The Time Course of Processing of Natural Language Quantification

Paterson, Kevin Brisbane (1996) The Time Course of Processing of Natural Language Quantification. PhD thesis, University of Glasgow.

Full text available as:
Download (10MB) | Preview


This thesis examines the processing of quantificational semantics during reading. It is motivated by some formal observations made by Barwise and Cooper (1981), and a psychological theory proposed by Moxey and Sanford (1987,1993a). Barwise and Cooper classify quantifiers in terms of the directional scalar inference they license. Quantifiers like a few and many are described as monotone-increasing which means that what is true about a subset is also true of the superset. For instance, if a few student passed the exam with ease, this entails that a few students passed the exam. Other quantifiers like few; and not many are monotone-decreasing, and license inferences in the opposite direction. If it is true that few students passed the exam, this entails [hat few students passed the exam with ease. Moxey and Sanford found that these categories of quantifier produce contrasting patterns of focus. They used an off-line production task to demonstrate the monotone-increasing quantifiers, like a few and many focus processing attention on that subset of the quantified NP which is true of the sentence predicate (called the 'refset'), and that subsequent pronouns are interpreted as referring to this set. This means that given a fragment like (1), the plural pronoun will be interpreted as referring to the set of students who passed the exam. (1) A few of the students passed the exam. They . .. In contrast, monotone-decreasing quantifiers like few and not many exhibit a more diffuse pattern of focus, and permit subsequent reference to either the refset, or the complement of this set (called the compset), which is false of the sentence predicate. This means that the plural pronoun in (2) can be interpreted as either referring to the set of student who passed the exam, or those who failed it. (2) A few of the students passed the exam. They ... However, the off-line nature of the Moxey and Sanford studies limit them as descriptions of reading processes, so this thesis reports a series of experimental investigations of pronominal reference during reading. The first two studies used materials like (3) in a self-paced reading experiment to demonstrate that reference is easier when the anaphor describes a property of the refset {their presence) following monotone- increasing quantification, but that reference to either a property of the refset (their presence) or compset (their absence) is possible following monotone-decreasing quantification, although there is a preference for compset reference. (3) [A few I Few] of the MPs attended the meeting. Their [presence | absence] helped the meeting run more smoothly. A second two experiments monitored subjects' eye movements as the read passages like (3) in order to determine the locus and time course of referential processes. In line with other studies (eg. Garrod, Freudenthal and Boyle, 1993), it was predicted that the anaphor would be immediately interpreted as anomalous when it describes the unfocused antecedent. However the studies failed to find any evidence of punctuate effects. A further eye movement study was conducted with a revised set of materials but still failed to find evidence of punctuate anomaly detection in the two quantificational conditions. It was concluded that pronominal reference to a quantified noun-phrase is not processed on-line, i.e. it is not processed as the anaphor is read. Chapter eight presents two experiments on the interpretation of the non-monotonic quantifier only a fezo. It was suggested that this has the simple function of marking a set relative to expectations, and that focus is pragmatically determined. Focus is maintained on the refset when the quantified sentence describes a situation which is consistent with expectations, but the compset is placed in focus when these expectations are violated. Experiment six uses a sentence-continuation task to demonstrate these preferences, and an interaction with sentence connectives. Experiment seven monitored subjects' eye movements are they read sentences which referred to either the refset or compset of a sentence quantified by only a few. There was no evidence that the processing of pronominal reference is contingent on the focusing properties of this quantifier. Chapters nine and ten make a digression to consider the interpretation of sentences with more than one quantifier. The resulting scope ambiguity has been the subject of considerable theoretical interest, but limited empirical research. The existing literature is reviewed in Chapter nine, and a preliminary off-line sentence-continuation study is reported in Chapter ten which examines the interaction of quantifier and pragmatic constraints on a doubly-quantified sentence. The experimental findings are summarised in Chapter eleven, where an effort is made to accommodate the quantifier focus and scope ambiguity strands of this thesis within a common representational framework.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Additional Information: Adviser: Linda Moxey
Keywords: Quantitative psychology, Linguistics
Date of Award: 1996
Depositing User: Enlighten Team
Unique ID: glathesis:1996-74904
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 27 Sep 2019 15:25
Last Modified: 27 Sep 2019 15:25

Actions (login required)

View Item View Item


Downloads per month over past year