The present study investigated if disfluencies arise from difficulties in lexical access and whether disfluencies in speech serve a social function to an utterance. Number and type of disfluencies were measured for a video description task in either one of two different audience design conditions. Previous research has used the network description task to study if disfluency is a product of difficulties with lexical access. This involves participants describing a network of pictures with paths connecting them, while manipulating the name agreement of the pictures. The present study develops this approach by asking speakers to describe motion animation with the intention to better reflect the context and complexity of the real world, while also manipulating name agreement of the pictures in the task. Consistent with our experimental hypothesis, there was more disfluency before low name agreement pictures than high name agreement pictures. In addition, previous research has implied the production of disfluencies provide a social function to an utterance. We manipulated audience design to explore this. Participants were either in a social condition which emphasised the need to communicate effectively with another person or in a non-social condition which did not emphasise the need to communicate effectively with another person. However, inconsistent with the research hypothesis, there was no effect of condition on the proportion of disfluencies produced by speakers. The significant effect of name agreement on disfluency production implies that difficulties in language production, namely lexical access, result in a greater proportion of disfluency. The present study informs us that lexical properties influencing disfluency production can be manipulated in a more realistic description task than previous methods used to clarify the primary influencing factors.