Scalar implicatures or focus: an experimental approach

Abstract

This dissertation is about one of the most robust and widespread types of pragmatic meaning in natural language: scalar implicatures (SIs). A scalar implicature is the inference that when a speaker uses a linguistic item that is on a scale with another item, the hearer concludes the sentence with the stronger scale member does not hold. For instance the inferences from 'It was okay' to 'It was not excellent' and from 'Some students passed' to 'Not all students passed' are scalar implicatures. It is widely acknowledged that these inferences are context-dependent, yet little is known about the properties of the context that determine their presence or absence. This thesis aims to contribute to the filling of this gap by considering the relation between SIs and the contextual property of information focus. Additionally, it addresses the psychological reality of the view that the exclusive reading of 'or', on which 'A or B' means 'A or B but not both', comes about by an SI. The thesis addresses these questions through a set of 11 experiments, both off-line (questionnaires) and on-line (processing). Next to the relation between SIs and focus, these experiments consider relevance of the stronger scalar alternative, the processing cost of SI-calculation, the assumption of speaker expertise and the relation between SIs and exhaustivity. The results support the prediction that information focus affects SI-calculation. However, the array of data taken together raise doubts about the view that the exclusive reading of 'or' is the result of a scalar implicature. The author discusses and tests theoretical insights from truth-conditional semantics, logic, pragmatics and reasoning as well as insights and results from (experimental) psycholinguistics, so this dissertation should be of interest to scholars in any of these domains.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 75 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.