The distribution of a special agreement pattern known as the “inverse” varies across the Algonquian languages. This article shows that, under an interaction-and-satisfaction model of ϕ-agreement (Deal 2015, 2021), twelve distinct distributions of inverse agreement can be derived simply by varying the interaction and satisfaction conditions of the probe on Infl. This syntax-based approach to agreement variation has strong predictive power: it not only accounts for the twelve attested patterns but also rules out the unattested patterns, a result that a less restrictive postsyntactic approach cannot deliver. The Algonquian data thus provide support for a syntactic model of agreement patterns in general, and for the interaction-and-satisfaction model in particular.

Using data from agreement in three Algonquian languages (Ojibwe, Cheyenne, and Plains Cree), this squib shows that effects typically attributed to Chomsky’s ( 2000 , 2001 ) Activity Condition (AC) can vary not only across languages, as in Baker’s (2008b) macroparametric proposal, but within a language as well. AC effects are thus another instance in which an apparent macroparameter turns out, on closer inspection, to be a microparameter instead, as in prominent cases such as the pro-drop parameter and the polysynthesis parameter ( Kayne 2005 , Baker 2008a ).