Towards a Probabilistic Grammar of Collective Noun Agreement in British English

Abstract

This thesis examines collective noun agreement, using the notion of probabilistic grammar. Most accounts on the phenomenon appears not to have been fully explained the factors that determine collective noun agreement in English suggesting the existence of multiple factors that inform collective noun agreement relations. The most established explanation (that plural agreement is the preferred agreement in cases that the individuals in the collection are the focus and singular when the collection is the referent) is taken as circular reasoning and as the gap. And the study departs from this line of reasoning by examining other nuance factors that impact collective noun agreement. The study used the corpus linguistic methodology to study 6,467 clauses of singular and plural collective noun agreement forms from the British National Corpus. The data has revealed accessibility, humanness, definiteness, distance, part-whole and number features of predicates as variously informing agreement choices. These factors translate into two-level predictors which inform singular and plural agreement at relative significance and confidence levels. The results show that there is a higher tendency for singular agreement choice when the CNs are conceived as ‘whole’. Plural agreement is significantly determined by human of humanness and part of part-whole constraints. The observation that the predictors were sensitive to specific CNs suggests that CNs differ somewhat from each other. The results also show that it is easier to predict plural agreement than singular agreement and plural agreement choices are often primed. Further studies are encouraged to examine the impact of the constraint herein identified as determining collective noun agreement in other varieties of English.