9. PLAN DE GESTIÓN AMBIENTAL PARA EL PROCESO PRODUCTIVO DE LA
9.2 PLANEACIÓN ESTRATEGICA
9.2.2 PERFIL DE OPORTUNIDADES Y AMENAZAS (POAM)
French WH in situ is not discussed by either Szabolsci and Zwarts (1992-1993) or Honcoop (1998), so only receives a principled account in terms of scope in the present study. Their respective accounts do, nonetheless, account well for the freezing property of adjuncts (Szabolsci and Zwarts) and for the locality effects produced by the movement of bare operators (Honcoop). In the present section, I discuss why my analysis in terms of scope is superior to these alternatives.
Szabolsci and Zwarts (1992-1993) argue that weak-island effects arise because, in contrast to the trace left by arguments, the trace left by adjuncts does not range over discrete individuals. Rather, the trace left by an adjunct ranges over a partially ordered
domain. Certain Boolean operations, such as complement and meet (negation corresponds to taking Boolean complement', universal quantification corresponds to taking Boolean meet), cannot be performed on impoverished domains such as the ones corresponding to adjuncts. When an expression E scopes over some operator O, the operations that define O need to be performed in E’s denotation domain. For instance, in calculating the denotation of (103), I take the complement of the set of those whom you saw:
(103) Who didn’t you see?
This is possible because who ranges over individuals, and individuals form sets, on which complementation can be performed.
Szabolcsi and Zwarts argue that, by contrast, the denotation domains of weak- island-sensitive items (manners, methods, etc.) do not lend themselves to complementation and/or intersection. Thus, they cannot scope over negation or universal quantifiers, for example:
(104) a. *How didn’t you see the car?
b. How did everyone arrive? *WH>V
According to Szablocsi and Zwarts, they can, however, scope over existentials (whose definition is in terms of union) or intensional verbs (whose semantic contribution is not Boolean in nature):
(105) a. How did a boy behave?
b. How did you want me to behave?
In sum, they propose that weak-island violations are semantically incoherent, in much the same way as *six airs is, where a numeral is applied to a mass term.
One problem for Szabolcsi and Zwarts’s analysis is that in-situ questions with existentials or intensional verbs are not very good. The existential tm enfant and the verb chercher ‘seek/to look for’ blocks the licensing of French WHl in situ;
(106) ?* Un enfant a fait quoi? (French)
a child has done what ‘What did a child do?’
(107) ?* Tu cherches quoi? (French)
you seek what
‘What are you looking for?’
On their account, existential and modals are not expected to block the licensing of French WHl in situ, the reason being that the definition of existential is in terms of union and the semantic contribution of intensional verbs is not Boolean in nature. On my analysis, the near-ungrammaticality of (106) and (107) is accounted for as follows: the existential and the modal are of a higher order than the stranded indefinite, so the stranded indefinite cannot depend on them.
I now turn to Honcoop (1998). Honcoop reduces the explanation of weak-island effects to that of cross-sentential anaphora. He notes that the expressions that create
weak-island effects and those that block cross-sentential anaphora are the same. In the examples in (108), the intervener ‘freezes’ the dynamic potential of the indefinite;
( 108) Inaccessibility
a. John has a can. It; is too expensive.
b. * John doesn’t have a cari. Iti is too expensive. c. *No student has a can. Iti is too expensive.
d. *Exactly five students have a can. Iti is quite expensive. e. *Most students have a can. Iti is quite expensive. f. *Every student has a can. Iti is quite expensive.
(Honcoop 1998:17)
An indefinite contained inside the scope of any of the operators (negation, quantifiers, etc.) can no longer introduce a ‘discourse’ referent that can be picked up by a pronoun in subsequent sentences. Following standard terminology, Honcoop refers to this phenomenon as inaccessibility. His claim is summarised in (109):
( 109) Claim: Weak-island Inducers are Inaccessibility biducers
The class of expressions that induce WIs [i.e. weak islands] coincides with the class of expressions that create inaccessible domains for dynamic anaphora.
(110) The Intervention Generalization
* ... Q/[cp... [weak island Operator ... indefinite,... J ... ] ... (Honcoop 1998; 19)
Honcoop (1998) follows the framework of Dynamic Semantics (cf. Groenendijk and Stokhof 1991, Chierchia 1992, 1995, Dekker 1993) according to which indefinites introduce an existential quantifier. This differs from Discourse Representation Theory (cf. Kamp 1981) according to which indefinites never introduce an existential quantifier. Dynamic Semantics accounts for the well-known quantificational variability of indefinites as follows: when an indefinite behaves like a (restricted) bound variable rather than a (restricted) existential quantifier, the operation that applies is Existential Disclosure. So, for example, whereas the indefinite introduces an existential quantifier in (Ilia ), ED applies to the indefinite in (111b). In the latter case, usually binds the variable introduced by the indefinite expression:
(111) a. A man came in.
b. Usuallyi, a new car\ is expensive.
In the case of cross-sentential anaphora, the indefinite introduces an existential quantifier. The quantifier can thus bind discourse markers in subsequent discourse:
4.3
If the indefinite is inside an inaccessible domain created by some operator, the indefinite is ‘closed off or ‘deactivated’, i.e. it undergoes Existential Disclosure (ED). The indefinite is therefore unable to bind discourse markers in subsequent discourse.
Honcoop argues that in a split construction the stranded indefinite is related to the bare operator dynamically. It is thus associated in the same way as the adverb of quantification usually in (111b) above is related to the indefinite it binds. In both cases, ED is required. If a quantifier intervenes between the bare operator and the indefinite, then the indefinite is deactivated. Therefore, it can longer be dynamically bound by the bare operator.
Honcoop’s analysis is very appealing. However, there are a few problems with his account. First of all, it is not clear how his analysis can account for the weak-island sensitivity of how. This is because there seems to be no reason to assimilate it to split constructions under the ED approach (this argument is made by Szabolsci and Zwarts). In order for Honcoop’s analysis to work for how, one would need to postulate a split structure for derivations involving how. However, presumably an adjunct like how does not contain an indefinite expression which could undergo ED. So, it is difficult to imagine how adjuncts involve splitting.
Another problem for Honcoop’s approach is that, whereas focus is typically a weak-island inducer, it does not seem to block cross-sentential anaphora;
J-3'
(113) Only JOHN received a letten. Iti was fi’om his mum.
(114) John twice saw a filnii. Iti was the best film he’d ever seen.
Finally, as shown by Honcoop, adverbs of quantification create inaccessibility domains (the judgment concerns the possibility of a token-interpretation of the indefinite antecedent):
(115) *Last year, when it rained, John often bought an Italian newspaper^. Iti reminded him of his summer love.
(Honcoop 1998:17)
However, it was shown that adverbs of quantification do not create opaque domains in the case of French single-WH questions in situ or split combien-de constructions.
In sum, although there are overlaps between inaccessibility in cross-sentential anaphora and blocking in split constructions, the interveners in split constructions and the interveners in cross-sentential anaphora are not totally equivalent.
In the next section, I capitalise on the Skolem function analysis presented in the previous section and generalize the present theory to other constructions.
2.5.5 Stranded nominals as predicative indefinites (Generalized Semantic