On the Notion of Subject in Ergative Languages

In traditional grammar, syntactic analysis is almost exclusively based on categories revealed directly in surface structures. In particular, morphologically unified categories of constituents are often taken to be the only ones that could possibly have any importance for the description of sentence structures. If a notion like “subject of” is to have any syntactic importance, then, it must be possible to associate it with a category of the language’s morphology. In most of the familiar languages of Europe which form the basis for this tradition, it is fairly easy to provide some set of morphological criteria which will (perhaps with a little fudging, such as the introduction of “notional” categories) pick out just the class of subjects which seems syntactically significant. The question of whether these properties actually have anything essential to do with “subjectness” or not, however, is seldom raised: having served their purpose, they are assumed ipso facto to be significant.