A demand exists to better prepare elementary teachers to improve math content and pedagogical competence for increasing numbers of English language learners (ELLs) (Grossman, Schoenfeld, & Lee, 2005). This paper will share how QTEL grant graduates demonstrate linguistically and culturally responsive math teaching (LCRMT) strategies in their classrooms with ELLs. A LCRMT framework is developed based on research on linguistically responsive teaching (Song & Simons, 2014; Commins & Maramontes, 2006), culturally responsive pedagogical content knowledge of mathematics (Aguirre, Zavala, & Katanyoutnant, 2012; Schleppegrell, 2007; Turner & Drake, 2016), and sociocultural constructivists' perspectives on student interactions (Vygotsky, 1978; Lemke, 1990; Engestrom, 2001; Polman, 2004). The LCRMT framework has two dimensions; three mathematics-related competences for the 1st dimension and three how-to actors for the 2nd dimension. The first dimension has math content competence, math discourse competence, and math pedagogical competence (Halliday, 1978; Schleppegrell, 2007; Aguirre, Zavala, & Katanyoutnant, 2012). The 2nd dimension includes know in-depth, develop cognitive and procedural demands, and examine and develop crosscultural and sociopolitical beliefs (Song & Simons, 2014; Commins & Maramontes, 2006). The three how-to actors under each of the math-related competences will be described. The dichotomy of math content, math discourse, and/or math pedagogy is complex and not easily separated (Moschkovich, 2007), and the LCRMT framework can be used as a planning and assessment tool for teaching mathematics to ELLs. This qualitative study's research question is: "How do elementary inservice teachers demonstrate their LCR math teaching practice in classes with ELLs after they graduated from QTEL program?" Twelve elementary preservice and eight inservice cohort teachers with ELLs, who were mostly white, female, and monolingual, participated in the 2015 QTEL program. Data analysis continues. Preliminary results from an inservice teacher were summarized; a full research methods and results will be presented at the 2017 convention. Activity theory was used for examining and reporting qualitative data (narratives and observations) through identified "activity systems" (Yamagata-Lynch, 2010, p.90 & Engestrom1987). Coding provided two activity systems (Engestron, 1987) to compare: the university teacher training (Triangle 1 - e.g., theory-based LCRMT) and what graduates actually taught (Triangle 2). The first activity system had texts, LCRT strategies, and math content pedagogy knowledge for Tools, teachers in QTEL training for Subject, LCRMT competences for Object/Goal, cooperative learning; LCRT, and LCRMT for Rules, and TESOL instructors for Division of Labor. The second activity system for QTEL graduates had lessons, manipulatives, books, and computers for Tools, teachers for Subject, math teaching for Object/Goal, classroom, order, routine, rules for talking for Rules, and teacher or learner-directed teaching for Division of Labor. Graduated teachers preferred school-provided routines and rules over LCRMT strategies; used SIOP, but did not give individualized language anchors for ELLs. Researchers were puzzled to find LCRMT strategies for ELLs lacking A contribution of this paper is this juxtaposed finding, a gap between University training and actual teaching, which will provide a benchmark for conversations about teaching academic content like Math and social languages to the increasing diverse student populations.