Arguments and adjuncts in O’dam : language-specific realization of a cross-linguistic distinction

This dissertation examines the properties that distinguish argument and adjunct dependents in the O’dam language (Tepiman<Uto-Aztecan) of Durango, Mexico. Verbal dependents, which express the participants involved in the eventuality described by the verb, are divided into different grammatical relationships with that verb (e.g. subject, object, oblique, etc.). Such grammatical functions are commonly assumed to be grouped into two overarching functions: arguments, which express core participants of a predicate and are closely tied to the verb, and adjuncts, which express peripheral participants of a predicate and lack any special morphosyntactic status in regards to the particular verb. There has been a long been an attempt to identify a cross-linguistically valid set of grammatical properties that will cross-linguistically distinguish arguments from adjuncts. I show that O’dam adds a typologically new type of language that does not conform to the standard view of the argument/adjunct distinction. Head-marking underpredicts the number of arguments that ditransitives and denominal verbs have, while most other standard cross linguistically-applied tests for different grammatical function in a large part do not distinguish dependents at all. Instead, the evidence for a thematically-rooted distinction between arguments and adjuncts found in argumenthood tests that mostly constitute wholly language-internal properties. I propose two new language-specific tests of argumenthood specific to O’dam: preverbal (discontinuous) quantification and applicativization. In addition to subjects and objects, preverbal quantification distinguishes different types of benefactive objects, and distinguishes recipients from recipient benefactives. The output of applicativization is hierarchically determined by the valency and argument structure of the verb, providing another probe into underlying argument structure. However, while there is overlap among the various argumenthood tests, the subsets of dependents each test identifies as an argument are not co-extensive. Valency effects on applicativization do not match such effects on head-marking, nor do either line up with preverbal quantification. Rather than finding a uniform behavior for arguments, I ultimately show that adjuncts are the only grammatical function with uniform syntactic behavior, purely because they are the only set of dependents that consistently fails every test. Notable among these are instruments and locatives, which behave as adjuncts regardless of their semantic relation to a predicate. Additionally, I show that O’dam realizes many of the properties predicted to hold for a Pronominal Argument Language (Jelinek 1984), suggesting that argument saturation is done within the verb. However, the interpretation of overt and covert nominals suggests that such argument saturation is not done through an equivalent to a lexical pronoun. This investigation of the argument/adjunct distinction in O’dam adds a more comprehensive empirical account of O’dam verbal syntax, and suggests that the cross-linguistically useful notion of distinctions between grammatical function can sometimes play out through almost entirely language-specific properties.