In this section I focus on the presuppositional behavior of HEs. In section 1.2.1, I discuss the phenomenon of projection and present an alternative classification of projective content put forward by Tonhauser et al. (2013). In section 1.2.2, I discuss rejection, that is, the various strategies that speakers employ for rejecting the contents that they do not share and are not willing to accept; whether such contents are descriptive or evaluative determines different strategies of rejection. Finally, in section 1.2.3, I discuss an alternative way to explain the projection of the evaluative content that does not rely on presupposition.
A well-known feature of both thick terms and slurs is projection (see Väyrynen 2009, 2013, Eklund 2011, 2013 for thick terms, Croom 2011, Camp 2013, and Jeshion 2013a, 2013b for slurs). Consider the following examples (Väyrynen 2013: 64, 70, 78):
1 7.Madonna’s show is lewd.
2 8.Madonna’s show isn’t lewd.
3 9.Is Madonna’s show lewd?
4 10.If Madonna’s show is lewd, tabloid press will go nuts.
5 11.Madonna’s show might be lewd.
6 π7.Things that are sexually explicit beyond conventional boundaries are bad because of being so.
In (7)–(11), ‘lewd’ seems to be associated with a negative value judgment along the lines of π7,6 even when the thick term is embedded under negation, questions, conditionals or modals. The same happens with slurs:
1 12.Madonna is a wop.
2 13.Madonna is not a wop.
3 14.Is Madonna a wop?
4 15.If Madonna is a wop, she may be late for her own concerts.
5 16.Madonna might be a wop.
6 π12.Italians are bad because of being Italian.
For both thick terms and slurs, the presupposed content triggered by HEs scopes out of semantic embeddings. It is hard to perceive whether a certain utterance presupposes a certain content, when such a content is something that speakers and addressees already take for granted (see Hare 1963), so it could be helpful to test these intuitions by employing an objectionable rather than unobjectionable evaluative term. However, because in principle any thick term could be seen as objectionable, projection data speak for thick terms in general (see Väyrynen 2013: 56).
Projection is typically taken to be the distinctive feature of presuppositions (see Chierchia and McConnell-Ginet 2000). However, recent work in linguistics emphasized the heterogeneity of projective content. In particular, Tonhauser et al. (2013), while leaving aside notions such as ‘presupposition,’ ‘conventional implicature,’ etc., individuate four classes (A, B, C, D) of locutions that give rise to projective contents. The four classes of triggers are distinguished by two properties, (i) strong contextual felicity and (ii) obligatory local effect: class A has both, class B has none, class C has obligatory local effect, but lacks strong contextual felicity constraint, and class D lacks obligatory local effect, but lacks strong contextual felicity constraint. Let us now see what these properties are.
The strong contextual felicity requirement demands that the trigger at stake imposes a strong constraint on the context, i.e. that the occurrence of the trigger requires that a certain content is already entailed by the context before the trigger is introduced, in order for the utterance to be felicitous. Take a context where Lucy is on the bus eating a cucumber salad and a woman whom she does not know starts talking to her and says:
1 17.Your salad looks nice. I never make cucumber salad at home ’cause my daughter hates cucumbers.
The expression ‘my daughter’ triggers the presupposition that the speaker has a daughter. Even though Lucy did not know whether the speaker had a daughter, the utterance is still acceptable:7 this is evidence that the projective content triggered by ‘my daughter’ does not have a strong contextual felicity requirement. Now imagine that the woman goes on and says:
1 18.Our bus driver is eating empadanas, too. (Tonhauser et al. 2013: 78)
This utterance is not felicitous unless a salient person in the context—other than the bus driver—is eating empanadas, which—as far as Lucy knows—is not the case. Because of ‘too,’ (18) requires that a certain content is already part of the context for (18) to be felicitous: ‘too’ imposes a strong contextual felicity constraint.
It is disputable whether slurs and thick terms impose a strong contextual felicity requirement. Take a context where the HE-evaluations of ‘wop’ and ‘lewd’ are not common ground: the conversation’s participants can make sense of the use of these expressions, but it is unclear whether they would be ‘felicitous.’ Suppose that Lucy and the woman on the bus are talking about a movie. The woman utters something like (19) and (20):
1 19.That movie is lewd!
2 20.The director is a wop.
Lucy can easily make sense of (19) and (20), even if she does not share what the woman is presupposing about Italian people and sexual explicitness: this leads Tonhauser et al. (2013) to analyze expressives in general as lacking a strong contextual felicity constraint. However, whether these utterances are felicitous or not seems an open question to me, as the fact that (19) and (20) make sense does not automatically mean that they are felicitous or acceptable.8 In sections 1.2.2 and 1.2.3, I will come back to this point, to discuss what happens when speakers do not share the presuppositions that the inter-locutor is taking for granted.
The second parameter with respect to which Tonhauser et al. (2013) classify projective content is obligatory local effect, that is, they test whether the projecting content needs to be part of the local context of a belief-predicate (i.e. the attitude holder’s epistemic state) or not. For instance, suppose that Mary says:
1 21.Jane believes that Bill has stopped smoking.
The projective content—namely, that Bill used to smoke— has to be ascribed to Jane (and it can be ascribed to Mary, but that is not obligatory). Tonhauser et al. (2013) conclude that the projective content triggered by ‘stop’ shows obligatory local effect. Now suppose that Mary says:
1 22.Jane believes that Bill, who is Sue’s cousin, is Sue’s brother. (Tonhauser et al. 2013: 92)
The projective content of the appositive—namely, that Bill is Sue’s cousin—is not ascribed to Jane, but only to Mary: appositives do not show obligatory local effect. Slurs and thick terms do not show obligatory local effect either. Suppose that Mary—a prudish person and ‘lewd’-user—has a conversation with her friend Jane, who does not share Mary’s perspective on sexuality. While commenting on a movie, Jane remarks that the film was too prim to be appreciated. Later on, Mary reports Jane’s comment to their mutual friend Bill and says:
1 23.Jane told me that if the movie were lewd, it would have been more popular.9
Bill knows both Jane and Mary and is aware that the negative evaluation associated with ‘lewd’ is not to be ascribed to Jane but to Mary. Similarly, imagine that, talking about the same movie, Jane tells Mary that the director is Italian. Later on, Mary reports this information to Bill, in her racist terms:
1 24.Jane