We report an experiment in which participants first heard a context sentence and saw a picture, and then had to produce a coherent discourse by describing the action of a previously introduced referent in a target picture. They described the picture either to an addressee or in the absence of an addressee. The results showed that participants produced more pronouns and fewer repeated noun phrases when the referent was mentioned as the subject than the object in the preceding sentence and when a competitor was not mentioned in the preceding sentence than it was. These linguistic saliency effects were not affected by the presence or absence of an addressee. In contrast, the effect of visual presence of a competitor did interact with addressee presence. In the presence of an addressee, speakers produced more pronouns when there was no visually present competitor than when there was, but no such visual saliency effect occurred in the absence of an addressee. We conclude that speakers take into account the referent's visual saliency for the addressee's benefit, but the linguistic saliency effect is independent of whether the addressee is present.