¿Por Qué No Los Dos? Multilingual Spanish Speaking Student Practices in Mathematics Classrooms

• In Event: Straight Talk: Natural Language Processing Approaches to Characterizing Student Opportunities to Learn in Mathematics

Abstract

Historically, the language practices of bilingual and multilingual youth in the United States have been treated with a deficit lens as part of a racial hierarchy based on cultural inferiority (Flores, 2024; Rosa, 2019). In schooling, this has been exhibited through bilingual and English-Only programs that worked to assimilate youth into a monolingual English dominant culture rather than encourage youth to use their full linguistic abilities as an asset. In contrast to these deficit-oriented programs, the field of Mathematics Education has emphasized the need for students to engage in rich disciplinary discourse practices using all of their multilingual resources (Erath et al., 2021). As students use different language varieties to engage in mathematical discourse practices, they often engage in translanguaging, a concept that describes the ability of multilingual speakers to pull from a fluid linguistic repertoire (García and Li, 2014; Poza, 2016). Recent studies have shown that computational methods, like Natural Language Processing (NLP), can be used to study student and teacher talk moves in classroom conversations (Author et al., 2023). NLP models can be used to outline the linguistic patterns that might not be otherwise apparent in student and teacher communication and connect this to learning outcomes (Author et al., 2023; Chen et al., 2015; Reitman et al., 2023). In this project, we use NLP to investigate patterns in how middle school students engage in various mathematical language practices across different language varieties, i.e. while translanguaging. This analysis was done as part of a larger study investigating the math language practices of multilingual Spanish speaking youth in a large urban school district in the San Francisco Bay Area. The data include 65 transcripts of grade 6-8 students collaborating on math tasks. Each group includes at least one student who reports speaking Spanish at home. Expert mathematics educators qualitatively coded a set of these transcripts to compare with and validate NLP outputs. Transcripts were coded for the presence of mathematical language practice features in utterances tagged as English, Spanish, or both. In this project, we analyze patterns in how students use different language varieties to achieve mathematical discourse practices. We also investigate how the home languages and language use of peers influence how other Spanish speaking youth speak during these observations. This research contributes to better understanding of how multilingual Spanish speaking youth utilize their linguistic repertoires in mathematics classrooms, with implications for how to support the noticing and encouragement of these practices.

Authors

• Liliana Santos-Deonizio, Stanford University

• Jim Malamut, Stanford University