The boy ate what? Gradient effects of context in sentence processing
A pervasive issue in psycholinguistics is the question of how information arriving at any instant (e.g., an incoming word) is integrated with the “contextual” information that came before it (e.g., the sentence context). While results from a number of classic studies support late integration (such that context-inconsistent representations are initially activated; e.g., Swinney, 1979; Tanenhaus et al., 1979), results from recent studies using the visual world paradigm support rapid, immediate context integration (such that context-inconsistent representations are fully suppressed; e.g., Dahan & Tanenhaus, 2004). Here, we report on a series of visual world experiments that show gradient effects of sentence context information. For example, listeners heard sentences like “The boy will eat the white…, ” while viewing visual displays with objects like a white cake (i.e., a predictable direct object of “eat”), white car (i.e., an object not predicted by “eat,” but consistent with “white”), and unrelated distractors. We found that while listeners fixated the white cake most, they also fixated the white car more than unrelated distractors. These results suggest that in addition to (anticipatorily) activating representations that are consistent with the sentence context, comprehends also gradiently activate (lexically-related) representations that are inconsistent with the sentence context. We show that this gradient pattern is predicted by an artificial neural network framework.