An experiment involving object naming within connected speech examined the effect of codability on disfluency. Participants described cartoon video clips in which the characters interacted with objects of both high and low codability. This method was utilised as an alternative to the Network Task (Levelt, 1983; Oomen & Postma, 2001) in an attempt to examine disfluencies in spontaneous speech for moving images rather than the still images presented in the Network Task. A social manipulation was included to examine whether the implication of a social usage for participants’ speech affected the number and pattern of disfluencies compared to a non-social usage. Consistent with the hypothesis, the experiment revealed that the probability of at least one type of disfluency was higher when codability of the object was low. Additionally, the probability of a pause was also higher when the codability of the object was low. However there was no effect of the social manipulation on number or pattern of disfluencies. These results suggest that objects with low codability cause more difficulties during lexical processing, leading to disfluency.