The “language of content”—the specialized language that students use in math and science—is vital for historically marginalized learners, including bilingual and multilingual, students of color, students with limited or interrupted formal education (SLIFE), economically disadvantaged, and other learners. By learning the language of content, students can unlock their full potential and become the owners and critical communicators of their knowledge (Durham & Ingram, 2016).
Unfortunately, content language is rarely taught in K-12 math and science classrooms. Students are oftentimes taught to achieve English proficiency, rather than being scaffolded to achieve both English and content language proficiencies (Brown & DiRanna, 2012). While it is not a surprise that bilingual and multilingual students are likely to encounter barriers in developing both proficiencies (Bailey et al., 2018), those who have achieved sufficient English proficiency also benefit from the facilitation of content language learning to effectively gain comprehension of the subject (Brown & DiRanna, 2012). When the facilitation of content language is absent, student academic achievement is put in jeopardy and thereby leading to a learning gap (Brown & DiRanna, 2012).
Why is this the case? Many math and science teachers know the value of content language. After all, it’s embedded in all aspects of science, technology, engineering, and math (STEM) content standards, from explaining math reasoning to crafting and defending scientific claims. But teachers face challenges (de Jong & Barko-Alva, 2015) and often feel that they lack the time, resources, and expertise to explicitly teach the academic language their students need.
To address this challenge, Speak Agent collaborated with the Learner Variability Project (LVP) to source research-based strategies they could embed into instruction. Learner variability is the idea that each learner has a unique set of strengths and challenges that interconnect and vary within each individual context (see Rose, 2017; Pape, 2019). The Learner Variability Project, and its flagship free, web-based tool, the Learner Variability Navigator, synthesizes research on the factors that matter for learning and strategies that support those factors to help educators and educational technology (edtech) developers serve diverse learners.
The Learner Variability Navigator includes learner models divided by developmental groups from grades PreK-12 in literacy and math, as well as an adult learner model, all embedded in research and developed by researchers and practitioners through an iterative and collaborative process. Each model includes four core whole learner domains and their key factors:
Through the Learner Variability Navigator, we distilled the research about effective strategies into actionable information that informed Speak Agent’s product designs. During the collaboration, our support of Speak Agent included a design review at each step of the development process to search for opportunities to integrate learning science and research-based design strategies for promoting the content language proficiency of all learners.
As a result of our collaboration, the Speak Agent development team integrated 36 strategies into the learning activities that they built. For example, a math strategy allowing student-generated problems was integrated to help students connect math concepts to their background knowledge and lived experiences. Dialogic reading was integrated to foster students’ content language and literacy skills through both guided and independent reading activities. You can find detailed descriptions of many of the research-based strategies used by Speak Agent that were selected from our Learner Variability Navigator:
In addition, Speak Agent has provided a resource webpage where educators can learn more about the strategies and download pencil and paper resource packs to support classroom instruction—without any technology needed.
A recent study commissioned by Prince George’s County Public Schools (PGCPS) in Maryland and conducted by Leanlab Education at 1,597 class sections representing 24,180 unique PGCPS students shows both evidence of the link between using Speak Agent supports for language and academic success and a pathway to narrowing the learning gap. The study provided two important findings.
First, multilingual learners in grades 6 to 8 who used Speak Agent’s Math+Language program for the recommended dosage (approximately 10 one-hour weekly activities per quarter) improved their English language proficiency scores by 5 percent. The following chart shows predicted overall ACCESS (summative English language proficiency assessments) scores for English Learners based on the number of Math+Language learning activities completed.
The predicted overall ACCESS scores for PGCPS English Learners, based on the number of Speak Agent Math+Language learning activities completed. Source: Huebert, 2022
Second, the general population of learners in grades 6 to 8 who used Speak Agent’s Math+Language program for the recommended dosage improved their math assessment scores by 10 percent, or double the rate found with English language proficiency. This provides evidence for the link between supports for academic content language and academic achievement. What’s more, the population in the study overwhelmingly consisted of historically marginalized learners (92 percent Black & Brown, 21 percent multilingual, 66 percent economically disadvantaged).
The following chart shows predicted math benchmark percentage scores for all learners at PGCPS based on the number of Math+Language learning activities completed.
The predicted math benchmark percentage score for PGCPS students, based on the number of Speak Agent Math+Language learning activities completed. Source: Huebert, 2022
The results of this collaboration demonstrate how using the Learner Variability Navigator’s research base of factors and strategies can help provide opportunities for students who are bilingual, multilingual, or not exposed to academic language to succeed in the classroom and beyond. Speak Agent’s findings highlight the positive association between academic content language and academic success. They also suggest a model of how integrating research-based, whole-learner supports can provide a pathway to improve learning for all students, perhaps especially those who face particular barriers to learning in traditional education contexts.
Embracing how learners vary across a whole learner spectrum can help to meet the needs of a broader diversity of students, especially those who have been historically and systemically excluded from opportunity. The finding that using more Speak Agent supports for language was associated with a larger gain in students’ math scores than their language proficiency is a telling example of how interventions that consider the whole learner, not just a limited view of their content knowledge, can better serve learners. These research-based design considerations may be particularly well suited for edtech, where products designed with learner variability in mind can supplement traditional instruction and optimize their impact on student learning.
Integrating research-based strategies into digital educational products requires additional effort as well as a depth of learning sciences expertise that many edtech startups may not have the capacity for. Partnerships between edtech companies like Speak Agent, and free resources like Digital Promise’s Learner Variability Navigator, can provide ready-to-implement research-based strategies to support learners of all varieties.
Today’s learning environments are dynamic, accommodating learners with a broad diversity of needs, cultures, and experiences. Using actionable research and collaboration to inform the design of digital learning platforms is critical to supporting this diversity. Integration of learning sciences research, therefore, is both necessary and attainable for edtech developers seeking to make a significant impact on K-12 educational outcomes.
To learn more about the Learner Variability Navigator, visit https://lvp.digitalpromiseglobal.org/.
References
Bailey, A. L., Maher, C. A. & Wilkinson, L. C. (2018). Language, literacy, and learning in the STEM disciplines. In A. L. Bailey, C. A., Maher & L. C. Wilkinson (Eds.), Language, literacy, and learning in the STEM disciplines: How language counts for English learners (pp.1-10). Routledge, NY: New York.
Brown, Z. A., & DiRanna, K. (2012). Equal access to content instruction for English learners: An example from Science. Region IX Equity Assistance Center at WestEd. https://files.eric.ed.gov/fulltext/ED580362.pdf
de Jong, E., & Barko-Alva, K. (2015). Mainstream teachers in two-way immersion programs: Becoming content and language teachers. In Y. S. Freeman & D. E. Freeman (Eds.), Research on preparing inservice teachers to work effectively with emergent bilinguals (Vol. 24, pp. 107-126). Emerald Group Publishing Limited. https://doi.org/10.1108/S1479-368720150000024005
Durham, P., & Ingram, J. M. (2016). Viewing content curriculum through the lens of language acquisition: A content analysis. Read: An Online Journal for Literacy Educators, 1(2), 6-17. https://www.shsu.edu/academics/school-of-teaching-and-learning/journals/read-journal/documents/READ%20Journal%20June%202016%20Final.pdf
Huebert, E. (2022). Effect of Speak Agent on Math and English language proficiency scores at middle grades in Prince George’s County Public Schools. Learnlab Education. https://www.speakagent.com/evidence
Pape, B. (2018). Learner variability is the rule, not the exception. https://digitalpromise.dspacedirect.org/server/api/core/bitstreams/a2e9f53f-d353-4a2b-875a-b176467f926c/content
Rose, T. (2016). End of average: Unlocking our potential by embracing what makes us different. Harpercollins Canada.
