|dc.description.abstract||This thesis investigates the distribution of long-distance anaphors (LDAs) in Latin and proposes an analysis which takes into account both syntactic and pragmatic factors. It is generally assumed in the grammatical literature that complement clauses of reported speech/thought constitute a relevant domain for long-distance anaphora in Latin. This claim has been challenged in Benedicto (1991), as there are examples of LDAs in non-reported environments. I show that reported speech/thought is relevant for long-distance anaphora in Latin, and that LDAs outside of reported environments need a separate treatment. The LDAs in reported complements, which I have called the normal Latin LDAs, obligatorily take as an antecedent the noun referring to the person whose thought the clause expresses. This happens regardless of the syntactic position of this noun. The group of LDAs which occur outside of reported contexts, the special Latin LDAs in my terminology, have their own domain restrictions and binding properties.
Giorgi (2006) and (2007) propose a syntactic account of long-distance anaphora which links long-distance binding to the temporal anchoring of complement clauses expressing propositional attitudes. The predictions this theory makes are in part borne out in Latin: The distribution of the normal LDAs seem to be sensitive to the syntactic distinction between complements and adjuncts. It might also be correct that the relevant complement clauses are those which express propositional attitudes. It is probably not correct, however, that long-distance anaphora is related to temporal anchoring in Latin. Moreover, the special LDAs are unexpected in this approach. A discourse approach to long-distance anaphora based on Sells (1987) can account for the attested patterns in a descriptively better way. While Sells theory makes empirically good predictions, it needs to be adapted in some way to a modular view of language. Syntax should play a part in such an adaption, as the complement/adjunct distinction is relevant to long-distance anaphora in Latin. I have therefore suggested an approach to long-distance anaphora in Latin which combines insights from both theories. In this approach, LDAs are indexical pronouns anchored to internally specified contextual coordinates. Certain verbs, notably those which take reported complements, specify a new set of contextual coordinates, referring to the speech or thought event. The normal LDAs are anchored to internally specified agent-coordinates in such complements. I also tentatively suggest that the space-coordinates can be internally specified in certain non-reported environments, and that the special LDAs refer to such coordinates.||en