Two experiments were performed to determine the extent of increase in articulation times with nonsense words containing strings of phonetically similar consonants, compared with when the successive consonants belonged to different articulatory categories. An average increase in excess of 33 per cent was found even when stuttering and other speech prolongations were controlled. This figure exceeded the increase in articulation times obtained under delayed auditory feedback. The absence of an interaction between the latter and the degree of articulatory difficulty of the material was considered consistent with a motor hypothesis, rather than with a model emphasizing such sensory effects as feedback or trace monitoring. Articulation was considered as a complex motor skill subject to such phenomena as interference between components and process overlap.