Referential and lexical factors in alignment variation of trivalent verbs

Argument marking with trivalent verbs exhibits a much larger variation than argument marking with bivalent verbs. In many cases, this variation-stemming both from referential and lexical factors-presents a problem when attempting crosslinguistic comparison of alignment patterns of trivalent verbs. Often, this problem results in picking just one of a number of patterns as representative for comparative purposes and thus ignoring the rest of the variation. This paper addresses these general challenges by discussing a case study of trivalent verbs in Yakima Sahaptin, a language with a large amount of alignment variation in indexing and flagging. In doing so, the paper elaborates the recently developed method for alignment typology called exhaustive alignment, adjusting the method to meet the challenges of constructions with trivalent verbs and pointing out its limitations.