The Weekly Newsletter of MIT Linguistics

Syntax Square 11/14 - Isa Bayirli

Speaker: Isa Bayirli
Title: On gender and concord
Time/date: Nov. 14, 2016, 1:00-2:00pm
Location: 32-D461

Suppose, for the sake of argument, that the Feature Assignment Rule (Pesetsky, 2013: 8) is defined for every grammar and that it is the only mechanism responsible for the concord phenomenon (i.e. the feature co-variance between a noun and the modifying adjectives)

(1) Feature Assignment (FA), version 1 of 6:
a.Copying: when α merges with β, forming [α β] with the label α, the grammatical features of α are immediately copied on β
b.Realization:…and are realized as morphology on all lexical head items dominated by β

Suppose, moreover, that in a language where the overt effects of the FA Rule is invisible (i.e. the non-concord languages), the NP (with all the AdjPs inside it) is a protected domain of some kind.

(2) The Absence of Concord: A language lacks concord on the adjectives only if the NP in this language is a protected domain

To support these suppositions, one would need to provide evidence of (at least) three types:

Type 1: Evidence indicating that concord is universally calculated over syntactic representations as implied by the FA Rule (with no reference to feature values)
Type 2: Evidence indicating that whenever we do not observe concord in a language, the NP in this language is, indeed, a protected domain
Type 3: Evidence indicating that whenever NP cannot be protected from the application of the FA Rule in a language we do observe concord in this language

Previously at Syntax Square, I presented evidence of Type 1 (The Concord Hierarchy) and of Type 2 (The Concord-Suspension Complementarity), both of which I will briefly summarize.

In this talk, I will present some evidence of Type 3. NPs can be protected from the features coming from the functional projections but not from the features that start on the noun head. That is, if a feature f starts out on the noun head, then the universality of the FA-rule implies that – in the usual case – it will show up on all the adjectives adjoined to the NP. Assuming the gender feature, in those languages where it is idiosyncratic, starts out on the noun head, we get:

(3) Idiosyncratic Gender Generalization (IGG): A language with an idiosyncratic gender system is a language with gender concord

I first discuss some evidence for the validity of IGG. I then report a typological survey based on World Atlas of Language Structures (WALS) in support of the IGG generalization. I finally discuss some problematic cases (Gur languages of Niger-Kongo Family). I argue that the solution I sketch for these problematic cases are motivated on independent grounds.