A naming paradigm was used to explore the effect of the syntactic structure of a sentence fragment on recognition of a subsequent target word when the syntactic context trials were mixed with the semantic context trials. Target words (Japanese verbs) were visually presented 500 msec. (Experiment 1) or immediately (Experiment 2) after context Japanese sentence fragments disappeared. In Experiment 1, the target words following the syntactically congruous contexts were named faster than those following syntactically incongruous contexts, even if the target words semantically unrelated to the contexts. Semantically context effects were not observed. In Experiment 2, both syntactic and semantic context effects were marginally significant. The target words in CR (syntactically congruous, semantically related) condition were named faster than those in any other condition. The implications of these results for the relationship between task structure and sentence integration processing are discussed.