Visually situated spoken words activate phonological, visual, and semantic representations guiding overt attention during visual exploration. We compared the activation of these representations in children with and without developmental language disorder (DLD) across four eye-tracking experiments, with a particular focus on visual (shape) representations. Two types of trials were presented in each experiment. In Experiment 1, participants heard a word while seeing (1) an object visually associated with the spoken word (i.e., shape competitor) together with a phonologically related object (i.e., cohort competitor), or (2) a shape competitor with an unrelated object. In Experiment 2 and 3, participants heard a word while seeing (1) a shape competitor with an object semantically related to the spoken word (i.e., semantic competitor), or (2) a shape competitor with an unrelated object. In Experiment 4, children heard a word while seeing a semantic competitor with (1) the visual referent of the spoken or (2) with an unrelated object. The visual context was previewed for three seconds before the spoken word, except for Experiment 2, where it appeared at the onset of the spoken word (i.e., no preview). The results showed that when a preview was provided both groups were equally attracted by cohort and semantic competitors and preferred the shape competitors over the unrelated objects. However, shape preference disappeared in the DLD group when no preview was provided and when the shape competitor was presented with a semantic competitor. Our results indicate that children with DLD have a less efficient retrieval of shape representation during word recognition compared to typically developing children.