4.2 Soft presuppositions and intervention effects
126.96.36.199 The intervention effects by soft presuppositions: a first pass
The present analysis treats soft triggers on par with strong scalar terms, so we expect that they should give rise to intervention effects in the same way. As I will show below in detail, this prediction is indeed borne out. Consider the case of because clauses: because leads to the inferences that its propositional arguments are true and these inferences appears to project as presuppositions, as shown by (14a)-(14c) which all give rise to the inferences in (14d) and (14e).6
(14) a. John is going to NY because Mary is there.
b. Is John going to NY because Mary is there?
c. It’s not true that John is going to NY because Mary is there.
d. Mary is in NY.
6Some of the speakers I have consulted did not share the judgement that (14)[a] gives rise to the inferences in
(14)[b]-(14)[e]. This could be accounted for by analyzing because not a soft trigger but as a regular scalar term like every. In this case, the theory of intervention by Chierchia (to appear) would apply to it without modifications or extensions. I leave the exploration of this hypothesis for further research.
e. John is going to NY.
Furthermore these inferences appear to be soft presuppositions, as the ignorance diagnostic applied in (15a) and (15b) shows.
(15) a. I don’t know where Mary is now, but if John is going to NY because she is there,
they have an affair.
b. I don’t know whether John is going to NY, but if he is going, because Mary is
there, they have an affair.
Therefore, because appears to be a soft trigger and we expect that it should intervene in the licensing of NPIs, in the same way we saw that everyone intervenes. Indeed, it has been known since Linebarger 1987 that because clauses do intervene. Consider the contrast betwen (16a) and (16b).
(16) a. *Dogs don’t hear because they have any eyes. They hear because they have ears.
b. Dogs don’t hear because they have eyes. They hear because they have ears.
This is precisely what we expect in the present account: the configuration we have in (16a) is (17). Given minimality and the fact that because is an alternative bearer, it needs to enter into a checking relation withEXHbefore the NPI. It’s only upon computing the exhaustification of the becauseclause that the NPI’s alternatives can be exhaustified, but by this point the environment is no longer DE given the soft presupposition that arises from theEXHof because.
(17) EXH[not[ because [ ... NPI ]]]
In the following, I discuss the intervention of presuppositions and the previous proposal by Homer (2010) and then I come back in detail to how the proposal here can account for the intervention by soft presuppositions.
4.2.1 The Intervention effects of presuppositions and previous accounts
Homer (2010), building on Linebarger 1987 and Chierchia 2004, observes that beyond scalar implicatures also presuppositions give rise to intervention effects. He discusses precisely the case of because-clauses and also cognitive factives in French, which behave in the same way.7
(19) (Context: Peter broke your Chinese vase.)
You are mad at Peter, not because he broke anything, but because he won’t own up to it. Homer 2010:ex.43
(20) *Pierre Pierre n’ NEG a has pas NEG d´ecouvert found-out que that Marie Marie a has ´ecrit written
quoi que ce soit anything `a to sa her m`ere. mother Homer 2010:ex.26
‘Pierre hasn’t found out that Marie has written anything to her mother.’
In both cases, he argues, that the intervention effects should be traced back to the presupposition. Building on Chierchia’s (2004) account, he proposes to extend the idea that scalar implicatures disrupt the licensing of NPIs to presuppositions. In other words, the proposal is that in order to check whether an NPI is in a downward entailing environment, we should not only look at the assertion (and its scalar implicatures) but we should also factor in its presuppositions. This means that if a sentence φ containing an NPI presupposes p, we should not ask whether φ creates a downward entailing contexts for the NPI but whether the conjunction p∧ φ does. To illustrate, consider the example above in (19): we saw above that because leads to the inference that its propositional arguments are true. Assuming that these inferences are presuppositions, it is easy to see how they are predicted to disrupt the licensing of the NPI in Homer’s (2010)
7Importantly, he also discusses the case of the strong trigger too, exemplified by (18a), which is degraded (for
some speakers) as compared to the minimal variant in (18b).
(18) (Context: Mary said something interesting during the meeting.)
a. *I doubt that John said anything interesting too. Homer 2010:ex.21 b. I doubt that John said something interesting too.
system. In a case like (19), repeated schematically in (21a), the relevant meaning for NPI licensing is not simply (21a) but it is (21b).
(21) for any propositional argument φ[NPI], ψ:
a. ¬(ψ, because φ[NPI])
b. ψ∧ φNPI∧ ¬(ψ, because φNPI)
One occurrence of the NPI is in a positive (non-DE) environment (the second conjunct in red), thus the licensing is predicted to be disrupted.
In sum, Homer’s (2010) proposal predicts that the presuppositions of because and French cognitive factives intervene in the licensing of NPIs. Notice, however, that while this is an important step towards the understanding of the intervention effects of presuppositions and while it presents some analogies with the case of scalar implicatures, it is not clear how far the analogy goes. More specifically, in the first case, given an account of scalar implicatures as entailments of exhaustified sentences based on the presence of operators in the syntax and an independently motivated minimality constraint we immediately predict intervention effects. In other words, we have a syntactic and a semantic aspect going hand in hand: an agreement relation between an operator and its targets obeying minimality and the result of exhaustification giving rise to contradictions if the semantic environment does not have a certain property (DE- ness). It is the combination of these two factors that gives rise to an explanatory account in the case of scalar implicatures. In the case of presuppositions, instead, we have to assume that a different component of meaning, the presuppositional level, should play a role in the notion of monotonicity required for the licensing of NPIs. This, in turn, conflicts with claims in the literature which requires exactly the opposite: the presuppositional level should be ignored for NPI-licensing purposes (see von Fintel 1999, Homer 2012b for discussion) The present proposal, on the other hand, directly extends the syntax-semantics account of the intervention effects by scalar implicatures to the ones of soft presuppositions, thereby explaining the latter without further assumptions.