Investigation of Plausibility, Voice and Implicit Cognition within Shallow Processing

Abstract

Shallow Processing (Underspecification or ‘Good Enough Representations’), is a psycholinguistics concept that suggests comprehenders of language can fail to build, or retain, fully specified representations of sentences. This can result in erroneous interpretations that are based on heuristics at the expense of a full and detailed analysis using syntax. Shallow Processing can also lead to pragmatic normalisation of the true sentence, resulting in ‘real-world’ interpretation. However, heuristics are faster in situations where there are time constraints or sentences are incomplete and comprehensive analysis might be neither necessary nor desirable

This thesis develops the work of Ferreira (2003) and Ward (2007), investigating the impact of plausibility (which created semantic anomalies in a sentence) and sentence arrangement (voice: which could be either active or passive) on participant’s tendency to Shallow Process.

There is disagreement with respect to implicit cognition in Shallow Processing and whether participants register semantic anomalies subconsciously rather than at a conscious level (Bohan & Sanford 2008 and Van Oostendorp & De Mul 1990). This study aimed to resolve these conflicting findings.

Our results demonstrate a significant main effect of both plausibility and voice in accuracy. Participants showed greater accuracy in plausible and active sentences than their implausible and passive counterparts. We also found participants Shallow Processed in these instances pragmatically normalising their interpretations. Implausible and passive sentences took longer to comprehend than their plausible and active counterparts. A significant interaction between plausibility and verbal IQ was also found in decision times. When reading sentences participants slowed down reading implausible sentences when compared to plausible sentences. That participants showed no difference in reading times between plausible and implausible sentences when questions were answered incorrectly, suggests there was no implicit cognition.

Future studies might use our methodology to sample larger populations and clarify out findings.