The grammaticalization of grammatical relations: A typological and historical study involving Kashaya Pomo, Old English, and Modern English

This dissertation presents a typological and historical investigation of grammatical relations (GRs), with data drawn from Kashaya Pomo and Old and Modern English. I assume a theory-neutral conception of GRs as empirically instantiated entities irreducible simply to semantic or discourse-pragmatic correlates. I argue that GRs are language-specific and derived constructs in several respects. Languages differ in the degree to which semantics and discourse-pragmatics are conflated in their syntax and morphology; this in turn relates to cross-linguistic differences in the degree to which languages map arguments onto a single neutralizing pivot (Dixon 1979) in order for that argument to participate in the sorts of construction generally taken to serve as 'tests' of subjecthood (Keenan 1977). GRs are also language-specific in that their semantic and discourse-pragmatic correlates may vary in a manner which poses severe problems for any attempt to identify a single universal subject (or object) prototype.; In Chapters 1 and 2, I review the status of GRs in several theories of grammar. Chapter 3 contains a critical of Role and Reference Grammar, a model which avoids the pitfalls of the other models discussed here, and which I argue to offer a superior array of tools for the analysis of data otherwise interpreted in terms of imprecise notions of subject and object.; Chapter 4 presents an analysis of the active-stative language Kashaya Pomo. Kashaya not only fails to conflate intransitive arguments with either the agent-centered or the patient-centered argument of transitive verbs, but also extends its morphological distinctions into its syntax, resulting in a paucity of behavioral evidence for a 'subject' category as distinct from a semantic agent-based category ('Actor'). The limited evidence I do find for discourse-motivated neutralization of case-sensitive semantics points to a category bearing little resemblance to an English-like subject.; In Chapters 5 through 10, I offer an analysis of Old English as a transparent language showing little evidence for GRs. In its case morphology Old English fails to neutralize as extensive a range of verbal semantics in a single transitive pattern. It also resists operations which manipulate morphological case in response to discourse considerations, which in turn coheres with a severely limited role of passive in mapping non-Actor arguments onto pivot function.