2.2 Orientaciones de la Sexualidad.
2.2.1.1.3.3 TEORIA SOCIOLÓGICA.
4.11. Roles Parentales.
T his section pursues a syntactic analysis of pas that sheds light on the fact that this element is used in contexts of both sentential and constituent negation. Neg- ative pas is an adverb, that is, it serves to modify something. In fact, Pollock (1989), in the context of his analysis of pas in sentential negation, notices the distributional parallels between pas and adverbs. H e notes (1989: 370, 377) that there is “a significant correlation in French between the placement of negation and that of adverbs” in both tensed and infinitival clauses. Cardinaletti and Gua-
5. For contexts in which pas negates a sentence containing a transitive verb governing an indefinite direct object, see section 2.2.
6. In her discussion of bipartite sentential negation in Navajo, Speas (1991b: 394!95) suggests that the post-verbal da marker is the overt realisation of NegE while doo, which canonically appears immediately before the direct object, occupies either SpecNegP or an adverbial position.
sti (1992) also observe that pas may function as an adverbial projection. Further- more, as the examples in (3) illustrate, the adverbial/modifier function of pas is not restricted to sentential contexts:
(3) a. A: Ça va? How are you?
B: Pas mal. Not bad.
b. A: Qui est-ce qui veut un café? W ho’s for coffee?
B: Pas moi. Not me.
c. Pas vrai! Never!
d. Pas possible! Impossible! e. A: T’as du fric? Got any money?
B: Pas un sou! Not a penny.
Zanuttini (1996: 184) notes further that the distribution of the equivalent of pas
in other Romance varieties (e.g., Piedmontese nen and Milanese minga) also overlaps with some adverbs. She concludes that all these items are lexical adver- bial elements generated in an adjoined position lower than NegP. (See footnote 2.) (See Zanuttini 1997a for a more recent and articulated analysis.)
In the context of a sentence (negated with pas) that contains either an intran- sitive verb or a transitive verb governing a definite direct object, I suggest that5 negative pas is generated adjoined to the lexical projection it modifies, that is, VP, as in (4). This reflects the fundamental relationship between 6 pas and the predicate; pas is a negative predicate adverb. (Sportiche’s Adjunct Projection Principle and Chomsky’s general theory of adjunction, together, oblige “modifi- ers” to appear adjacent to their nonargument XP “modifiee” or the head of their “modifiee”.)
VP VP
(4) [ pas [ ]]
W hile this analysis of pas is conceptually attractive, it has one major problem that is avoided within Pollock’s model, namely the fairly convincing evidence discussed in sections 1.1.2, 1.1.6, and 1.2.1, that, superficially, pas occupies SpecNegP, that is, above the VP-adjoined position. I take this to suggest that pas
raises from its VP-adjoined base position to SpecNegP in order to convert con- stituent negation into sentential negation; by raising to SpecNegP, pas endows NegE with the feature [+N EG]. (See the discussion in the previous section.)
7. In some cases, corresponding tonic and atonic pronouns are homophonous. I assume that this is a matter of coincidence and does not detract from the conclusion that there are two independent sets of pronouns, with their own distinct properties.
2.1.1
Synchronic evidence: negative imperatives
In this section, I present synchronic data that, I argue, are incompatible with a syntactic analysis of pas such as the one proposed by Pollock (1989), in which the element is assumed to be generated in SpecNegP. It is argued that, in cross- linguistic terms, the morphologically truncated structures of so-called true im- peratives suggest that such paradigms project truncated syntactic structures in which no functional structure above and including NegP is projected. The fact that such imperatives in French are nevertheless compatible with pas but not ne
(as well as the fact that such true imperatives in numerous other languages are compatible with adverbial negative markers but not head negative markers) un- dermines the claim that pas is uniquely associated with SpecNegP in verbal con- texts.
The analysis hinges on the assumption that there are two kinds of (negative) imperative in French and is based on the data in the next section and on recent work by M aría-Luisa Rivero (1994) and Raffaella Zanuttini (1990, 1991, 1994a, 1996). Following these authors, I assume that the difference between the two kinds of imperative is position: one kind of imperative occupies one position, while the other kind occupies another. In the two subsequent sections, I discuss theoretical approaches to the distinction proposed by Rivero and Zanuttini, re- spectively. W hile I ultimately reject Rivero’s analysis, I show that Zanuttini’s offers interesting insights into the morphological and syntactic properties of im- peratives. Finally, I show how the syntax of negative imperatives suggests that
pas in French is not uniquely associated with SpecNegP.
2.1.1.1
The data
Negative imperatives in French appear with either tonic or atonic complement pronouns: tonic forms are post-verbal, as in (5); atonic ones are pre-verbal, as in (6). I assume that the nature of complement pronouns (tonic/post-verbal versus7 atonic/pre-verbal) is determined by the syntactic properties of the imperative verb, more specifically the position of the verb. This approach is supported by the fact that pre-verbal complement pronouns cannot co-occur with post-verbal ones. If the pre-verbal atonic pronouns are licensed by virtue of the imperative occupying one position, while the post-verbal tonic ones are licensed by virtue of the imperative occupying a different position, then I expect pre- and post-ver- bal pronouns to be mutually incompatible; the imperative cannot occupy both positions simultaneously. I conclude, then, that the choice of pronoun is deter- mined by the position of the verb.
Another feature associated with negative imperatives that seems to be deter- mined by the position of the verb (and therefore co-varies with pronoun posi-
tion) is the (un)availability of “negative” ne. French imperatives can always be negated by pas alone, but they cannot always be negated by bipartite ne . . . pas. Significantly, the (un)availability of bipartite ne . . . pas patterns identically with the choice of complement pronouns: imperatives that license post-verbal tonic pronouns are incompatible with ne, as in (5c), while no such incompatibility pre- vents imperatives that license pre-verbal atonic pronouns from co-occurring with
ne, as in (6c). Given that the pre-/post-verbal position of the pronouns is deemed to be determined by verb position, this suggests that the (un)availability of ne is also determined by verb position. (The patterns in (5) and (6) are based on M ul- ler 1991: 142):
(5) a. Regarde- moi. / Donne- le lui. watch-IM P me / give-IM P it to-him ‘Look at me.’ / ‘Give it to him.’
b. Regarde- moi pas. / Donne- le lui pas. watch-IM P me pas / give-IM P it to-him pas
‘Don’t look at me.’ / ‘Don’t give it to him.’
c. Ne regarde- moi pas. / Ne donne- le lui pas.
ne watch-IM P me pas / ne give-IM P it to-him pas
(= (5b))
(6) a. M e regarde. / Le lui donne. me watch-IM P / it to-him give-IM P
(= (5a))
b. M e regarde pas. / Le lui donne pas. me watch-IM Ppas / it to-him give-IM Ppas
(= (5b))
c. Ne me regarde pas. / Ne le lui donne pas.
ne me watch-IM Ppas / ne it to-him give-IM Ppas
(= (5c))
Observe that the pre-verbal pronouns are incompatible with positive impera- tives: compare the grammatical (5a) with the ungrammatical (6a). I assume that this restriction is purely syntactic: the position necessarily occupied by a positive imperative is not the one that licenses pre-verbal pronouns. The inability of the verb to occupy the necessary position to license pre-verbal pronouns could be attributed to economy if the derivation of (5a) is less costly than (6a), for exam- ple, if (6a) involves (gratuitously) generating more functional structure than (5a) and if Hyams (1986) is right that UG favors minimal derivations. (See section 2.1.1.3 for details.) In contrast, the examples in (6b) could be argued to be fine for the simple reason that, as examples of sentential negation, NegP is non- empty; generating the required structure is not therefore deemed to be gratuitous. Consider now negative imperatives. W here the verb is negated by pas alone (without ne), it can be accompanied by either tonic or atonic pronouns (but not both)see my earlier discussion): (5b) and (6b) are both grammatical. I assume
8. Where infinitives are used with imperative force, as in (i), we assume they share the syntax of infinitives used in other contexts. Note that ne is present and that pas is pre-verbal.
(i) Ne pas marcher sur la pelouse.
ne pas walk-INF on the lawn ‘Keep off the grass!’
that the same position is occupied by the verbs in (5a, b) and that this is reflected in the position of the pronouns. In (6b), I assume that the atonic pre-verbal pro- nouns are licensed because the verb occupies a different position from the verb in (5a, b). W here the imperative is negated by bipartite ne . . . pas, the accompa- nying pronouns must be atonic and pre-verbal: compare the ungrammatical (5c) with the grammatical (6c). I assume that the unavailability of the tonic post-ver- bal pronouns together with ne in (5c) can be explained in terms of incompatible requirements on the imperative: the verb needs to be in one position to license the post-verbal pronoun(s) and in another to licence ne. G iven that the two con- straints cannot be met simultaneously, the string is ungrammatical. In summary, then, the imperatives in (5) are compatible with neither ne nor pre-verbal com- plement pronouns, while the imperatives in (6) are compatible with both.
I assume that the absence of pre-verbal ne in (6b) is due to optional “ne- drop”, discussed in chapter 1, footnote 3. The acceptability of (6c) shows that ne
is possible in such a structure. In (5b), in contrast, I assume that the absence of
ne is the result of some deeper grammatical incompatibility; hence the ungram- maticality of (5c).
These and similar data from other Romance languages have been considered by Zanuttini (1990, 1991, 1994a, b, 1996, 1997a, b) and Rivero (1994). Both authors exploit the distinction drawn by Joseph and Philippaki-W arburton (1987) between “true” imperatives (e.g., (5)) and “surrogate” imperatives (e.g., (6)). Informally speaking, true imperatives represent a distinct verbal paradigm, while surrogate imperatives are verb forms taken from another morphological paradigm, for example, the subjunctive or the infinitive, used with imperative force. True imperatives have distinctive structural properties; surrogate impera-8 tives adopt the structural properties of their source morphological paradigms. In the next two sections, I consider first Rivero’s then Zanuttini’s analyses of the distinction between true and surrogate imperatives.