We report an effort to replicate and extend Levelt and Kelter's (1982) Experiment 3. In their study, experimenters phoned businesses and asked, "At what time does your shop close?" or "What time does your shop close?" Participants were more likely to produce prepositional responses (At 7 o'clock) to questions containing a preposition than to questions that did not. Across three experiments, we replicated Levelt and Kelter's method, and included a number of other communicative contexts (asking questions by phone, in person, or by text message, and asking people both known and unknown to the experimenter). Our efforts produced mixed results. Across experiments, and within each communicative context, we found the predicted pattern of results (more prepositional responses following prepositional questions). Although this pattern was not statistically reliable in our individual experiments, it was significant in an analysis conducted across all of our experiments. Overall, our results suggest that the structural repetition effect observed by Levelt and Kelter is replicable, but is weaker than the effect reported in the original paper.