pp. 39-56
Alain Kihm
Abstract
Construct State Nominals (henceforth
CSN) in the Semitic languages (Arabic and Hebrew essentially) have been the
object of renewed interest in the past few years, because of the crucial
questions they raise in relation with the structure of genitive
constructions in particular and of the noun phrase more generally. In this
paper, I will explore one particular issue about them which has, I believe,
far-reaching consequences.
First, I will briefly review two
properties which are widely held to be defining properties of CSN, namely
head-complement adjacency and definiteness spread, showing that they
conspire to yield analyses that view CSN as resulting from some kind of
incorporation. Then, I will introduce data from Classical Arabic which,
although not new in themselves, have never been taken into consideration in
previous generative studies of CSN, at least to the best of my knowledge.
These data are patent counterexamples to the adjacency property and, as such,
are not accounted for by existing analyses. I will also argue that
definiteness spread is not a mechanism we want to allow, given more basic
assumptions. Given this, I will then present an alternative analysis of CSN
formation that (i) accounts for the non-adjacency facts, and (ii) allows one
to derive all properties of CSN as distinct from prepositional genitive
constructions, as well as to systematically relate both types of
construction. Finally, I will make a tentative proposal in order to explain
what I call the 'referential uniformity' of CSN, an apparent effect of which
is captured by definiteness spread, wrongly so in my opinion.