In this section I argue for an at least partially syntactic approach to ni-phrase licensing. Section 2.1 looks at the scope properties of ni-phrases and concludes that no operator may scopally intervene between negation and ni-phrases. (8) Based on this, a theoretical argument is made in section 2.2 for treating ni-phrases as universal quantifiers. Section 2.3 offers additional empirical motivation for treating ni-phrases as universal quantifiers and shows counterarguments from the literature to be inconclusive. Section 2.4 finally examines the locality conditions on ni-phrase licensing. I conclude that ni-phrases cannot escape their containing TP. This turns out to be an additional argument for treating ni-phrases as universal quantifiers.
2.1. The Scope of ni-Phrases
In this section I argue that the scope of ni-phrases tracks the scope of negation (see Babyonyshev, Ganger, Pesetsky and Wexler 2001 for a closely related point). What I mean by this is that no scope-bearing element can, in the general case, intervene between negation and ni-phrases; negation and ni-phrases scope together relative to other operators. Suppose there is a scope-bearing element "OP" in a clause. If negation takes scope above OP ([logical not] > OP), then ni-phrases also take scope above OP (ni-phrase > OP), never below it (*[logical not]- > OP > ni-phrase). On the other hand, if OP takes scope above negation (OP > [logical not]), then it also takes scope above ni-phrases (OP > ni-phrase), never below them (*ni-phrase > OP > [logical not]). In this sense, the scope of ni-phrases "tracks" the scope of negation.
Notice that there are two logically independent claims here. The first claim (*[logical not] > OP > ni-phrase) is illustrated in example (10), the second claim (*ni-phrase > OP > [logical not]) is illustrated by example (12) below. The logic of the argumentation is exactly parallel, but the points the examples make are independent. Consider example (10) first.
(10) Ja nicego ne zapretil emu chat'.
I NI-what NEG forbade him to-read.
a. [check] 'There is nothing that I forbade him to read.'
= It is not the case that there is an x such that I
forbade him to read x.
= For every x, it is not the case that I forbade
him to read x.
b. * 'I didn't forbid his reading (something).'
= I did not forbid it to be the case that (there
is an x such that) he reads (x).
There are three logical operators of interest in the example: negation, the intensional verb zapretit' 'to forbid', and the operator associated with the ni-phrase. Although eventually I will conclude that ni-phrases are universal quantifiers, this conclusion has not been established yet. I therefore consider two different possibilities, namely that ni-phrases might be universals with scope over negation and that they might be existentials with scope below negation. The scopal relations between negation and zapretit' are fixed. This gives rise to two distinct possible interpretations, as shown in (11).
(11) a. [logical not] > [there exist] > zapretit' [check] [for all] > [logical not] > zapretit'
b. [logical not] > zapretit'> [there exist]
Example (11a) corresponds to (10a) and (11b) corresponds to (10b). The paraphrases after the equal signs are intended to correspond to the existential and the universal interpretation of the ni-phrase respectively. Semantically they are equivalent. The example only has the reading in (10a), where no operator intervenes between negation and the ni-phrase. This reading is stronger than (10b), since (10b) is compatible with an injunction against specific books, while (10a) is not. Example (10) is judged false if there are specific books that I forbade. The representation [logical not] > OP > ni-phrase must then be ruled out. …