This article reports on mixed methods classroom research carried out at a British university. The study investigates the effectiveness of two different explicit teaching frameworks, Illustration--Interaction--Induction (III) and Present--Practice--Produce (PPP) used to teach the same spoken discourse markers (DMs) to two different groups of Chinese learners and compared to a control group. Univariate analysis of the pre-and post- tests indicated statistically significant differences between the PPP group and III/control groups in terms of a higher mean usage of the target DMs in the immediate post-test. Qualitative results demonstrated that the PPP group generally found this method to be more useful, which tallied with their better performances in the tests. Both groups also articulated a desire for a different kind of practice to be used in class, based on rehearsal forreal world tasks. This suggests a need to re-conceptualise practice within III, PPP or other teaching frameworks.