Thom van Gessel, Alexandre Cremers & Floris Roelofsen (2018)
“Polarity sensitivity of question embedding: experimental evidence”

Main information:
Full paper in pdf
BibTeX reference
Experiment 1 Experiment 2 Experiment 3
Result file in csv format Result file in csv format Result file in csv format
R script for the main analyses R script for the main analyses R script for the main analyses

Abstract:

Attitude predicates can be classified by the kinds of complements they can embed: declaratives, interrogatives or both. However, several authors have claimed that predicates like be certain can only embed interrogatives in specific environments. According to Mayr, these are exactly the environments that license negative polarity items (NPIs). In his analysis, both NPIs and embedded interrogatives are licensed by the same semantic strengthening procedure. If this is right, one would expect a correlation between acceptability of be certain whether and NPIs. The analysis also predicts a contrast between antecedents vs. consequents of conditionals and restrictors vs. scopes of universal quantifiers. This paper tests these predictions experimentally through an acceptability judgment task. We find that judgments for be certain whether do not clearly correlate with judgments on NPIs, which suggests that be certain whether and NPIs are in fact licensed by different mechanisms.

R source file for mixed models