The role of quantifier restriction in illusory NPI licensing

Thesis

Negative polarity items (NPIs) like ever have been of particular interest and much debate in psycholinguistics because they rely on semantic and pragmatic processes which are not involved in the processing of other dependencies. There are currently two competing accounts that explain how NPIs are processed. One theory considers that NPIs are licensed through memory retrieval mechanisms (Vasishth et al., 2008), a second theory considers that NPI licensing is primarily a matter of semantic and pragmatic properties of the sentence which contains it (Xiang, Dillon, & Philips, 2009; Xiang, Grove, & Giannakidou, 2013).

Given their specific semantic structure, universal quantifiers such as every are able to license NPIs within their restrictor and not within their scope. Therefore, they offer an interesting yet unexplored testing bed for both licensing hypotheses. We examined cases of illusory licensing to see whether clearly establishing a restrictor could reduce illusions in offline and online measures. We found that illusions arise with the universal quantifier, but that manipulating the restrictor can subvert illusions in offline ratings. We also examined whether licensed NPIs could reduce garden path effects by reducing structural ambiguity that produces them.

We found that licensed NPIs can help the parser chose the reduced relative clause interpretation at ambiguity regions and pre-empt the later appearance of garden path effects. We conclude that NPI licensing is driven primarily by semantic/pragmatic properties but memory retrieval plays a role. The universal quantifier provides the parser with information about an upcoming downward-entailing context and that the NPI itself provides information to the parser that the context where it is located is the appropriate one, regardless of whether it is located inside the appropriate one, which explains the rise of illusions. At this point the parser checks the availability of a context-providing element possibly through a memory retrieval mechanism.

Authors: Hildebrandt Belmont, LA

• DOI: 10.5287/ora-w4ddae7nj
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: NPI; semantics; sentence processing; linguistics; downward entailing; quantifier restrictor; syntax; psycholinguistics; ambiguity resolution; pragmatics; negative polarity items; illusory licensing; self paced reading; universal quantifier; garden path; speeded acceptability ratings; quantifiers; quantifier scope
• Subjects: Linguistics; Negative Polarity Items; Garden Path; Linguistics, Experimental; Pragmatics; Sentence Processing; Quantifiers; Syntax; Semantics; Psycholinguistics