The Algonquian person prefix is an agreement affix, not a pronominal clitic

Halle & Marantz (1993) propose that the person prefix that appears on Independent Order verbs in Potawatomi is a pronominal clitic and not an agreement affix, and this analysis has been followed for a variety of Algonquian languages by a large number of researchers. With data primarily from Passamaquoddy-Maliseet (Eastern Algonquian), I show that the prefix is not a pronominal clitic, it is an agreement affix, according to all diagnostics for distinguishing them. This conclusion is problematic because the prefix can be separated from the verb stem and the rest of the inflection, which is suffixal. I propose an analysis according to which all the inflectional suffixes are heads low in the clause, through which the main verb moves. The prefix, in contrast, is an Agr(eement) head adjoined to a head higher in the clause. This head attracts the highest verbal element, which will be the highest preverb if there is one, or the main verb if not. Preverbs are syntactic heads above the main verb which never form a complex head with it. This is amply justified by their separability, their prosody (they are often separate prosodic words), and coordination facts. Word order facts also indicate that the verb does not move very high in Algonquian languages, contrary to many analyses since Halle & Marantz (1993).