| This work proposes a formal account of the syntactic devices languages use to express the meronymic (part-whole) relation. The languages treated--Korean, Lakota, Amharic--have a clause type that allows for meronymic pairs only: a semantic constraint excludes pairs of nouns expressing kinship and ownership, while the two types of nouns underlying the meronymic relation, i.e. body parts and spatial nouns, are allowed. This clause type is semantically constrained in another respect: the event or state expressed by the predicate must be one involving not the whole, but just a part of the referent that the noun denotes. Thus predicates like 'hit' turn out to be compatible with the meronymic construction, while predicates like 'carry' are not. Alongside these semantic constraints, syntactic tests reveal an asymmetry between the noun denoting the whole (holonym) and that denoting the part (meronym). Only the former can be the target of relativization, topicalization, and passivization.;The structure envisaged here is a union--Meronymic Union (MU)--that automatically entails both the semantic and syntactic limitations characterizing the meronymic construction. The syntactic inertness of the meronym is accounted for by analyzing it as a noun predicate, while the inaccessibility of non-meronymic pairs and the exclusion of such predicates as 'carry' follow from the presence of an scENLARGED ARGUMENT, namely of a noun apparently denoting the whole, but whose actual referent is only a part of the whole. This result is attained with a new notation for heads of arcs which makes the referent of a holonym a sub-area of what the same noun would normally denote.;The union produced by the serialization of the noun predicate (meronym) with the ensuing predicate (verbal or adjectival) gives rise to a 2-hood constraint: the enlarged argument can only be the object of a transitive or unaccusative predicate. The incompatibility of unergatives with MU is expected and derives from the Union Law and the Compactness Principle.;Chapters 2-4 treat Korean, Lakota, and Amharic respectively. Chapter 5 shows the reasons to prefer MU over possessor ascension, while the appendix discusses other languages with constructions involving meronyms and holonyms. Certain clause types of Mandarin Chinese appear to be amenable to an MU analysis, whereas others present in some Australian languages don't. |
