While mathematics education researchers have long characterized student performance marked by mathematical explanations, arguments, and justifications as evidence of mathematical reasoning and understanding (e.g. Schoenfeld, 1992), current education policy has begun to move in a similar direction, emphasizing sense making and mathematical communication as features of mathematics education (Common Core State Standards Initiative, 2010). However, designing and implementing instruction with these features is challenging (e.g. Lampert, 1990). Furthermore, classroom and instructional norms norms must be carefully developed for this discursive-heavy instruction to be equitable (e.g. Boaler & Staples, 2008). If mathematical discussions are to be a feature of the mathematics classroom, then teachers must learn to learn from their own teaching to enact practices that promote discussion (Hiebert, Morris & Glass, 2003).

This study is a qualitative investigation of how three novice middle-school mathematics teachers learned to promote in-class student discussion, with a focus on the features of and strategies for instruction to which they attended, as well as their negotiation of challenges that arose during practice. Supported by a mentor, these teachers participated in a reflective teaching cycle that included a continuing teacher seminar, planning sessions, classroom observations, and reflection sessions over the course of 5 months. Through case studies, these teachers' instructional planning, practice and reflection were analyzed. Each case offers a perspective addressing how a teacher approached promoting student sense making through discussion, the challenges faced, and how those challenges were negotiated.

Cross-case analysis yielded five findings. First, the teachers found that building relationships with their students encouraged student participation in discussion. Second, the teachers were able to leverage accountability in the design and implementation of their lessons. Third, school context either supported or impeded the teachers' ability to engage students in discussion. The fourth finding illuminated the ways in which the organizational practice of tracking students impacted teacher perceptions and eventual decision-making. The final finding clarified the effect of mentoring support on teacher efficacy and self-efficacy. These findings have implications for mathematics teacher education, as well as induction mentoring programs.