Language processing is normally rapid, incremental and driven by online prediction-making. However, the phenomenon of negation presents an interesting possible exception to this case. Although at least some utterances involving negation seem to induce effects on processing in which predictions cannot be made or their accuracy breaks down, evidence suggests that this is not the case when negations are presented in contexts providing adequate pragmatic support. This thesis presents a series of experiments using several methodologies (EEG, computer mouse-tracking, and eye-tracking) to test the idea that this effect of pragmatic felicity can be attributed to its association with predictability: that is, to investigate whether the predictability of later material in a sentence can influence the extent to which negation can be incorporated incrementally into the comprehender's interpretation of the partial sentence. This is achieved through the use of episodically-varying contexts, presented prior to the accompanying linguistic input, which manipulate the predictability of critical material in pragmatically licensed sentences. The findings lead to the overall conclusion that even pragmatically licensed negations can incur more processing costs and result in the generation of more inaccurate predictions than equivalent affirmatives. Furthermore, the most reliable strands of evidence (from a sentence completion mouse-tracking task) suggest that, in the type of paradigm in which prediction is manipulated using episodic contexts, reducing predictability detrimentally affects affirmatives to a greater extent than negations. This may indicate that accurate prediction-making is relatively difficult, even in the easiest cases, for negative sentences when (1) predictions must be formulated on the basis of episodic rather than long-term semantic associations and (2) the combination of sentence structure and context mean that there is a clash between the concepts activated by association with local components of the sentence and those relevant to its global interpretation.
|Date of Award||23 Jan 2019|
- The University of Bristol
|Supervisor||Nina Kazanina (Supervisor) & Chris Kent (Supervisor)|