On the Nature of Presupposition: A Normative Speech Act Account

In this paper I provide a new account of linguistic presuppositions, on which they are ancillary speech acts defined by constitutive norms. After providing an initial intuitive characterization of the phenomenon, I present a normative speech act account of presupposition in parallel with Williamson’s analogous account of assertion. I explain how it deals well with the problem of informative presuppositions, and how it relates to accounts for the Triggering and Projection Problems for presuppositions. I conclude with a brief discussion of the consequences of the proposal for the adequacy of Williamson’s account of assertion.


Introduction
Some natural language expressions are conventional indicators of illocutionary types: the interrogative mood conventionally indicates a question; appending the appositive clause 'I promise' to a declarative sentence S-S, I promise-is a conventional indicator of a promise. According to Searle (1969), referential expressions such as proper names, indexicals and demonstratives (and perhaps definite and even indefinite descriptions, in some uses) share an expressive feature that is a conventional indicator of an ancillary speech act, referring-'ancillary' in that it is an auxiliary for the performance of another speech act. In this paper, I argue for an account of presuppositions as speech acts-ancillary ones, in Searle's terms, and I provide a normative account of this ancillary speech-act on which it is constituted by a common knowledge norm. I will assume in my argument a view that, although controversial, many researchers share and I have defended elsewhere: 1 that some expressions traditionally considered presupposition-triggers, such as it-clefts or definite descriptions, are conventional indicators of this ancillary speech act.
An ancillary speech-act view is compatible with rival accounts of presuppositions, such as the influential Stalnakerian picture on which they are propositional attitudes of speakers, or the view that they are logico-semantic relations among propositions. Besides, the evidence discussed here does not decisively confirm the normative proposal that I'll make. However, it shows at least that the view is sufficiently interesting for it to be worth having on the table. In further support of this, I will discuss at the end some implications of the view for Williamson's influential normative account of assertion, on which it is an act answerable to a knowledge norm. I will elaborate on these foundational issues in the remainder of this section.
I find it useful to classify theories of speech acts along two dimensions. First, they can be normative/prescriptive or descriptive, depending on whether or not they posit norms or rules as constitutive of such acts. Secondly, they can be individualist or social/communicative, depending on whether or not they characterize the acts in terms of relations between speaker and hearer, or rather appeal only to mental states of the speaker not involving those relations. For my purposes here, I do not need to go into a more articulated characterization of those conditions; I will limit myself to providing illustrative examples. Let us consider first Williamson's account of the speech act of assertion, and compare it with a paradigm Gricean account, such as the one to be found in Bach and Harnish (1979), for they differ along the two dimensions I am interested in emphasizing. Williamson (1996Williamson ( /2000 claims that the following norm or rule (the knowledge rule) is constitutive of assertion, and individuates it: (KR) One must ((assert p) only if one knows p).
In the course of the debate that Williamson's proposal has generated, other writers have accepted the view that assertion is defined by constitutive rules, but have proposed alternative norms; thus, Weiner (2005) proposes a truth rule, (TR), Lackey (2007) a reasonableness rule, (RBR), and I myself (García-Carpintero 2004, forthcoming) a knowledge provision rule, (KPR): (TR) One must ((assert p) only if p). (RBR) One must ((assert p) only if it is reasonable for one to believe p). (KPR) One must ((assert p) only if one's audience gets thereby to be in a position to know p).
A few clarifications are in order before moving to the contrasting Gricean picture. The obligation (KR) imposes is not all things considered, but prima facie and pro tanto; in any particular case, it can be overruled by stronger obligations imposed by other norms. It has a sui generis source, specific to speech acts. Given this, we can tighten them up. Perhaps the reason why Williamson gives only what I take to be a necessary condition for correctness, and not a sufficient one, lies in the fact that he thought that knowing a proposition does not suffice for making an assertion of it correct. And this is true, of course, with respect to all things considered correctness; however, no unacceptable consequences follow from assuming that there is nothing prima facie wrong with asserting anything that one knows -even if, all things considered, there may well be many reasons for such incorrectness. 2 Thus, we can simplify the analysis by tightening up the rules; instead of (KR), the proposal would be this: (KR 0 ) For one to assert p is correct if and only if one knows p.
Note that both (KR) and (KR 0 )-as the alternative proposals (TR), (RBR), and (KPR)-are intended as characterizations of what is essential or constitutive of assertion (and not, as it might wrongly seem at first glance, merely of correct assertion). The proposal is that assertion is an act essentially constituted by its being beholden to the relevant norm. Thus, on Williamson's view assertion is the unique representational act such that, in performing it, one is thereby committed to knowing the represented proposition (on the alternative views, to its truth, or to its being reasonable for one to believe it); in other terms, the propositional act such that, if one performs it without knowing the intended proposition, one is thereby infringing an obligation. As in Searle (1969) well-known framework, there might be additional features or rules contributing to a full characterization of assertion (i.e., ''sincerity'' or ''preparatory'' conditions, etc.), but (KR)/KR 0 ) characterize in an illuminating way what an act must ''count as'' for it to be an assertion (i.e., what Searle describes as its ''essential rule''). The act must count as being beholden to (KR)/(KR 0 )-and hence, as the agent incurring the commitment to knowing the represented proposition. This leaves unspecified what utterances count as being subject to (KR)/(KR 0 ), but this is, I think, as it should be. If, as Williamson assumes (op. cit., 258) the declarative mood of the whole sentence is a default conventional indicator of assertion, then uttering a sentence in that mood in a default context is such an act; but, even so, there might well be other, indefinitely variegated nonconventional ways of making assertions, similarly counting as being subject to (KR)/(KR 0 )-for instance, by means of indirect speech acts as in the case of rhetorical questions.
As Hindriks (2007) notes, although it is indeed a feature of our assertoric practices that we criticize those that violate rules like (KR) or (TR), this fact about our practices of appraising assertions is by itself insufficient to justify prescriptive accounts. For we also evaluate assertions relative to (invoking Rawls' (1955) wellknown distinction) merely regulative norms, norms that regulate, relative to certain purposes, acts in themselves constitutively non-normative-for instance, as witty, polite or well-phrased. Hindriks shows that norms for assertion could be merely 2 Cf. DeRose (2002) and Hawthorne (2004). Of course, the sufficiency is only for the correctness of the assertion, i.e., for its being permitted, like the necessity -if the condition is not met, the assertion is incorrect, or forbidden. Objections have been raised involving examples of subjects who intuitively know p, but their knowledge is felt to be fallible enough for them not to correctly act on it (in particular, not to assert p) in their circumstances (see, e.g., Reed 2010). I think these objections can be answered by taking into consideration the prima facie character of the correctness at stake; cf. Simion (2016) for elaboration. regulative of a constitutively non-normative practice, definable in the psychological Gricean account that Bach and Harnish proposed, GA below ('R-intending' there is to be explicated in terms of Gricean communicative intentions). The regulative norms in question would then be derived from an ultimately moral sincerity rule such as (SR): (GA) To assert p is to make an utterance thereby R-intending the hearer to take it as a reason to think that the speaker believes p and intends the hearer to believe it. (SR) In situations of normal trust, one ought to be sincere. Bach and Harnish's (1979, 15-6) Gricean account (GA) is a descriptive account, not a prescriptive one like the others mentioned: unlike them, it does not mention norms, but only certain psychological states. Of course, normative consequences would follow, derivative from further rules such as (SR), or perhaps from the fact that the mental states in question are themselves constitutively normative, but this does not suffice for an account to count as normative. By itself, as Bach (2008) insists, (GA) is descriptive, not normative.
Take now the other dimension of variability. Williamson's (KR) account, as (TR) and (RBR), are individualist in that they only mention obligations or commitments of the speaker, obligations or commitments not in regard to anybody else. (GA) and (KPR) are instead social/communicative, in that they invoke relations between speakers and their (perhaps merely presumed) audiences. Promises, as they are usually understood, are the paradigm case of normative and social acts; they constitutively impose obligations on their performers, and they are constitutively addressed to an audience -whose ''uptake'' is required, on most accounts, for them to have taken place.
In this paper I provide a normative and social account of linguistic presupposition, understood as an ancillary speech act. The suggestion, to be developed and more precisely stated in Sect. 3, is that it is an act individuated by this norm: (PR) For one to presuppose p in a context is correct if and only if p is common knowledge in that context.
As I argue in the concluding section, this provides a consideration for a similar kind of account for assertion-one along the lines of (KPR). I believe that norms such as (KR), (TR) and (RBR) are good candidates to characterize the individual act of judging, but not the communicative one of asserting. If (PR) was on the right track, (KR) would not be properly individuative. For, if (KR) defined assertion and (PR) presupposition, whenever it was correct to presuppose p, it would not be incorrect to assert it-indeed, it would be correct to do so, if we assume the stronger (KR 0 ). But this seems wrong.
In the next section I provide a brief characterization of presuppositions. Against skeptical claims to the contrary, the characterization suggests that they are a robust kind, in need of elucidation. I'll also present the influential Stalnakerian account. I'll assume a partially conventionalist solution to what is known as the Triggering Problem for presuppositions. This is the problem of explaining ''why presuppositions arise to begin with'' (Abrusán 2011, 491). 3 This helps the main claim of the paper because, although it is not required that for a norm to be in force there should be a conventional procedure indicating it, there is a good correlation between norms and conventions indicating when an act is to be beholden to them. Also, norms are not mere regularities in the behavior of some subjects; they should guide such behavior. Norms for speech acts are, of course, not codified anywhere; they are like the norms of traditional games that nobody has cared to articulate yet (Williamson 1996(Williamson /2000. If there is a default conventional indicator that an act is bound by the norm, we can check whether in such cases speakers manifest some sensitivity to the alleged norm. In the third, main section of the paper, I suggest that some problems that have been raised for the Stalnakerian account are a particular case of familiar difficulties for descriptive accounts of speech acts. Although there is no principled incompatibility between a normative speech act account like the one I offer and Stalnaker's influential picture, Stalnaker himself (2014, ch. 2) advocates a Gricean descriptivist view of assertion and presupposition. I show how a normative account deals with the problems, thus providing support for the main claim of the paper, which goes as follows. There are serious qualms with descriptive accounts of assertion, and good motivation for a normative account thereof. Analogous considerations provide motivation and justification for a normative-cum-social account of presuppositions. The fourth section critically discusses the most influential contemporary forms of a skeptical attitude about presuppositions, on which natural languages merely codify a distinction between ''at issue'' and ''background'' contents, but not specific requirements on a contextual common ground. In closing, I discuss the consequences of the proposal for the adequacy of individualist accounts of assertion such as Williamson's, suggesting instead that a proper account should also be communicative, i.e., social.
A normative account along the lines of the one I defend here, according to which presupposing is a conventional, institutional practice, is taken for granted by many linguists. 4 However, as far as I know it has never been presented in the way I do, using recent proposals in theories of assertion, and adapting arguments that have been deployed in that literature. Philosophers sympathetic to the anti-normativist view perspicuously articulated by Stalnaker himself in his recent work (2014, ch. 2) will resist it. As far as I know, the consequences I point out for the adequacy of most recent accounts of assertion are equally new and relevant. 3 Abrusán (op. cit., 492) distinguishes two answers in addition to agnosticism: ''The first … is that presuppositions are just an arbitrary special type of meaning specified by the lexicon, requiring their own set of rules for combining with other elements when embedded in larger contexts. According to the second view … presuppositions might arise via pragmatic means from assumptions about rules that rational interlocutors follow, just like conversational implicatures.'' The first is the account I favor for most cases of linguistic presupposition. As has been pointed out before, however, there also are pragmatically triggered presuppositions; I'll provide illustrations below. There are thus cases to which the second view applies, but I am not persuaded that it can be generalized to all cases. Abrusán goes on to defend what she takes to be a third, ''cognitivist'' account. See below, Sect. 4 for discussion. 4 Cf. fn. 14 for an example.

Presupposition
Our semantic competence underwrites the validity of inferences such as the following, for both (1) and (2): (1) John infected the PC.
(2) It was John who infected the PC. ; Someone infected the PC.
However, there is a difference between the syntactic constructions in (1) and (2); unlike the less marked way of expressing the same content in (1), the cleft construction in (2) also validates (at least, in default contexts) the inference when placed under different embeddings, such as negation-(3), conditionals-(4), modals-(5), and still others; presuppositions are said to be thereby ''projected'', i.e., inherited by the embedding constructions: (3) It was not John who infected the PC.
(4) If it was John who infected the PC, the Mac is also infected. (5) It may have been John who infected the PC. ; Someone infected the PC.
Other presuppositional constructions exhibit this behavior; for a second illustration consider the case of definite descriptions: (6) The station newsstand sells The Guardian. ; There is exactly one station newsstand.
Presuppositions do not project in some cases; hence, they are not there (as this is usually put in the literature) ''globally''-the whole utterance lacks the presupposition -, although they remain ''locally''-the embedded clause still carries it: (10) If someone infected the PC, it was John who did it. (11) Someone infected the PC, and it was John who did it.
Conventional implicatures, which intuitively differ from presuppositions, share their projection behavior in the embeddings in (3)-(5); following Potts (2007), I use non-restrictive wh-clauses as illustrative examples: (12) John, who infected the PC, teaches in Oxford. (13) It is not the case that John, who infected the PC, teaches in Oxford. (14) If John, who infected the PC, teaches in Oxford, he will attend the conference. (15) It may be the case that John, who infected the PC, teaches in Oxford. ; John infected the PC.
Geurts (1999,(6)(7)(8) uses the projection behavior illustrated by (3)-(5) and (10)-(11)-the projection of the presupposition in the former and its filtering in the latter-as an intuitive test to characterize presuppositions. As he acknowledges, however, the test is defeasible, which shows that it provides only an initial characterization of the phenomenon it singles out, as opposed to a full-fledged definition. 5 Following Shanon (1976), von Fintel (2004 proposes an alternative hey, wait a minute (HWAM) test to distinguish presupposition and assertion. Consider the following dialogues, with '#' being an indication of impropriety or infelicity: (16) It was not John who infected the PC. (17) # Hey, wait a minute, I had no idea that John did not infect the PC. (18) Hey, wait a minute, I had no idea that someone infected the PC. (19) It is not the case that John, who came this morning, teaches in Oxford. (20) # Hey, wait a minute, I had no idea that John does not teach in Oxford. (21) #? Hey, wait a minute, I had no idea that John came this morning.
Intuitively, the reason why the HWAM test tracks presuppositionality is this. Assertions are presented as providing new information; thus targeting asserted contents with the HWAM objection does not feel right. In contrast, presuppositions are presented as information already shared by the conversational participants, hence the HWAM objection is fully pertinent. 6 Together with the projection behavior, the HWAM test is thus a useful instrument for singling out the phenomenon of presuppositions. A proper identification should be done by the correct theoretical account. Given current knowledge, including cross-linguistic evidence, this should in my view involve a finer-grained delineation of the projection behavior of presuppositions, in contrast 5 See also Chierchia and McConnell-Ginet (1990, 283), Beaver (2001, 19-22), and Kadmon (2001, 13). Data suggesting that presuppositions might fail to project include the ''metalinguistic negation'' cases of ''cancellation'' illustrated in fn. 34, but also the data allegedly showing a distinction between ''hard'' and ''soft'' triggers, discussed below in Sect. 4. Related, it is to be expected that pragmatically triggered presuppositions (see next footnote for illustration and references) fail to project in some contexts. A very good reason suggesting that presuppositions might fail to be filtered is given by Roberts' (2004) compelling arguments that focus presupposes questions: not propositions stating that a given question has been asked, but that the questions themselves are in the common ground. For such presuppositions can hardly be filtered, given that the antecedent of a conditional can merely express a proposition, not a question (see García-Carpintero 2017a). 6 The judgment reported in (21) is far from unanimous. Cummins, Amaral and Katsos (2012, 4) contend that ''it seems felicitous to use ''Hey, wait a minute'' to object to any precondition of the utterance, no matter how obscure (and perhaps even to an aspect of foregrounded meaning, if it is particularly surprising)''. A referee illustrates this: ''Suppose I say 'John can drive me to the airport.' You can reply: 'Hey wait a minute! I had no idea John had a driver's license.'. (The referee goes on to say, ''That John has a driver's license is not, however, a presupposition of what I have said (it doesn't project out of standard holes).'' It is certainly true that this is not a conventionally triggered linguistic presupposition-a presupposition standardly attached to a phrase or construction; hence, not one of ''what is said'', as this is usually understood. However, I think it is a pragmatic, conversational one (Chierchia andMcConnell-Ginet 1990, 285 ff, Simons 2005, 150-1). Whether or not it projects thus depends on pragmatic issues, but, to my ears, it standardly does in 'John, can you drive me to the airport?', or in 'John will not drive me to the airport'). The issue should be investigated empirically, even though methodologically this is no easy matter [cf. von Fintel and Matthewson (2008, 184) and Tonhauser et al. (2013, 81)]. In any case, as emphasized in the main text, I only take the two tests as good intuitive indication of a distinctive phenomenon in need of theoretical explanation. Its proper definition, if the indicated phenomenon proves to exist, should be given by the correct theoretical account.
with other background assumptions such as conventional implicatures (cf. Potts 2007), and a three-valued semantic framework. But by itself the tests do reveal intuitions showing that the skepticism expressed by some writers is prima facie unreasonable. 7 Such intuitions are quite robust, and they are robustly related to grammatical constructions like those we have used for illustration, as the reader might establish by considering variations on them, or others in the lists given by Levinson (1983, 181-5) and Beaver and Geurts (2011). This at least prima facie suggests that we are confronted with a sufficiently ''natural'' kind, amenable to a precise characterization. 8 What we are after is an adequate definition, which, if it is good, should help us to delineate the phenomenon and to distinguish its true instances. This is the task to which I will try to contribute below; I conclude by presenting Stalnaker's proposal, on which I'll base mine.
In influential work starting in the early 1970s, Stalnaker (1973Stalnaker ( , 1974Stalnaker ( , 2002 has provided an account of the phenomenon of presupposition. The account has been slightly modified along the years; here I will just present what I take to be its core, as stated in recent works. 9 Stalnaker's proposal-as he more emphatically explains in recent foundational work, cf. Stalnaker (2014, ch. 2)-is in the spirit of Grice's account of phenomena such as conversational implicature in particular and meaning in general: it purports to explain those phenomena as a specific form of rational behavior involving communicative intentions, averting irreducibly social notions such as conventions or (socially construed) norms. 10 Bach and Harnish's account of assertion (GA) above is an example of such Gricean accounts.
Stalnaker bases his analysis on a notion of speaker presupposition, which he then uses to define a notion of sentence presupposition. Speaker presupposition is explained in terms of common beliefs about what is accepted by the conversational partners; and common belief follows the pattern of Schiffer's and Lewis' proposals about it and about common knowledge: p is common belief in G just in case (almost) everybody in G believes p, believes that (almost) everybody in G believes p, and so on. Acceptance is in its turn defined by Stalnaker (2002, 716) as a category of mental states ''which includes belief, but also some attitudes (presumption, assumption, acceptance for the purposes of argument or enquiry) that contrast with belief and with each other. To accept a proposition is to treat it as true for some reason.'' The need to invoke acceptance in the definition derives from many cases in which, intuitively and according to our initial characterization above, p is presupposed while not commonly believed. Thus, consider Donnellan's (1966) example: the secret conspirator asks the usurper's minions, ''Is the king in his counting house?'' The speaker does not believe that the intended referent is king, nor perhaps that there is a king, and hence does not believe that these propositions are commonly believed in the context; but nonetheless it is presupposed that the referent is king (and hence that there is a king). 11 Nonetheless, belief is the default setting for acceptance: ''divergences between what is believed, and mutually believed, and what is accepted, and commonly accepted need to be explained'' (Stalnaker 2014, 45). 12 This is thus the current form of the account. We first define a proposition p to be in the common ground in a group G-CG G (p)-and then we define speaker presupposition: (CG G ) CG G (p) if and only if everybody in G accepts p, everybody accepts that everybody accepts p, … (SpP) Speaker S presupposes p (relative to G) if and only if S accepts that CG G (p) Stalnaker (1973Stalnaker ( , 451, 1974 then defines a notion of sentence presupposition based on this notion: (SnP) Sentence S presupposes p if and only if the use of S would for some reason be inappropriate unless the speaker presupposed p Stalnaker (1978) complements this analysis of presuppositions with an analysis of assertion also deservedly influential, on which an assertion is a proposal to update the common ground, which, if accepted, is ''added'' to it (i.e., it then becomes common belief that every participant accepts it). 13 He combined the two accounts to suggest intuitively plausible explanations of some aspects of the projecting behavior we presented in the previous section. This (together with the related independent work of Lauri Karttunen) was the origin of the new important tradition of Dynamic 11 I believe this is an intuitively correct characterization of what is presupposed in this case, which I take to be a referential use of the description. In general (see García-Carpintero 2000, 2017b all cases of reference involve in my view ''identification'' presuppositions. This fits Searle's (1969) view that the presuppositions of definite descriptions and other referring expressions are the felicity conditions for the ancillary speech act of referring. 12 Stalnaker does not provide any justification for the claim, which I take to address the concerns that led Stalnaker (2002, 716) to a stronger definition of common ground in terms of beliefs about what is accepted. It is a one more consideration in favor of my view below that it does: presupposing constitutively involves a norm of common knowledge; manifest violations by rational and competent speakers must be non-default, and thus explainable. 13 I take Stalnaker's account of assertion to be compatible with normative views; Stalnaker himself wouldn't like the combination, given the already mentioned reservations he has about them, fully articulated in his more recent work (2014, ch. 2) but present from the start. Semantics, DS, developed for instance in Heim (1983), Beaver (2001) or von Fintel (2004. This tradition has the resources to provide the theoretical account that is needed, for reasons given above. Unlike the traditional account of presuppositions as conditions on the truth and falsity of claims, it can explain the selective projection behavior we have seen to be characteristic of presuppositions, and it can distinguish them from conventional implicatures, accounting also for the adequacy of the HWAM test; it also has the resources to explain the phenomenon of noncatastrophic presupposition failure (cf. von Fintel 2004). Geurts (1999, 17), however, is right in pointing out the important conceptual differences between the DS tradition and Stalnaker's viewpoint. Relinquishing Stalnaker's Gricean reductive aims, in this tradition presuppositions are taken to be, both with respect to their triggering and projecting behavior, a constitutive feature of the semantics of natural language expressions. 14 I will henceforth assume it. 15 Normatively constituted speech acts can be made with purely pragmatic resources, without invoking conventional procedures established in order to help making them; presuppositions, normatively understood in my account, are no exception (Chierchia and McConnell-Ginet 1990, 285 ff;Simons 2005, 350-1). 16 However, there typically are conventional devices for such purposes-such as the declarative mood of a whole sentence in the case of assertion. Moreover, Williamson's strategy in arguing for a normative account of assertion-which mine for the case of presupposition replicates-assumes conventional indicators of the speech act. Assuming that there is a default conventional indicator that an act is bound by a norm, we can check whether when using it speakers manifest some sensitivity to the putative norm and are to that extent guided in their behavior by its being in force. In brief, my point will be that considerations analogous to those that have been used to justify normative accounts of assertion similarly suggest one for the conventional devices that trigger presuppositions. 14 While Geurts distances himself from Dynamic Semantics on account of their betrayal of Stalnaker's truly pragmatic stance, like Stalnaker he helps himself to a notion of expression-presupposition, defined in terms of the pragmatic notion of speaker presupposition: ''In the previous section we defined presuppositions as inferences that are triggered by certain expressions, and that exhibit projection behavior. For Stalnaker, a presupposition is an assumption which a speaker takes for granted. These two definitions may seem to contradict each other, but the contradiction is only an apparent one. For we can plausibly say that a given form of words requires that the speaker presuppose something or other.'' (Geurts 1999, 14;my emphasis). Like Stalnaker's, this definition uses normative notions; unlike Stalnaker, Geurts also appeals to the normative notion of commitment in characterizing speaker presupposition: ''a speaker who presupposes something incurs a commitment … regardless whether he really believes what he presupposes'' (ibid., 11). Geurts never explains where those requirements and commitmentsprima facie at odds with the Stalnakerian stance he vows to adopt -come from. The account I'll provide in the next section does explain it. 15 I have defended this elsewhere, see (García-Carpintero 2016a). There is a serious concern that I cannot confront here, in that, like Stalnaker's, but unlike DS, my account below prominently features traditional propositions. If pressed, I could assume instead Schlenker's (2009) bivalent and static account, with which the proposal below is also compatible. 16 See fn. 6 for illustration. I will present now the normative account I announced in the first section, and I will then motivate it and provide justification for it. In parallel with the presentation of the Stalnakerian account above, to present my proposal I first modify the previous definition of common ground in a group G, going back to Lewis' original account. Then I'll define in the normative terms outlined in the first section the ancillary representational act of presupposing, and sentence presupposition in terms of it. I follow Tonhauser et al. (2013) in calling triggers expressive devices such as words (definite descriptions, names and indexicals, factive verbs, change-of-state verbs, manner adverbs), phrases and structures (clefts, temporal clauses) or intonation patterns (focus, contrast).
For one to presuppose p (relative to G) is correct if and only if CG 0 G (p) (SnP 0 ) Sentence S presupposes p if and only if some trigger in S conveys that acts of uttering S are subject to instances of the norm (PR). 17 As just announced, I'll follow Williamson's (1996Williamson's ( /2000 path in his proposal for assertion in order to motivate this parallel view for presupposition. He assumes that the declarative mood is a default conventional indicator of the speech act he wants to theorize about: ''In natural language, the default use of declarative sentences is to make assertions' ' (op. cit., 258). This gives us an independent specification of the phenomenon that we aim to characterize: it is the act, whatever its proper definition is, that is in fact associated with the indicative mood in natural languages as used on some occasions (the default cases), and which speakers intentionally purport to make by such means on such occasions. Similarly, (SnP 0 ) assumes that there are conventional indicators of an ancillary speech act, presupposing.
Note however that, although sentence presupposition has in this way heuristic precedence over speaker presupposition, the present proposal agrees with Stalnaker's in considering the latter as explanatorily fundamental. 18 It nonetheless differs from his proposal, in that (1) (PR) makes explicit the assumption that a speaker presupposition is (when linguistically triggered) a speech act (and not just a propositional attitude) with a normative character, which he rejects; and (2) (SnP') makes a commitment to there being expressive resources that conventionally 17 I am assuming here that whether or not a trigger conveys that utterances of sentences including it are bound by instances of the norm (PR) depends on the compositional projective behavior of operators and other constructions. A trigger that conveys it in an atomic sentence might fail to do so when embedded under some operators (Kattunen's filters and plugs). What it is for a trigger to convey a presupposed content should thus be established by a compositional account of the triggering and projection of presupposed contents, such as those offered by DS, or by Schlenker (2009). 18 Simons (2013) classifies views on presuppositions into utterance and clause level. But as she notes, Stalnakerian accounts may be both, as in fact the present proposal is. They are both utterance level in a fundamental sense, because they explain presupposing pragmatically-as a normatively constituted speech act, on the present proposal; in terms of the attitudes that uttering speakers are required to have, in Stalnaker's. They may be clause level, in that they are compatible with conventional triggers, and compositional ways for projecting them. Of course, while I hypothesize conventional triggers, Stalnaker is much less enthusiastic, appearing to prefer to get rid of them. convey acts of speaker presupposition, which he allows, but would rather avoid. It thus rejects the skepticism about the phenomenon of linguistic presupposition discussed in Sect. 2, by acknowledging that the intuitive impressions we have of the phenomenon -manifested in the intuitive tests mentioned there-do track a real, substantive kind. 19 Presupposing is on this view a pragmatic notion involving attitudes of speakers-a (linguistically ancillary) speech act-but it is one defined in normative terms. The norm (PR) is specifically illocutionary (not moral, or prudential) and defeasible by stronger moral or prudential norms, as in the assertoric cases discussed in the first section. Indication that it applies can be triggered in different ways, even when standardly associated with a sentence-type-allowing for conventional triggering in such cases, but not requiring it. 20 I will now provide abductive justification for this proposal, as Williamson does for his (op. cit., 239). I'll argue that it offers better explanations for the phenomena than alternative accounts. I will present this justification in parallel to Williamson's for his prescriptive account of assertion. As said at the outset, there are serious qualms with descriptive accounts of assertion, and good motivation for a normative account thereof. I will argue that analogous considerations provide motivation and justification for a normative-cum-social account of presuppositions. 21 As a first motivation for his account, Williamson (1996Williamson ( /2000 mentions intuitive conversational patterns: we challenge assertions politely by asking ''How do you know?'' or, more aggressively, ''Do you know that?'' 22 Austin (1962, 138) pointed out these patterns: [I]t is important to notice also that statements too are liable to infelicity of this kind in other ways also parallel to contracts, promises, warnings, &c. Just as we often say, for example, 'You cannot order me', in the sense 'You have not the right to order me', which is equivalent to saying that you are not in the appropriate position to do so: so often there are things you cannot state -have no right to state -are in no position to state. You cannot now state how many people there are in the next room; if you say 'There are fifty people in the next room', I can only regard you as guessing or conjecturing.
We observe a similar pattern in the case of presuppositions, addressing here not just whether the speaker knew the presupposed proposition, but whether it is shared knowledge. Thus, we find annoying the practice of disparaging our views by presupposing the criticism, thus pretending that we agree on the disparaging: ''How 19 If I understand him properly here, I take it that this gives to the ''mysterious relation X'' the sort of explanatory and descriptive role that Stalnaker (2002, 712-3) disparages. 20 I discuss in the next section the alternative account by Simons et al. (2011). 21 The social/communicative character of the present account of presupposing lies of course in its appeal to requirements on common knowledge. The prescriptive character of presupposing could also be defended in terms of weaker notions, such as common belief or acceptance, and consequences for assertion would still follow, especially if its proper analysis also invoked weaker notions. It is only to facilitate my presentation, which draws on the analogies with Williamson's (KR) account, that I run the discussion in terms of knowledge. 22 Williamson refers to Unger (1975) as a precedent for his view; Unger (1975, 263-4) mentions as data such conversational challenges. Cf. also Turri (2010). was the talk?-You know, as usual with this guy, premised on the typical Fregean confusion of semantics with psychology'', told to us in full awareness that we share the putative confusion. I think we find this annoying because, while an assertion is presented as something we are entitled to reject, a presupposition is presented as something we already know, and thus are not supposed to question- Langton and West (1999, 309) make a similar point. The annoyance manifests an intuition that plays for the case of presupposing a similar role than the one revealed by Austin's point does for assertion. It is also this intuition that the HWAM objection appears to be tracking; for in saying ''I didn't know/had no idea …'' one is questioning that the proposition is in the common ground in the sense characterized in (CG' S ), or at the very least that it is shared knowledge-knowledge that all in the group have. 23 As I said above, Hindriks (2007) notes that these facts about our evaluative practices are by themselves insufficient to justify prescriptive accounts. For we also evaluate acts relative to merely regulative norms, norms that regulate, relative to certain purposes, acts in themselves constitutively non-normative-for instance, as witty, polite or well-phrased. In the case of assertion, there are, however, important considerations telling against ''regulative norms'' accounts such as (GA), and in favor of prescriptive accounts. There are well-known objections to Gricean accounts of speaker-meaning in general, of which (GA) is a special case for assertoric meaning, which suggest that prescriptive accounts are worth taking seriously. 24 For instance, the clerk in the information booth uttering ''The flight will depart on time'', or the victim saying to his torturer ''I did not do it'', or any of us saying to our neighbor in the lift ''nice weather, isn't it?'', may well lack the Gricean intentions that (GA) requires for them to assert: they may not care about what their audiences think they believe, know it is useless to try, or irrelevant. But they are asserting all right. Normative accounts would capture this, for, no matter their intentions, they are still committed to knowing what they say (or having justification for it, or being truthful, or, in the (KPR) I favor, to put their audiences in a position to know). Defenders of Gricean accounts consider objections like these, and they contend that they have good responses to them. 25 I do not think so, but I lack the space here to properly examine the matter, as I lack the space to critically examine alternative descriptive accounts. I hope it is enough for me to remind the reader of the problems, and to point out the similarities between them and the difficulties I have discussed at length elsewhere (García-Carpintero 2016a) for the influential descriptive Stalnakerian account, as I am about to do. The goal of this paper is to articulate an alternative normative view modeled on Williamson's account of 23 As indicated above (fn. 6), some researchers think that the HWAM objection just tracks non-at-issue content. Given that such contents project from embedding under negation, we cannot object to them with a straightforward denial; we need more roundabout ways. I myself feel that the 'hey, wait a minute, I didn't know/I had no idea …' objection, with focus on the main verb as opposed to the embedded clause (cf. Simons et al. 2017 for the significance of this), is particularly pertinent for presuppositions. It can also be adequate for conventional implicatures if focus is on the embedded proposition; but it is then this that is being objected to. Otherwise I would choose some other circumlocution making this clear, ''hey, wait a minute, are you sure that John came this morning?'', say, instead of (21) above. In any case, this is only one piece of evidence, intended just as part of an abductive argument. 24 Cf. Vlach (1981), Davis (1992), Alston (2000, ch. 2), Green (2007, ch. 3). 25 Cf. for instance Hindriks (2007, 400). assertion, and to motivate it additionally in parallel with Williamson's abductive justification for his account.
The well-known phenomenon of informative presuppositions tells against descriptive accounts of the phenomenon in parallel ways to those mentioned for the case of assertion. As other writers have pointed out, 26 it poses a serious problem for common knowledge accounts. As Stalnaker (1973Stalnaker ( , 449, 1974 noted in his early writings, it is frequent for speakers to communicate information by uttering sentences that carry presuppositions with such contents. These are examples from Abbott (2008, 531, cf. sources there): (22) The leaders of the militant homophile movement in America generally have been young people. It was they who fought back during a violent police raid on a Greenwich Village bar in 1969, an incident from which many gays date the birth of the modern crusade for homosexual rights. (23) If you're going into the bedroom, would you mind bringing back the big bag of potato chips that I left on the bed?
Speakers who utter sentences (22) and (23) do not typically assume their presuppositions-that some people fought back during a violent police raid on a Greenwich Village bar in 1969, and that there is exactly one big bag of potato chips that the speaker left on the bed, respectively-to be in the common ground. To utter sentences with those presuppositions is just an expedient resource for speakers to tersely inform their audiences about them, in addition to what they assert. 27 Most writers, however (Stalnaker certainly among them), would like to count the relevant contents as nonetheless somehow presupposed. 28 The examples above help us to appreciate the extent of the phenomenon, but usually simpler cases are discussed. 29 We are to assume that the speaker utters (24) in the knowledge that his audience knows nothing about his pets: (24) I cannot come to the meeting-I have to pick up my cat at the veterinarian. 26 Cf. Abbott (2008), Gauker (2008) and Simons (2006). I of course disagree with their contention that the appeal to accommodation to deal with informative presupposition renders common knowledge accounts of presuppositions vacuous. On the argumentative line I will sketch, the phenomenon poses problems to purely pragmatic views, but accounts that assume semantic triggering (including Stalnaker's official own, which officially allows for it) can escape them. Cf. also the discussion in the next section. 27 In some cases, the primary goal of the speaker is to convey the presupposition, as in the notorious exchange: 'The new boss is attractive-yes, his wife thinks so too'. Alan Ryan's review of John Stuart Mill: Victorian Firebrand (''The Passionate Hero, Then and Now'', New York Review of Books, 2011, 19, 60) contains the following quotation from the book with a nice real-life example; it refers to Mill's first encounter with Harriet Taylor, who would become his very special friend for twenty years until the death of her husband, and then his wife: ''In many ways, it was not a surprising match. Harriet Taylor was intelligent, pretty, vivacious, progressive, open-minded and poetic. But his admiration was shared by others -not least by her two children, and her husband''. 28 My main reason for acknowledging the presence of a conventionally triggered presupposition in these cases is given by the goal of providing a systematic compositional account of their semantics. As explained below, this is also required to properly account for the presence of a genuine speaker presupposition along the lines suggested by Stalnaker. 29 Stalnaker (1974, 52, n. 2) attributes an example like (24) to Jerry Saddock.
Here the speaker ''presents herself'' as if she thought that the presupposed proposition was in the common ground, exactly as the clerk in the information booth ''presents himself'' as if he had the information that his audience needs. They need not have the attitudes that descriptive accounts ascribe them.
Stalnaker accounts for these cases by pointing out that when an utterance triggers (semantically or pragmatically) a presupposition, evaluation of the utterance ideally involves two moments, an earlier ''presupposition evaluation time'' PT at which the presuppositions are checked, a later ''assertion evaluation time'' AT at which the assertoric content is appraised. If an utterance of (24) triggers the presupposition that the speaker has a cat, this is to be in place at PT for it to be correct. Nevertheless, pointing this out does not explain how the presupposition was triggered in the first place. 30 The account he provides of the inferential process that the audience is intended to follow in his more recent work (2014, makes this manifest, because the crucial premise is the recognition by the audience that the speaker was making the relevant presupposition (''you manifestly are presupposing that you have a cat' ', op. cit., 48). How does the speaker communicate that she is doing that? 31 Even if the speaker can be said to have the attitude-believing that the informative presupposition will have become common knowledge at presupposition evaluation time, PT -, her act must still provide the means for their audience to recognize it.
There are clear parallels we can draw between assertion and presupposition here; consider this case 32 : A observes B get in a muddle, and inadvertently commit the fallacy of affirming the consequent. She responds by asserting the law modus ponens. This is a criticism because such an assertion is appropriate only if B does not know what modus ponens is; so, by asserting this A suggests that B does not know what modus ponens is. Just as B can pick-up on A's criticism only if he can identify A's speech act as an (inappropriate) assertion, so the addressee of (24) can pick-up on the speaker communicating to her that he has a cat only if she can identify his speech act as a presupposition, which would be inappropriate unless accommodated at PT. The semantics of the definite NP explains how the addressee can recognize that the speaker's act is that of presupposing, despite the fact that it would be inappropriate to presuppose it unless accommodated. 30 Cf. Abbott (2008), Simons (ms) and García-Carpintero (2016a). 31 This is why Geurts's (1999, 11) normative characterization of speaker presupposition, ''a speaker who presupposes something incurs a commitment … regardless whether he really believes what he presupposes'', which is along the right lines, is doubtfully compatible with his Stalnakerian avowals, as indicated in a fn. 14. The commitment is incurred independently of the speaker's attitudes, in that she ''presents herself'' as having them. We need to explain how this happens, and I think the evidence suggests that in some cases only a conventional triggering of the presuppositions, with normative implications in itself, provides adequate explanation (cf. Tonhauser et al. 2013;Abrusán 2016). The evidence shows that whether or not presuppositions are felt to be projected depends in complex ways on issues involving focus and question-under-discussion (cf. Simons et al. 2017;Abrusán 2016), but I take conventional triggering and compositional projection to be compatible with a deeper explanation of how linguistic presuppositions work, in part in terms of their ''not-at-issue'' character. See Sect. 4 for further discussion. 32 Thanks to an anonymous referee for suggesting the example.
Let us see now how the parallel reemerges when we consider some of the additional justification that Williamson provides for his specific normative proposal. First, he notoriously relies on intuitions about lottery cases, in which, knowing that you hold a ticket in a very large lottery, I assert ''your ticket did not win'' only on the basis of the high probability of the utterance's truth (ibid., 246-52). (KR 0 ) predicts, in accordance with our intuitions, that the speaker is doing something wrong. But presupposing the relevant proposition, as opposed to asserting it, sounds equally objectionable in those situations: ''don't you realize that your ticket didn't win?''; ''you can throw away that losing ticket that you bought''. 33 Second, Williamson argues that (KR 0 ) explains what is wrong in a version of Moore's paradox with 'know' instead of 'believe': A, and I do not know that A (ibid., 253-4). But we can easily obtain analogues of Moorean paradoxes when the speaker is not asserting but presupposing, which would be similarly explained by the account I have proposed: ''It was not John who infected the PC, and I do not know that anybody infected the PC''; ''Peter doesn't realize that his wife betrays him, and I do not know that Peter's wife betrays him''. 34,35 As in the case of assertion assuming (KR 0 ), given the present characterization there will be many cases in which speakers make a mistake in presupposing something that in fact is not common knowledge, while intuitively they are not to be blamed. Some can be explained along the lines suggested by Williamson for the corresponding cases involving assertion (ibid., 256-9). In some situations, it is reasonable for us to think that something is common knowledge, and hence that we are not violating the norm in presupposing it, even if it is not and we in fact are; what we do is not permissible, but it is exculpable. In some other cases additional 33 An anonymous reviewer pointed out that the latter utterance sounds much better if made after the speaker says, ''I don't know why I bought this ticket, it's obviously not going to win''. To my ears, it sounds acceptable only if I understand it as something like ''then/if so, you can throw away that losing ticket of yours''. The ''then/if so'' here makes it clear that we are taking the utterance to be justified only under the assumption that the speaker's utterance allows, hence in a context in which the presupposition is satisfied. 34 There are contexts-described in the literature as ''presupposition cancellation''-in which nothing wrong would be felt with those utterances, as in this variation on Keenan's well-known examples: ''It was not me who infected it, it was not Mary who infected it, it was not John who infected it … in fact, I do not know that anybody infected the PC''. Given that the presuppositions of it-clefts are-I thinkconventionally triggered, I do not accept that they can be contextually suspended the way that conversational implicatures do. I agree with Chierchia and McConnell-Ginet's (1990, 314-5) diagnosis: the speaker rhetorically contradicts a remaining presupposition for the purposes of challenging and changing the contextual assumptions. Cf. the discussion of indirection below. 35 A reviewer points out that these cases are also explained ''if there is a knowledge norm on assertions and presuppositions are entailed by asserted material, assuming that the knowledge norm is closed under at least known entailments''. This is correct, and I agree that there is a sense of ''assertion'' on which what is asserted entails what is presupposed, for conventionally triggered presuppositions-one corresponding to what Bach (1999) takes to be what is said: a content explicitly articulated to whose truth the speaker represents herself as committing, including for instance conventional implicatures. But this notion simply disregards the distinction between at issue and not at issue content; only the former is properly assertoric content. In order to fully account for the data, we also need this stricter notion of assertion -concerning only ''at-issue'' content-that I am here contrasting with presupposing; see García-Carpintero (2016b), forthcoming for further elaboration. values (saving someone from danger, enjoying a relaxed conversation) are at stake, similarly allowing for exculpation. 36 Other apparent violations are instances of pretense, or of indirection. Although a presupposition is not common knowledge, speakers make as if it is so for different reasons: creating and enjoying fictions, examining suppositions, facilitating conversation, etc. Conventionally (and semantically, on the view of the semantics/pragmatics divide assumed here), whoever utters 'Who the heck would want to see that film?' is asking a question, but the speaker is not asking anything, she is in fact asserting. Conventionally, and semantically, an utterance of 'Paul is a good friend' is an assertion that Paul is a good friend; in some contexts, it might be perfectly clear that the speaker is not making such an assertion, but in fact one with a contrary content. Conventionally, and semantically, 'Thanks for not browsing our magazines' is an expression of gratitude, but when we find an utterance of it in the train station kiosk, we know that its author was doing no such thing, but in fact making a request. In related work (García-Carpintero ms) I have provided an account of what is going on in such cases, based on Grice's account of conversational implicatures. Speakers utter sentences that, taken literally in their contexts, would be used to make specific speech acts; but they do not (or do not primarily) make these speech acts. Their primary aim is to make other speech acts, counting on conversational maxims to determine them and for their audiences to adequately interpret them.
Something similar applies, on the present account, to cases in which speakers use sentences that presuppose contents that they know are not common knowledge or have any chance to become so, as in the example by Kripke that Stalnaker (2014, 65-66)  The current proposal explains informative presuppositions as relying on accommodation understood as a repair strategy, along the lines of Lewis' (1979) account. 38 The process can also be understood using Stalnaker's distinction between the two times of evaluation PT and AT. Speakers who utter ''I am sorry I am latemy car broke down'' use a sentence that presupposes that they have a car. They thereby present themselves as performing an act-ancillary to their main speech act-which is correct if, and only if, it is mutually known to all involved that she has a car. Following Stalnaker, speakers can be described as not merely presenting themselves as carrying out this act, but as correctly performing it. But for this they 36 This might be what is going on in Donnellan's ''the king in his countinghouse'' example; or it could be a case of pretense or indirection like those discussed below. Both accounts are in accordance with the explanation I suggested above in fn. 12 for the assumption about belief and acceptance that Stalnaker (2014, 45) makes without justification. 37 As indicated in the main text, I take these to be examples of indirect speech acts on the assumption that I am arguing for here, that presuppositions are ancillary speech acts. 38 Cf. Simons (2003, 258-61) on differences between Stalnaker's and Lewis' views. On Stalnaker's ''presupposition evaluation time'' account outlined above, the presuppositions of a sentence are requirements on speakers' attitudes, which might be in place in these cases, so there may well be nothing to repair; speakers at most need to adjust their attitudes to get them in line. On the present proposal presuppositions are instead requirements on the common ground (see also von Fintel 2008, 140). are relying on their audiences' awareness of a presuppositional requirement attached to the form of words they are using, and their awareness that, in their context, the requirement prima facie is not met, the speaker knows that it is not, and knows that the audience knows that it is not, in order to indirectly provide them those pieces of non-controversial background information in a conveniently brief and non-verbose way. 39 In order to appreciate that the speaker presupposition of (24) prima facie is not in place, note that in the ordinary case of the presuppositional requirements of definite descriptions (unlike the case of those of indexicals) no inference based on the utterance is needed for them to be satisfied at PT. The speaker who utters ''the PM is giving a speech at the parliament'' can count on the presuppositional requirements of the definite to be satisfied before she starts speaking. The claim that the presupposition in (24) is after all satisfied relies thus on Stalnaker's distinction between PT and AT. It can be established using for the derivation a well-known argument based on the logic of common belief that Stalnaker (1998, 103-4, Stalnaker 2002 has provided, to the extent that we assume some independent triggering mechanism. 40 Alternatively, we could just put aside the project of justifying that a Stalnakerian speaker presupposition is after all met in these cases through the distinction between PT and AT, taking them simply as cases of indirect assertion. In any case, of course, this strategy for indirection is not going to work with contents that are controversial or surprising (e.g., the speaker utters a variation on (24) with 'my panther' replacing 'my cat'), which would invite instead the ''Hey, wait a minute!'' complaint, resisting accommodation.
This diagnosis of the case of informative presuppositions is thus close to Stalnaker's: (1) a presupposition is present; (2) the speaker is presupposing, in the way Stalnaker suggests, but (unlike in his account) by thereby becoming beholden to a norm constitutive of it; (3) there is nothing inappropriate in this, as there is nothing inappropriate in related cases of indirection. 41 As Karttunen (1974, 412) puts it, ''This is one way in which we communicate indirectly, convey matters without discussing them.' ' Stalnaker (1974, 51-2;cf. also 1973, 451) says something similar: ''In such a case, a speaker tells his auditor something in part by pretending that his auditor already knows it''. What I resist is his contention that such a ''pretending'' can be accounted for even if there is no ''mysterious relation X'' (Stalnaker 2002, 712-3)-no substantive kind conventionally indicated by some linguistic triggers, playing the crucial explanatory role in theoretically describing the intuitive data that he disparages.
There is no direct connection between a prescriptive account of a speech act, and the existence of conventional means for performing it. There are, I believe, indirect assertions, made without using conventional means for such a purpose-rhetorical questions, metaphorical assertions, assertions in fictions. 42 Our account of assertion should allow for that, and so does Williamson's normative account, which does not require any conventions for making assertions. The same applies to presupposing; speakers might presuppose without using conventional means for that, by means of indirect mechanisms like the one Grice described for conversational implicatures, or just reliance on the capacity for rational inference of their interlocutors (cf. Chierchia and McConnell-Ginet's 1990, 285 ff;Simons 2005, 150-1). The normative account allows for it. Thus, if a soccer fan tells another in the morning, out of the blue, ''you surely celebrated well into the night'', she indirectly conveys that she is presupposing that the second fan's team won the cup the previous night. On the just outlined account, an informative presupposition is also a pragmatically triggered one: the speaker uses a device that conventionally presupposes something in a context in which the conventionally triggered presupposition would by itself be unwarranted, in order to get the requirement to be after all satisfied at ''presupposition evaluation time''. This is thus a case of a speaker presupposition indirectly generated by a sentence presupposition.
However, it is natural to have conventional indicators that we are taking our behavior to be subject to specific rules. Perhaps the development of such conventional indicators is the only evolutionarily intelligible account for the existence of such a practice, given that, obviously, the norms envisaged here have not come to be in force by being explicitly codified. Hence, from the present point of view, it is to be expected that languages have specific resources to (by default) indicate specific acts of presupposing, which is what conventionalists suggest as a solution to the Triggering Problem, and what (SnP') envisages.
To sum up, in this section I have articulated a speech act account of presuppositions, on which these are constituted by common knowledge norms. I have offered an argument for such a normative account of presupposition, replicating the arguments that Williamson invokes for his normative accounts of assertion (KR 0 ). The proposal accounts on its own terms for some of the main features associated with presuppositions. It explains why the HWAM test tracks presuppositions. It explains how informative presuppositions are generated, invoking an independently justified model for indirect speech acts. Through the correlation between norms and conventional means to indicate liability to them, it supports conventionalist responses to the triggering problem. Finally, being a common ground account of presuppositions (like the Stalnakerian and DS accounts), it provides the sort of explanation of projection and their absence (under filters and plugs) that these accounts allow. In the next section, I'll briefly address views that cast doubt on common ground accounts.

Presuppositions Without Common Ground?
The proposal in the previous section shares the main idea in the Stalnakerian tradition, that presuppositions are requirements on the common ground. Abrusán (2011Abrusán ( , 2016, Simons et al. (2011), andTonhauser et al. (2013) suggest an alternative picture, on which presupposing is simply a matter of presenting some aspects of the total content of an utterance as background, not-at-issue. They do not contend that this is always a pragmatic matter, advancing semantic devices that contribute in some cases to its expression. As an appealing feature, presupposition triggering and projection is not left on these views as a brute lexical fact, but it is explained by a mechanism that distinguishes aspects of the full contribution of words and constructions. They show that linguistic presupposition depends in complex ways on mechanisms to indicate what is and is not at-issue, such as focus, questions-under-discussion, and other matters. But I don't think their results invalidate the distinction that my proposal assumes among types of not-at-issue content, including at least Potts' (2007) conventional implicatures and presuppositions, still less my account of the latter. Simons et al. (2011) and Tonhauser et al. (2013) appeal to empirical data that some presuppositions associated with referential expressions resist accommodation. On that basis, they provide a finer-grained taxonomy of projective contents than the one discussed in the first section, which only assumed a distinction between presuppositions and conventional implicatures. Simons et al. (2017) in fact argue for a purely pragmatic triggering for the presuppositions of factives, subtler than previous attempts by Stalnaker (1974) and others, and consistent with the empirical data studied by Spenader (2003) that suggests that such presuppositions are frequently accommodated. Now, as I have been insisting, the point that there are differences among triggers of not-at-issue contents doesn't contradict my main claims (cf. Simons 2013, 164). I also feel unconvinced by some of their arguments, and I think it might be useful to discuss them in this concluding section. 43 Tonhauser et al. (2013, 76) introduce a condition they call strong contextual felicity constraint, SCFC; utterances including triggers that satisfy it are only acceptable in contexts that entail their projective content, while those that don't are also acceptable in contexts that do not entail it. They use this condition to distinguish triggers such as definite descriptions like 'my dog', factives like 'know' and change-of-state verbs like 'stop', allegedly not meeting the constraint, from others that allegedly do, ''anaphoric'' triggers such as demonstratives and adverbs like 'too'. They provide data from Guaraní and from English, gathered using refined methodological procedures. But I'll try to show that it is unclear that SCFC underwrites a theoretically solid distinction.
The cases Tonhauser et al. provide showing that expressions fail to satisfy SCFC are, from my perspective, just instances of accommodation, along the lines of the 'my cat' example (24) above. They do establish that speakers of Guaraní, exactly like speakers of English, are happy to count as appropriate utterances including those triggers made in contexts that, previous to the utterance, did not include the relevant information. Now, Tonhauser et al. (2013, 80) dismiss appeals to accommodation, referring to arguments in their previous work (Simons et al. 2011) where, as far as I can tell, the only argument that they provide is that ''in the paradigm case of common ground constraint, accommodation is ruled out'' (313). Paradigm cases are anaphoric triggers like those already mentioned; they give an example involving 'she', uttered in a context in which no female is salient, and in the more recent piece they investigate such cases with their sophisticated crosslinguistic methodology. Thus, they point out that an utterance of 'our driver is eating empanadas, too' (or its Guaraní equivalent) uttered by a woman who just sat in a bus next to another she doesn't know, who is eating a hamburger, would be rejected as inappropriate, while it would be appropriate if the woman already at the bus was instead eating empanadas. They provide similar results with demonstratives such as 'that little boy', produced respectively in contexts with and without a salient little boy (Tonhauser et al. 2013, 78-9). 44 But I think we should be wary of the theoretical significance of a distinction between triggers based on SCFC. Notice that for a trigger to satisfy the constraint, it is not enough that we find one context that does not entail the presupposition of a clause in which an utterance including it feels inappropriate. Otherwise, all triggers would meet SCFC: just imagine (24) uttered in a context in which a no-pets policy has been severely enforced for years, and nobody is expected to have cats. Satisfying SCFC requires resistance to accommodation in all contexts. But this is a tall order: it is not so difficult to produce contexts not entailing the projective contents of utterances of ''anaphoric'' triggers in which speakers (also of Guaraní, I bet) would find the utterances perfectly appropriate. Imagine now that the bus in the previous example is running in a country where it is forbidden eating hamburgers (a symbol of capitalism!), or even referring to them. The locals, however, adore the food, and are ingenious enough to manage to systematically violate the prohibition, conning the informers that usually infest buses and other public places. The woman that just got on the bus now makes the same utterance as before to the woman next to her, who is watchfully eating her hamburger, while winking at her -a case rather like Donnellan's 'the king in his counting house' discussed above. It is easy to find similar examples in which uttering 'she' or 'that little boy' where only a male or a senior person is salient is felt to be perfectly appropriate.
It could be replied that these cases are ''pragmatic''. But on the sort of view like the present one that they are objecting to, this applies to accommodation in general: it is a pragmatic strategy. For all we can tell, the triggers we are discussing conventionally carry presuppositions; these conventionally conveyed presuppositions are present in all contexts, because they are semantic (in the relevant sense). In some cases, before making the utterances, the common ground does not satisfy 44 Abbott (2015, Sect. 3.1) accepts these results, subscribing to the significance of SCFC. those presuppositions. In some of these cases, speakers can count on their audiences to have conveniently come to accept them at presupposition evaluation time, PT. But this involves pragmatic processes; in some cases the needed resources will not be in place, and accommodation will fail (von Fintel 2008, 153-4). This suggests that SCFC at least requires a more refined formulation. Tonhauser et al. (2013, 80) conclude ''the strong contextual felicity diagnostic distinguishes implications that impose constraints on the utterance context from ones that do not''. I am not convinced. Semantically, the two sorts of triggers they distinguish appear to impose the normative constraints with which they are conventionally associated. As formulated, SCFC appears not to capture any devices, because we can only distinguish between contexts lacking needed pragmatic resources for accommodation, from those in which they are in place; as we have seen, a little imagination suffices to find such contexts for all triggers they discuss. I am not at all denying that there might be a significant classification in the vicinity, based on the difficulties that some triggers pose to accommodation 45 ; but it should be formulated in more complex ways, and hence it remains an open question what theoretically interest it might have for a semantically significant taxonomy of triggers vis-à-vis their projective behavior. In any case, even if there turns out to be one, that seems compatible with the general idea that there are linguistic presuppositions understood as imposing constraints on a common ground, and the specific account provided here.

Closing: Presupposing and Asserting
I will conclude with a brief indication of the consequences of the present account for our views concerning the nature of assertion. As Williamson (op. cit., notes, an account based on the weaker rules (TR) or (RBR) seems at first sight preferable: given that those rules are satisfied whenever the knowledge rule is, but not the other way around, it provides for a practice with fewer violations of its governing rule; some knowledge rule could then perhaps be explained as derived from TR, together with considerations not specific to assertion. However, those rules, he argues (ibid., 244-5), do not individuate assertion; alternative speech acts like conjecturing (which requires an evidential standard weaker than assertion) or swearing (which requires one stronger) also involve one or the other. Now we can see how the correctness of the account we have given for presupposing puts strain on Williamson's account of assertion: it suggests that (KR) runs into the same problem that he poses for (TR) and (RBR); to wit, that they do not properly individuate assertion. There is a difference between the rule we have invoked in order to characterize presupposing, and the knowledge rule: presupposing requires common or shared knowledge, while given (KR), asserting merely requires knowledge. But is this enough to distinguish them? I do not think so. If (KR) truly individuates assertion, and (PR) presupposition, then, if in a given case presupposing p is correct, asserting it would not be incorrect in that case-indeed, it would be correct to do so if we assume the stronger (KR 0 ), which provides a sufficient condition and not merely a necessary one for correctness. This is, I think, clearly not acceptable: when it is correct to presuppose p, it is incorrect to assert it. 46 (Putting of course aside non-default cases, such as small talk or ''making conversation'': ''nice weather, isn't?'') Intuitively, it is relatively clear why Williamson's account is shown to be inaccurate, given an account of presupposition like the one I have advanced: assertion (unlike judgment) is in central cases a communicative act, whose function is to put audiences in a position to gain knowledge that speakers are in a position to impart. This is in fact, ultimately, the contrast that the HWAM test brings out; the complaint is appropriate for presupposed contents, because the audience is presented as knowing them already; it is inappropriate for asserted contents, because they are presented as having been put forward precisely with the goal of the audience coming to know them. The rules (KR), (TR) and (RBR) miss this communicative aspect of assertion; as I indicated at the beginning, they are, I think, good candidates for characterizing the internal act of judging, but not the communicative act of asserting. This is what the contrast we have been examining here with presupposing brings glaringly to the fore. I have made a proposal elsewhere (García-Carpintero 2004, forthcoming) to capture the communicative nature of assertion in a normative framework by means of (KPR) instead; Pelling (2013) and Hinchman (2013) offer other considerations for a related view.
To sum up, in this paper I have provided a new account of linguistic presuppositions, on which they are ancillary speech acts defined by constitutive norms. After rehearsing the intuitive data characterizing the phenomenon, I have articulated the normative speech act account of presupposition, and I have argued for it in parallel with Williamson's considerations abductive for his analogous account of assertion. I have explained how the account deals with the problem of informative presuppositions, and how it relates to accounts for the Triggering and Projection Problems for presuppositions. I have concluded with a brief discussion of how the proposal suggests adding a communicative twist to Williamson's knowledge-based account of assertion.