In the U.S., Black and Latinx/e students face educational disadvantages, especially in subjects belonging to the STEM field (Gutiérrez, 2012). Schools have been identified as one source contributing to such disparities (Michelmore & Rich, 2023). Within the school system, teachers and their practices play a critical role. Previous research has shown racial-ethnic biases in classroom instruction (e.g. Tenenbaum & Ruck, 2007). In mathematics, teacher biases can be a result of lower expectations towards pupils from minoritized groups, particularly Black girls (Copur-Gencturk et al., 2019). Whereas teacher biases have been extensively investigated in the domain of academic judgments and disciplinary referrals, to our knowledge, teachers’ natural language in approximations of teaching, particularly in the domain of ambitious math instruction, have not been investigated experimentally. Teachers’ language in their in-the-moment responses to students is a site that may be particularly sensitive to biases, as biases are most salient in non-reflective, automatic processes. We thus aimed to investigate teachers’ spontaneous spoken responses to students’ mathematical explanations in an experimental setting. Participants were N=271 teachers of record in the U.S. The study had a one-factorial between-persons design in which participants were randomly assigned to one of three conditions: a classroom with predominantly Black, Latinx/e or white students, reflecting the relatively segregated nature of American classrooms. Teachers were presented with six hypothetical fourth grade classroom vignettes. The vignettes showed different mathematics tasks and student explanations of their problem-solving process. After reading the target student explanation, teachers were asked what they would say and do next. Using voice recording software embedded in our survey panel, we captured teachers' spoken responses. We employ natural language processing methods to decipher topics by cohesion, identifying unique topics ranging from mathematically-focused language (e.g. number line) to process-oriented language (e.g. explain, who thinks). We also apply a sentiment analysis using the BING dictionary. We find significant differences in the topics discussed by experimental condition. Teachers showed, for example, greater likelihood of affirmative but little mathematical language for the Black condition relative to the White condition. Additionally, we find a tendency in sentiment that teachers were positive toward Black students compared to white or Latinx/e students. Finally, we find more words spoken on average in the white classroom condition compared to the Black and Latinx/e conditions. Our findings have implications for anti-racist teacher education tied to specific instructional domains in ambitious math teaching.

Copur-Gencturk, Y., Cimpian, J. R., Lubienski, S. T., & Thacker, I. (2019). Teachers’ Bias Against the Mathematical Ability of Female, Black, and Hispanic Students: Educational Researcher. https://doi.org/10.3102/0013189X19890577 Gutiérrez, R. (2012). Context matters: How should we conceptualize equity in mathematics education? In B. Herbel-Eisenmann, J. Choppin, D. Wagner, & D. Pimm (Eds.), Equity in discourse for mathematics education: Theories, practices, and policies (pp. 17–33). Dordrecht, the Netherlands: Springer. doi:10.1007/978-94-007-2813-4_2 Michelmore, K., & Rich, P. (2023). Contextual origins of Black-White educational disparities in the 21st century: Evaluating long-term disadvantage across three domains. Social Forces, 101(4), 1918-1947. https://doi.org/10.1093/sf/soac098 Tenenbaum, H. R., & Ruck, M. D. (2007). Are teachers’ expectations different for racial minority than for European American students? A meta-analysis. Journal of Educational Psychology, 99, 253–273. https://doi.org/10.1037/0022-0663.99.2.253