Current reforms in mathematics education place dialogue at the heart of the development of conceptual understanding. Underlying these ideas is strong criticism of transmissive teaching styles, often referred to as "teaching by telling" However, there is little in terms of specific guidance for teachers about how best to achieve these reform aims, which has led to what Baxter and Williams (2010) describe as "the dilemma of telling" This article summarizes an exploratory reflective action research study into this issue that the author conducted in the context of a 1318 comprehensive (all-ability) school in England. This article addresses the following research questions, in order to demonstrate that such research-based work can lead to teacher change: RQ1: What does an analysis of the form and function of my utterances reveal about my scaffolding strategies? RQ2: Can the form and function of these utterances be changed? RQ3: What aspects of my scaffolding strategies did students value? There were two action-research cycles, 20 weeks apart. Participants were four Year 13 students (age 1718) selected by purposive sampling from six male volunteers in a class of 12. In order to address RQ1 and RQ2, the author used individual task-based clinical interviews in which my responses as interviewer were contingent on the participants' reactions to the task (Ginsburg 1997). In order to address RQ3, individual semi-structured interviews were conducted after each clinical interview. All interviews were audio-recorded in a quiet area adjoining the classroom and transcribed promptly. The author kept an analytical log to evaluate the interviews, also recording thoughts, feelings and insights in an attempt to unearth the intentionality behind my utterances. Results are summarized in this article.