Minimalist theories of spoken language planning hold that articulation starts when the first speech segment has been planned, whereas non-minimalist theories assume larger units (e.g., Levelt, Roelofs, & Meyer, 1999a). Three experiments are reported, which were designed to distinguish between these views using a new hybrid task that factorially manipulated preparation and auditory priming of spoken language production. Minimalist theories predict no effect from priming of non-initial segments when the initial segment of an utterance is already prepared; observing such a priming effect would support non-minimalist theories. In all three experiments, preparation and priming yielded main effects, and together their effects were additive. Preparation of initial segments does not eliminate priming effects for later segments. These results challenge the minimalist view. The findings are simulated by WEAVER++ (Roelofs, 1997b), which employs the phonological word as the lower limit for articulation initiation.