Volume 18, Issue 2

In this article two hypotheses about the relationship of verbal and adnominal agreement are proposed: first, it is claimed that languages tend to have adnominal agreement only if they also have verbal agreement, but not vice versa. Second, it is argued that languages with adnominal agreement are areally restricted, while languages with verbal agreement are not. A sample of languages was selected to test these hypotheses, and both are confirmed by the data. The article closes with an admittedly speculative explanation of the hypotheses involving patterns of grammaticalization of different types of agreement markers.

Cognate Head-Dependent Constructions (CHDCs) are employed across numerous genera in Africa to signpost alternations in the aspectual characteristics of a predicate or the information focus of a clause. The co-occurrence of a finite lexical verb (the cognate head) and an etymologically related (deverbal) noun or non-finite verb form (the cognate dependent) in such structures is interpreted with reference to the scalar semantics of events and properties. Within this areal typology, CHDCs are employed to indicate either (i) a high point relative to a norm on a semantic scale or (ii) a conventionally low-ranked possibility, in order to implicitly contrast possible alternatives.

This article describes the reality status system of Nanti (Arawak) and argues that it constitutes an instance of a canonical reality status system. The relevance of such a system is examined in the light of literature that casts doubt on the typological validity of reality status as crosslinguistic grammatical category. It is shown that reality status is an obligatory inflectional category in Nanti, and that the distribution of realis and irrealis marking across Nanti construction types hews closely to expectations based on notional understandings of “realis” and “irrealis” categories grounded in a contrast between “realized” and “un-realized” situations. It is also shown that the Nanti reality status system does not exhibit evidence of being based, either synchronically or diachronically, on semantically narrower notions that could account for the distribution of reality status marking in the language, without recourse to the more generalized notions of realized and unrealized events. It is suggested that the Nanti reality status system might serve as a suitable canonical system around which a canonical typology of reality status might be built.

Languages vary as to whether they overtly mark the referential status of noun phrases as well as the manner in which they do so. This article discusses a hitherto undescribed means of marking the (un)identifiability of referents found in Ėven, a North Tungusic language spoken in Siberia. Narrative data from two geographically and linguistically divergent dialects show that in this language (in)definiteness is expressed with sets of diminutive and augmentative suffixes which are in complementary distribution. One member of each set occurs with identifiable referents and one member occurs with unidentifiable referents; the dialects differ with respect to possessive-marked nouns.