first highlights the use of language-based translanguaging resources such as gesture and fingerspelling (level one of the TDEF, valuing individual ideolects). The second was the use of an array of translanguaging resources that were both tool-based and language-based to develop an explanation for an idea (level two of the TDEF, coming to a shared understanding). The last demonstrates how one group expanded their understanding of the content via engaging with their group mates, the information resources, as well as the other groups in the camp (level three of the TDEF, building (STEAM) knowledge). Taken together, these vignettes demonstrate episodes where providing an array of translanguaging resources supported the unique linguistic needs of each group.

4.7. Vignette 1—Use of Gesture and Fingerspelling (Individual Ideolects)

Group two’s vignette provides an example of how the various campers within and across groups both used their own and acknowledged one another’s individual ideolects. This was especially evident with group two, as this was the group that included a member that did not know ASL. Because of this, they appeared conscientious about how they communicated with one another to ensure that all were included. Group two included Kayla, Morgan, and Andrew, though Kayla is not present in this particular interaction.

Morgan (who does not know ASL) and Andrew (who does not use spoken English) are both standing in front of the whiteboard surveying their food web as it has been designed so far. They occasionally draw or modify arrows between different food web components. Andrew points to two spots in the food web and looks at Morgan, as if to say, ‘See these ones?’ and then he erases an arrow between them. Morgan picks up her phone and starts searching for information. She then points out two additional components of their food web and looks back at Andrew. He draws a line between the elements she pointed out. She points to two more, and he draws an arrow between them as well. This same activity continues for several seconds—they go back and forth between looking over the web, searching for information using the phone, and modifying their web based on the results. At one point, Andrew stops and taps Morgan on the shoulder. He then points to one circle, signs “Eat” and points to a second circle. Morgan nods and Andrew reverses the direction of the arrows. Andrew points to his paper and then to two more web elements, and Morgan once again adds an arrow. This occurs several more times, until Morgan gives Andrew the thumbs up, and they both take their seats at the table, satisfied with their work.

Here, we see extensive use of gesture with minimal signs—only the sign “eat” was used throughout the exchange—alongside written English (via internet searches and hand written web notes on the whiteboard). The prioritizing of gesture allowed the signing group member (Andrew) to communicate effectively with the non-signing group member (Morgan) as they came to a shared understanding of their web. However, this is not the only way that campers’ individual ideolects came into play. Below, we explore an example of Andrew providing a metacommentary about fingerspelling and scientific terminology.

Andrew is chatting with a camp staffer about the difficulty of communicating using scientific terms. “They’re so long to spell and there is no sign for a lot of them,” he complains. He straightens his stance and starts signing formally, in a clear imitation of a camp staff member. “Well, the deoxyribonucleic acid…” he spells. The staff brushes him off, and he returns to see what Morgan has written in his absence.

This second interaction offers an example of the complexity of using ASL in science learning contexts, which we have explored elsewhere (Scott & Cohen, 2023; Enderle et al., 2020), and Andrew’s affective response to that complexity. To fully recognize individual ideolects, both conceptual and affective dimensions must be considered as they can offer different contributions to discursive engagement and sensemaking. For example, although Andrew capably and accurately fingerspelled a sophisticated scientific term (deoxyribonucleic acid) not related to the food web activity, the complaint he expressed could plausibly foster a reticence to engage in further science learning activities in the broader camp experience.

4.8. Vignette 2—Use of Multiple Resources at the Same Time (Coming to a Shared Understanding)

Group one provided a strong example of how these campers, when given access to a large collection of communication resources, used several at once to explain and evaluate each others’ reasoning. This episode occurred about an hour into the food web activity. Michael, Tuan, and Hector were alternating responsibilities for collecting data from the paper and computer tools, while also using the whiteboard space to generate and evaluate their food web model.

Tuan is looking at the whiteboard at their handiwork, and then turns back to the table where Michael and Hector are standing. “Are we satisfied with all of these?” Tuan asks while gesturing to the hard paper copies of the various animals and plants in the food web that are laid out on the worktable before them. “These ones are all done”, he says while gesturing to some. “These ones we don’t need”. As he is signing, Hector picks up one of the papers and walks over to the whiteboard, looking at the information on the card alongside the web they have begun to create. “Hey, wait a minute”, Tuan signs. He picks up one of the papers from the table, and Hector returns to the table with the paper he was holding. The two hold the papers up, side by side, to compare information. Tuan looks between the two cards. Hector points to the name of the animal on the card. “It’s this specific type of animal”, he signed to Tuan. Tuan looks it over. “What’s the name of it? It’s a bird. But I think it’s not this one”, he points to another card, “but it’s that one”. He looks frustrated, puts the paper down on the table and pushes it away. “Get out of here”, he signs to it. He returns to looking over the various cards and starts signing to himself. “Perfect, perfect”. Hector reaches for another paper, which Tuan takes and hands to him. Hector moves to the whiteboard to start writing information from the card. Tuan approaches him to take the paper back, and turns to the computer where he starts a web search for the bird on his card. Hector joins him. “Type it there, these specific words in that order”, Hector signs. Michael joins them in looking at the information on the computer screen. “Is it a plant?” Michael asks. Tuan points to the search results, then opens a new window and begins a new search. Hector holds up a paper to Tuan and points to a specific word. “Add this”, he signs. Tuan types the new word into the search bar. “Oh, I see! Of course. I think these two are the same”. Hector agrees. “It does eat a plant, but the plant has a specific name. Which plant is it?” he asks his groupmates. They continue their web search seeking an answer.

The simultaneous use of paper and whiteboard tools facilitated Hector and Tuan’s back and forth consideration of the relationships they were representing in their model. Although they signed with each other using ASL, a language-based resource, the paper and whiteboard tools provided a significant element of their communication as they endeavored in collective sensemaking about their scientific model. The presence of computers as another translanguaging tool afforded the group another resource to extend this sensemaking effort, motivated by Michael’s return to the group discussion. This scene offers an example of how multiple linguistic resources supported both generative and evaluative work that fostered shared understanding among Michael, Hector, and Tuan.

4.9. Vignette 3—Evaluating Models Extends Sensemaking (Building Knowledge)

The final vignette we present here is focused on group three as they grapple with synthesizing information to update and refine their food web models. This activity occurred after the groups presented their findings to one another, and they were then tasked with returning to their group’s model and making any necessary changes to accurately reflect the ecosystem. This group consisted of Andre, Chris, Isaiah, and Ana.

Group three is standing in front of their food web, which they have just presented to their peers. A camper from group two, Kayla, asks them, “If you removed one species from your food web, which one would have the biggest impact or the least impact on the rest of the ecosystem?” Andre stands in front of his group’s web, his fingers tracing the lines of the arrows as he considers Kayla’s question. Chris turns to his group and signs, “We’ll have to discuss our answer again [based on the new information they received from watching the other groups’ presentations]”. Chris looks at Ana. “Do you think it’s the mosquito that has the biggest impact? Which animal is eaten the most by the others?” He picks up the keyboard and starts a web search. The head camp instructor approaches the group. “Did you decide to keep your first answer, or change to a new answer?” Chris responds, “We are changing our answer”. The instructor asks them to explain their rationale for the change. The group looks at one another without answering, so the instructor points to a section of their web. “You said before that most animals eat algae, right? How many of these lines connect with algae?” Chris counts them up and reports that it is only three. The instructor points out another section of the web and asks, “So is this one more impactful, or less? You may have misunderstood something, so you need to connect with your group and discuss it”. Chris turns to his group mates and suggests they do a deeper dive on the various elements of the food web. “We should be looking at what organisms eat algae versus mosquitoes”. He is searching online for the information. Andre joins him, pointing out different ecosystems in the search results and trying to direct Chris to the relevant information. Ana suggests that they look at which animals eat milkweed. The instructor tells the group to consider the evidence they might need to help them make a decision. “What information do you need to gather to make this decision?” The group continues to work while the instructor leaves them to their research.

The whiteboard models developed by all three groups generated spaces for externally communicating their analyses and the underlying reasoning. In this vignette, we can see how evaluating other groups’ models created productive uncertainty for group three that led to a reconsideration of the claims they generated. Encouraged by the instructor, this group further examined their own model, using this tool as a springboard to extend their evaluative sensemaking efforts. Considering the critical role of their model, as well as other groups’, in engendering further scientific reasoning, the campers were afforded opportunities to learn about the communicative power of tools in STEAM that also rely on the incorporation of artistic expressions. Through engaging with visual representations of information, which are common in scientific work, the groups are learning the language of STEAM. They are learning how to make sense of the food web model and how it communicates information using not only traditional language (e.g., ASL and English), but images and graphics.

5. Discussion

This study examined the translanguaging practices of deaf adolescents enrolled in a STEM camp designed for deaf learners. Overall, both the quantitative and qualitative data presented here point to several important ideas. First, all groups used an array of both language-based and tool-based translanguaging resources as they grappled with their task and communicated their understandings in both intra- and inter-group communication. Second, though certain types of translanguaging resources were more correlated with one another than others, there was no effect of sensemaking task on the selection of translanguaging resources. Rather, the campers seemed to select the translanguaging resources that met the needs of their interlocutors most effectively. Below, we explore our interpretations of these data within the context of the TFDE (Wolbers et al., 2023).

5.1. Limitations

Though this research is a unique contribution to our understanding of translanguaging among deaf learners specifically in the science context, there are limitations to this study that should be considered. First, though time chunk coding has been used in prior research (Pesch & Lumeng, 2017), like all approaches to analysis this has drawbacks. Specifically, we expect that there are certain codes that may be over-represented in certain types of sensemaking activities because they were present within that three minute chunk but potentially not related to the specific sensemaking activity in question. For example, if a three minute chunk included both generative and evaluative activities, and the whiteboard was used strictly for generative discussion by the campers, it would have been coded as present during both types of interactions. Out of 156 coded segments, 54 (35%) included more than one sensemaking code. Though this is only about a third of instances, the reality of potential over-representation of specific codes should be considered. Second, the data here represent a snapshot in time, and are not intended to represent translanguaging in all activity types, or even all STEAM activities as the food web activity is highly aligned with science practices, nor with all types of group compositions. Third, although we have collective experience with assessing the language proficiency of deaf learners, we were not able to collect formal ASL or English proficiency data for these participants. Finally, we are hesitant to draw specific recommendations for teachers and educators on the basis of these data for a number of reasons. First, because these activities did not occur in a classroom context, and second because this analysis is descriptive of what campers did, and did not evaluate which practices may have been more or less effective for understanding, as this was not measured. Despite these limitations, these data provide us with a starting point for understanding how translanguaging may manifest in these spaces, and future research in this area that expands to the area of classroom application could be a fruitful avenue to pursue in order to generate specific recommendations for the classroom.

5.2. Translanguaging in Science, STEM, and STEAM Deaf Education

This study explored the nature of translanguaging interactions as they manifested during science sensemaking activities within the context of a STEM-oriented summer camp for deaf teens. The data we presented here helps to elucidate our understanding of how translanguaging may play out as deaf learners collaborate with one another to solve a problem when provided with an array of options and resources for communication and expression. We present our findings below within the context of the TFDE framework (Wolbers et al., 2023), that is, how our findings support recognizing individual ideolects (how group members and their personal communication approaches seemed to impact how campers chose to communicate with one another), coming to a shared understanding (the use of many language-based approaches and tool-based approaches used in varying combinations to develop and communicate a collective understanding of the sensemaking task at hand), and building (STEAM) linguistic knowledge (how via these interactions campers developed an understanding not only of the concept they were learning about but also ways to communicate that understanding through multiple modes). Though the TFDE framework goes on to describe communicating with an external audience, because our data were limited to an internal activity, we do not explore this outermost layer of TFDE. We explore each of these discussion points below. A summary of these major findings and examples supporting them can also be found in

Table 7. Overview of Findings.

Finding	Examples
Recognizing Individual Ideolects: Nature of Group Interactions	Varied approaches to communicating in groups with varied language backgrounds Campers made adjustments based on their groupmates needs and preferences
Coming to a Shared Understanding: Translanguaging Activities	Co-occurrence of language-based strategies and tool-based strategies Use of gesture to communicate with peers Using fingerspelling to bridge between ASL and print English
Building Linguistic Knowledge: Sensemaking and Engaging with Content	Incorporation of graphics to convey information Evaluation of data and theories led students to reconsider their proposed solutions Model-building as a synthesis practice in STEAM learning environments

.

5.3. Recognizing Individual Ideolects: Nature of Group Interactions

A major finding of this study was that campers’ approach to communicating with peers, including the use of particular translanguaging resources, was at least partially influenced by the communication needs and preferences of the individual ideolects within the group. The t-testing revealed differences in all groups with regard to how they communicated. Because the task and materials were the same across groups, we posit that it was the ideolects of the group members that resulted in these differences. For instance, group one never used spoken language, because all group members were fluent ASL users. In contrast, group two used more gesture than groups one or three, because they had one group member who knew only spoken English, while the others used ASL exclusively. Because written English was the only language they had in common, the group made communication choices that prioritized a means of through-the-air communication that all could understand. Finally, group three used a greater mix of language modalities—though all the group members knew ASL, three of the four also communicated in spoken English, and they used it from time to time. We note that we did not dictate to the campers how to communicate with each other; we simply provided an array of resources and allowed them to select the ones that matched their needs (González-Howard & McNeill, 2020).

Research has previously found that children in mixed hearing/deaf groups are able to adapt to the communication needs of their interlocutor from a young age (Kanto, 2016). A large minority of (30%) deaf students with cochlear implants use signed as well as spoken language to some extent (Hyde & Punch, 2011). This modality flexibility is echoed in these findings. However, because ASL was used in 99% of the interactions, we were not able to determine any kind of statistically significant relationship between ASL and any other variable. Future research may wish to replicate this study with more learners who may not use ASL as exclusively.

5.4. Coming to a Shared Understanding: Translanguaging Activities

Throughout the activity, the campers made use of a number of different translanguaging resources to work their way through the information and come to a shared understanding of how the various pieces of the food web fit together. There were correlations between certain translanguaging activities, including both language-based (ASL, gesture, fingerspelling, spoken English) and tool-based (whiteboard, hard copy paper, computer, phone) events. Individual language-based translanguaging activities tended to be paired with the use of a tool. For example, gesture was correlated with the use of the whiteboard, phone, and hard copy papers. It’s possible that this was used to navigate text presented in written English (a non-native language for most of the campers) to supplement or enhance their understanding. There is prior research to suggest that gesture may be an important factor in the writing process among hearing children (Rowe, 2019). Gesture can even be found to take on language-like forms when it is used for direct communication (Goldin-Meadow, 1999; Pierson et al., 2021), as we saw with Andrew and Morgan. As translanguaging has been observed in the arts (Leung, 2019), there may be an artistic component to how gesture was combined with other means of communication.

Another language-based translanguaging activity that had correlations with tool-based activities was fingerspelling. The use of fingerspelling appeared more frequently when campers were also utilizing the hard copy papers, which had both pictures and printed sentences about the plants and animals in the ecosystem under study. It has been argued previously that fingerspelling is a bridge to printed English (Haptonstall-Nykaza & Schick, 2007; Stone et al., 2015), and even as a translanguaging product explicitly connecting signed and printed languages (Lee & Secora, 2022). A contrasting finding, however, was the negative correlation between fingerspelling and use of the whiteboard—a primarily writing focused activity where one might expect to see increased fingerspelling, as campers moved from their signed conversations to the written word. It is possible, however, that by the time the campers moved on to writing on the whiteboard that they had already made the necessary connections between what they were reading on the hard copy papers and how they wanted to express that in their own food webs. It is also worth noting Andrew’s commentary about fingerspelling being less helpful in accessing and understanding science content (Enderle et al., 2020; Clark et al., 2021).

5.5. Building Linguistic Knowledge: Sensemaking and Engaging with Content

In this study, informational and generative discourse was comparable in volume among the campers. This could suggest that these campers were not familiar with the epistemic practice of building and refining models (e.g., evaluative discourse, which appeared much less frequently across groups). A practical implication of this pattern could be that deaf students do not get many opportunities to do this in formal settings, although such rigorous work is recommended (Easterbrooks & Stephenson, 2006). A scaffolded generation of explanatory knowledge artifacts, including arguments and models, will be necessary. Using translanguaging as a framing for such scaffolding, linguistic and epistemic sensemaking can be leveraged to support each other, as seen in studies involving hearing students (Pierson et al., 2021; Suárez & Otero, 2024). However, similar to critiques offered by previous studies involving a variety of learners, generative work must not be the only type of science sensemaking opportunities offered, as engaging in evaluative work offers further enhancement of learners’ science proficiency and uptake of epistemic agency (Osborne, 2014; Schwartz et al., 2023; González-Howard & McNeill, 2020).

We also note that in vignette three, we see an instance of the campers changing their mind about a previously reached conclusion. This as well is a demonstration of the building of science knowledge via collaboration and the importance of students engaging in evaluation of their own work to determine whether their conclusions are appropriate (Kelly et al., 2023; Waldrip & Prain, 2017). The impact of the instructor’s targeted questions in conjunction with this group’s whiteboard model also demonstrates how teachers’ use of various linguistic and epistemic scaffolds enhances the sensemaking learners experience (Kelly et al., 2023; Licona & Kelly, 2020; Suárez, 2020). Prior research has suggested a model-based inquiry approach to science learning as a valuable experience for youth learning about science (Windschitl et al., 2008). Not only that, but the creation and revision of the food web models themselves allowed the campers an insight into the unique ways that science can be communicated via graphics (Scott et al., 2021). Future research may consider additional ways that the arts can be more thoroughly explored as they contribute to deaf students’ learning of science, math and engineering concepts.

Indeed, a focus on the epistemic practices of science, such as developing and using scientific models emphasized in this study, provides an opportunity to synthesize resources across the component disciplines of STEAM, including art. Bevan et al. (2019) provide a potential alignment of those practices in an effort to develop a set of STEAM epistemic practices. When considering the practice of scientific modeling, learners must consider the most critical elements of a system that should be included in a model to make it a useful tool for sensemaking. Such considerations involve abstraction, moving beyond just creating graphic or physical representations of important system elements to also determine ways to communicate the underlying concepts that explain how those elements interact. Similarly, the arts employ various principles to augment the meaning and messages being communicated by an artistic work, moving beyond simple considerations of the media and structure of the work. Both practices focus on creating tentative representations of complex ideas that can be refined through evaluative interactions, like those described above. As such, efforts to move from more one-sided instrumental visions of STEAM teaching and learning toward more mutually instrumental visions should consider ways to explicitly include both scientific and artistic principles in the generation and evaluation of knowledge products developed during learners’ sensemaking efforts (Mejias et al., 2021; Bevan et al., 2019).

6. Conclusions

This study extends our understanding of the way that deaf students use translanguaging activities to engage in conversations and sensemaking with their peers in the context of STEAM learning. Though translanguaging as a theoretical model has been argued as a potentially useful framework for understanding how deaf students use language resources (De Meulder et al., 2019; Hoffman, 2014; Swanwick, 2017), even specifically in disciplinary contexts like science, STEM, and STEAM learning (Scott & Cohen, 2023), this is among the first studies to observe how deaf learners use multiple linguistic resources in such contexts. Importantly, the campers in this study were not told which resources to use, or about the concept of translanguaging at all—they simply noted the ways that were present in their environment to facilitate their communication, and took advantage of them.

We found that most groups took advantage of the full array of communication options that were available, though they did not use them in parallel ways. Our findings indicate that deaf students naturally make affordances in their communicative approaches based upon their observation of their own linguistic needs, as well as the needs of their peers. It will be important for future researchers to continue the study of these group interactions and whether they vary across content and contexts. The potential for the natural use of translanguaging by deaf learners to connect with peers and engage with content is compelling, and may hold significant explanatory power for sensemaking activities.