Case marking and alignment * Balthasar Bickel and Johanna Nichols 1. Introduction Alignment is standardly illustrated with ergative vs. accusative coding of subjects as in Basque (1) and Russian (2): 1 (1) a. gizona etorr-i da man.NOM arrive-PERF.PTCP AUX.3sS ‘The man has arrived.’ b. gizona-k mutila ikus-i du man-ERG boy.NOM see-PERF.PTCP AUX.3sA3sO ‘The man has seen the boy.’ (2) a. muzhchina prishel man.NOM came.PST.MASC 'The man arrived' b. muzhchina uvidel devushku man.NOM see.PST.MASC girl.ACC 'The man saw the girl' In (1) the syntactic functions of S (gizona ‘man’ in 1a) and O (mutila ‘boy’ in 1b) are identically coded, while in (2) it is S (muzhchina ‘man’ in 2a) and A (muzhchina in 2b) that are identically coded. That is, the two languages differ in whether S is aligned with A or O. Henceforth we refer to syntactic functions such as A, O, and S as argument roles. Argument structure has to do with the number of arguments and their roles; valence has to do with the morphosyntactic treatment of these roles, including their marking with particular cases. Alignment is the identical vs. distinct coding or treatment or behavior of argument roles that are different at some other level or in some other part of the grammar. Put differently, alignment is neutralization of valence-specific argument roles in particular morphological or syntactic contexts. Alignment, from our perspective, holds between sets (usually pairs, sometimes triads) of argument roles that * Research for this chapter was facilitated by DFG grant No. BI 799-3/1 (Bickel) and NSF grant no. NSF 96- 16448 (Nichols). 1 Throughout this chapter we use nominative to refer to the morphological case that is citation form and marks S, regardless of whether S is aligned with A or O (though many linguists use nominative only for S=A and absolutive for S=O). As the rest of the chapter shows, there are so many cross-cutting alignment patterns that a terminology treating morphological cases as nodes in an alignment space and separately labeling every node would be cumbersome, as well as confusing morphological case paradigms with alignment patterns.