ABSTRACT Children with developmental language disorder (DLD) have significant deficits in language understanding and production. In particular, these children demonstrate deficits in processing sentences with noncanonical word orders like passives. However, some passives are more difficult to process than others for these children. Reversible passives like The cat was followed by the dog tend to be more difficult than non-reversible passives like The bone was eaten by the dog. In fact, children with DLD demonstrate similar levels of comprehension of non-reversible passives to children with typical language development (TLD) but show much more difficulty with reversible passives even when compared to younger children with otherwise similar levels of language development. The evidence currently suggests that the comprehension of non-reversible passives is aided by rich semantic-pragmatic contextual cues that are often unavailable in reversible passives, which often must be processed on the basis of syntactic cues alone. Important cues to the interpretation of passives are animacy and event probability. Although the literature in TLD suggests that these cues interact in sentence processing, it is unclear how these cues are separately used by children with DLD to improve their passive sentence processing. We propose to examine how this set of semantic cues influences the processing of passives in children with and without DLD by setting up expectancies for agents to participate in certain actions more than others. We will show agents participating in different sets of highly contrastive semantically coherent actions (e.g., clumsy actions: slipping, bumping, tripping vs. affectionate actions: hugging, kissing, petting), thereby biasing children to expect that an agent will participate in a similar action (e.g., falling over) more than another (e.g., tickling). In an eye-tracking task, we will present reversible and non-reversible passives to children along with videos and images depicting the target event and foils. Some sentences will have agents and verbs that match the biases previously set up. We expect that those sentences that demonstrate agent-verb matching will influence children’s processing of passive sentences, particularly for reversible passives. We will also measure children’s verb-specific semantic development using a repeated word association task. We hypothesize that the children’s performance on the semantic association task will be related to the degree to which children are able to use the expectancies for the agents in the sentence prediction eye-tracking task. We hope that this information will clarify sentence processing in these children and the role of lexical semantic knowledge. We also anticipate that the results of this study will help clinical practice by identifying potentially useful cues in the processing of passive sentences, cues that may be targeted in therapeutic settings.