ABSTRACT
Local pragmatic anomalies arise when we struggle to incorporate certain utterances into the context of our real world knowledge. For example, The cat picked up the chainsaw (Filik & Leuthold 2008) is anomalous with our world knowledge; the utterance is discourse independent and instead relies on our understanding that cats lack opposable thumbs or mechanistic know-how. Yet, in an alternate world, like Tom and Jerry, the above scenario would be perfectly plausible. If we hear an utterance like The cat picked up the chainsaw in a cartoon context, can we avoid the anomaly? And if we can, to what processing cost? And how much context would be required to override our background assumptions?
To address these questions regarding the incorporation of context and real world knowledge in assessing our understanding of an utterance, an experiment using the G-maze (i.e. grammaticality maze) task was conducted (Forster et al. 2009). In the maze task, participants see two words simultaneously displayed on a computer screen. Employing their working memory, they decide via keyboard response which word in the set represents the best grammatical fit for continuing the sentence; decision times are recorded. After reading a one-sentence context that was either neutral in its pragmatic content or at odds with world knowledge, participants were presented with a maze sentence which related to the context. Interestingly, results show that participants were faster to respond (by 33 ms) to anomalous scenarios than they were to neutral scenarios. This finding of pragmatic anomalies being processed faster than information that aligns with world knowledge has significant ramifications for current pragmatic frameworks.
