This dissertation explores the relation between the notion of subjecthood and syntactic ergativity through the lens of West Circassian (or Adyghe; of the Northwest Caucasian family), a polysynthetic morphologically ergative language. West Circassian displays a number of unusual syntactic ergativity effects, i.e. effects which indicate that the absolutive argument occupies the structurally prominent position which is generally believed to be associated with subjecthood. Through the close examination of two such effects – reciprocal binding patterns and constraints on parasitic gap licensing – I arrive at the conclusion that the absolutive argument in West Circassian may be generated in a variety of positions within the verbal theta-domain, but then uniformly undergoes A-movement to the highest argument position in the clause regardless of its theta-role. The derived status of the high absolutive explains why a number of subjecthood diagnostics, such as reflexive binding and obligatory control configurations, single out the ergative agent as the subject: this is because the syntactic properties of these constructions rely on the voice projection, which is merged directly above the verbal theta-domain, i.e. vP. Given the low position of Voice0, diagnostics which involve the use of this head single out the highest argument in vP as the subject, rather than the highest argument in the full clause. The proposed analysis provides support for the idea that subjecthood properties may be disbursed across several positions – in a syntactically ergative language like West Circassian, these positions are often occupied by two distinct nominals. In the absence of a single position or single nominal which could be labeled as a subject within a given clause structure, the term ‘subject’ becomes theoretically vacuous. The dissertation provides insight into the nature of syntactic ergativity, the structural status of subjecthood, and the syntax of anaphor binding, parasitic gap licensing and obligatory control constructions. Through the examination of argument asymmetries in a polysynthetic language, this project contributes to the line of research demonstrating that polysynthesis and non-configurationality cannot be defined macro-parametrically, but are rather a consequence of several micro-parameters at play, and that there is configurational structure at the core of such a language.