This talk presents puzzles concerning the representation of features in the agreement system of the Eastern Algonquian language, Mi’gmaq. A growing body of research converges on the idea that phi-agreement should be separated into distinct person (π), number (#), and sometimes gender (Γ) probes (e.g. Anagnostopoulou 2003; Béjar & Rezac 2003; Laka 1993; Shlonsky 1989). This separation has led to a wide range of empirical converge, in everything from Mayan Agent Focus constructions to Basque unergatives.

In light of the successes of the above accounts, the Mi'gmaq facts presented here initially come as a surprise. Mi’gmaq’s phi-indexing system suggests that probes for person and number cannot be completely distinct, as 1st and 2nd person plural arguments are privileged over all other arguments in competition for a specific verb stem slot. Evidence from Mi’gmaq’s dual/plural contrast nonetheless supports the proposal that π and # should be separated into distinct heads, which fuse together in the transitive animate paradigm. The two heads in the fused probe must probe separately, and the results of the search are conjoined in order to determine the agreement marker.

In addition to providing an account of the complex Mi’gmaq transitive animate agreement paradigm––which differs significantly from those of more commonly studied Algonquian languages––this talk provides additional evidence in support of proposals which argues that while traditional prominence hierarchies (e.g 1>>2>>3) may provide useful descriptive tools, they do not have a formal place in the grammar (see Bruening 2005; Brown et al. 2003; Harbour 2006; Coon & Preminger 2012; Oxford 2014, Preminger 2014).