Configurational syntax in the Sumerian clause: Evidence from cross-referencing
The Sumerian clause is characterized by a highly synthetic verbal complex that is preceded by free nominals, whose relative order is governed largely by information structure. Despite the over century-long study of Sumerian, though, the details of its clausal syntax are poorly understood. The large array of verbal properties that can be marked within the verbal complex, the tendency for pragmatically inferable nominals to be dropped, and the many different morphological analyses of the verbal complex in the literature have largely conspired against in-depth study of Sumerian syntax.
I argue that the behavior of two of the cross-referencing morphological slots in the verbal template—the final person-prefix (FPP) slot before the verb stem and the person suffix (PS) slot after the verb stem—points to the existence of an underlying configurational structure. In particular, I argue that the PS is not pronominal cross-referencing, as is commonly assumed, but rather true agreement, the result of a (Multiple) Agree relation (Chomsky 2001, Hiraiwa 2001, and others) with the structurally highest active nominal in the clause. In contrast, the FPP is a doubled pronominal clitic that moves into its position in the verbal complex post-syntactically.
In support of this, I provide an analysis of the ergative cross-referencing split between the ḫamṭu and marû aspects, adapted from previous analyses of other ergative languages. I then show that this analysis allows for a principled account of the third-person plural agent marking in ḫamṭu forms, which is distributed between an FPP /n-/ and a PS /-eš/.