In this study, we report results from a novel coding of the Measures of Effective Teaching (MET) Study data that offers evidence on a set of teacher discourse measures in the domain of teacher support including: public praise vs. admonishment, autonomy support vs. controlling language, strategy suggestion vs. lack thereof, and discourse supporting (vs. undermining) learning mindsets. Novel coding of these constructs is paired with extant measures of instruction and achievement in the MET data. Several of the newly coded discourse measures have promising features, including high lesson- and teacher-level variability, and convergent and discriminant validity with existing protocols. We also report possible associations with change in achievement over two years.