This study presents the first investigation of the development of possessive constructions in Northern East Cree, a polysynthetic language indigenous to Canada. It examines transcripts from naturalistic recording sessions involving one adult and one child, from age 2;01.12 to 3;08.24. Findings reveal that, despite the frequency of possessive inflection in child-directed speech, the child overwhelmingly produces a possessive construction that circumvents this morphology. This construction, named here the equational possessive strategy (EPS), is largely undescribed in existing literature but is the primary mechanism for the child to express possession. These findings have potential implications for the cross-linguistic acquisition of possessive morphology and the connections between child-directed speech and child language production.

Blackfoot verbs are marked with nominal agreement morphology in relative clauses, in place of verbal inflection. These relative clauses have previously been analyzed as nominalizations. The present study shows that a nominalization analysis makes incorrect predictions about the morphological composition of relative clauses, as well as the availability of non-agentive and possessive constructions, and adjectival modification. This study demonstrates that relative clauses can relativize subjects, direct objects, and indirect objects. Based on observations about obviation, recursion, long-distance extraction, inflection, and word order, I propose that relative clauses are participles: clausal entities with a nominal superstructure. This accounts for their mixed clausal and nominal properties, and provides an analysis of Blackfoot relative clauses similar to those proposed for related Algonquian languages.

This article discusses the morphological and syntactic structure of relative clauses in Ojibwe (Algonquian), in particular their status as wh-constructions. Relatives in this language are full clauses that bear special morphology, show evidence of A′-movement of a wh-operator, and consequently exhibit wh-agreement also found in interrogatives and certain types of focus constructions. Ojibwe shows the possibility of wh-agreement realized on T (in addition to C and v for other languages), as it appears on tense prefixes. We account for the realization of wh-agreement on T in Ojibwe via the mechanism of feature inheritance. We propose that while declarative matrix clauses are canonical in that C introduces φ-features in Ojibwe, the role of C in embedded or wh-contexts is to introduce δ-features (discourse features), such as [uwh], rather than φ-features. These δ-features can be introduced by C, but are transferred down to T where they spell out as wh-agreement.

Derived clusters in Northern East (NE) Cree are traditionally assumed to result from the deletion of an intervening vowel. I propose, rather, that they arise from vowel devocalization, a process involving gradient vowel shortening or weakening. First, I demonstrate the compelling similarity between the process leading to derived clusters in NE Cree and vowel devocalization in non-Algonquian languages. Second, I show that consonants in NE Cree derived clusters display phonetic lengthening, one of the phonetic correlates accompanying vowel devocalization. I suggest that the gestures of these lengthened consonants overlap the vowel to the point that it is not perceived