Regressions in Pragmatics (and Semantics)
Kent Bach
全能for Advances in Pragmatics
Noël Burton-Roberts, editor
Abstract
Influenced by the Wittgensteinian slogan “Don’t look for the meaning, look for the u,” ordinary language philosophers aimed to defu various philosophical problems by analyzing key words in terms of what they are ud to do or the conditions for appropriately using them. Although Moore, Grice and Searle expod this error – mixing pragmatics with mantics – it still gets committed, now to a different end. Nowadays the aim is to reckon with the fact that the meanings of a great many ntences underdetermine what we would normally mean in using them – even if the ntence is free of indexicality, ambiguity, and vagueness. This can be so becau the ntence express a “minimal” proposition or even becau it doesn’t fully express any proposition. Many theorists are led to defend “truth-conditional pragmatics” (or linguistic “contextualism”), to find a hidden indexical in every syntactic
nook or mantic cranny, or otherwi to pay undue respect to emingly mantic intuitions and intentions. This paper identifies various such moves and explains what’s regressive about them.
Regressions in Pragmatics (and Semantics)
Kent Bach孩子学
Remember the 20th century? Around the middle of it, so-called ordinary language philosophers made extraordinary claims about various philosophically interesting terms. Evidently they were operating under the influence – of Wittgenstein, that is – and his slogan “Don’t look for the meaning, look for the u.” In ethics, for example, it was (and sometimes still is) suppod that becau ntences containing words like good and wrong are ud to express affective attitudes, such as approval or disapproval (or, alternatively, to perform speech acts like commending and condemning), such ntences are not ud to make statements, hence that questions of value and morals are not matters of fact. This line of argument is fallacious. As G.E. Moore pointed out, although one express approval (or disapproval) by making a value judgment, it is the act of making the judgment, not the content of the judgment, that implies that one approves (1942: 540-45). Sentences ud for moral evaluation, such as ‘Gambling is wrong’ and ‘Greed is good’, are no different in form f
rom other declarative ntences, which, whatever the status of their contents, are standardly ud to make statements.1
The fallacious line of argument expod by Moore commits what John Searle called the “speech act fallacy” (1969: 136-141). Searle gives further examples, each involving a speech act analysis of a philosophically important word. According to such analys, the terms true, know, and probably do not express properties. Rather, becau true is ud to endor statements (Strawson), know to give guarantees (Austin), and probably to qualify commitments (Toulmin), tho us constitute the meaning of the words. In each ca the mistake is the same: identifying what the word is typically ud to do with its mantic content.
Searle also expos the “asrtion fallacy” (1969: 141-46), which confus conditions of making an asrtion with what is asrted. Here are two examples. You wouldn’t asrt that you believe something if you were prepared to asrt that you know it, so knowing does not entail believing. Similarly, you would not describe a person as trying to do something that involves no effort or difficulty, so trying entails effort or difficulty. Paul Grice (1961) had already identified the same fallacy in a similar argument, due to Austin, about words like ems, appears, and looks. Becau you would not say that a table looks old unless you doubted or were even prepared to deny that it is old,
the proposition that the table looks old entails that its being old is doubted or denied. This argument is clearly fallacious, since it draws a conclusion about ntence mantics from a premi about conditions on appropriate asrtion. You can misleadingly implicate something without its being entailed by what you say.
1 This leaves open the possibility that there is something fundamentally problematic about their contents. Perhaps such statements are factually defective and, despite syntactic appearances, are neither true nor fal. But this is a metaphysical, not mantic, issue about the status of the properties ethical predicates purport to express.宝宝身高标准
The claims from ordinary language philosophy were discredited decades ago. So why do I dredge them up now? Becau esntially the same mistakes keep getting made. There continues to prevail an illicit mixing of pragmatics with mantics. Yes, people are no longer rving up misguided analys of philosophically interesting expressions. Now they have a different concern: to reckon with the fact, appreciated only recently, that the meanings of a great many ntences, at least tho we are at all likely to u, generally underdetermine what we would normally mean in uttering them. This can happen even if the ntence in question is free of indexicality, ambiguity, and vagueness, and even if the speaker is using all of its constituents literally. In other words, what a speaker means
in uttering a ntence, even without speaking figuratively or obliquely, is likely to be an enriched version of what could be predicted from the meaning of the ntence alone. This can be becau the ntence express a “minimal” (or “skeletal”) proposition or even becau it fails to express a complete proposition at all.
Many theorists who appreciate the fact that ntence meaning underdetermines speaker meaning grant that mantics concerns ntences (and their constituents) and that pragmatics concerns acts of uttering them. But they will then go ahead and conflate them anyway. They just don’t em to appreciate what makes the pragmatic pragmatic and, in some cas, what makes the mantic mantic either. As a result, some theorists have been led to defend some form of “contextualism” or “truth-conditional pragmatics” and others to propo inflated conceptions of mantics. I will not attempt to examine specific theories in any detail but will instead identify a ries of ideas that are central to one or another of them. Each idea, I will suggest, in one way or another conflates the mantic and the pragmatic. Before taking up the ideas, I will first (§I) state my view on what makes the pragmatic pragmatic and then (§II) give a sample of ntences who u typically does not make fully explicit what one would mean in uttering them. Then (§III) I will identify nine suspect ideas underlying different ways of trying to account for the fact that such ntences are typically u
d in enriched ways. I will suggest that each of the ideas needlessly conflates something pragmatic with something mantic.
I. What makes the pragmatic pragmatic?岭南春早
校园趣事300字
A speaker can convey a thought without putting it into words. He can say one thing (as determined by ntence meaning, perhaps relative to context) and mean something el (speaker meaning). In order to communicate something to someone, the speaker has to come up with a ntence who utterance makes evident, even if the ntence itlf does not express, what it is that he intends to convey. The hearer’s task is to understand the speaker or, more precily, to recognize the speaker’s communicative intention in making the utterance and, in particular, to identify what the speaker means. The meaning of the ntence provides the hearer with only part of his basis for figuring that out. The hearer needs also to take into account the fact that in that situation the speaker uttered that ntence with that meaning.
延时自拍The very fact that a ntence is uttered gives ri to distinctively pragmatic facts. As Grice (1961) obrved, it is the fact that a speaker utters a ntence with a certain mantic content (or even that ntence rather than another with the same content) that generates what he would later call a conve
rsational implicature. His first example of this was an utterance of ‘Jones has beautiful handwriting and his English is grammatical’, made as an evaluation of Jones’s philosophical ability, to implicate that Jones is no good at philosophy. A different sort of illustration is provided by Moore’s paradox (so-called). If you say, “Snow is white, but I don’t believe it,” you are denying that you believe something you have just asrted. The contradiction here is pragmatic. That snow is white does not entail your believing it (nor vice versa), and there’s no contradiction in my saying, “Snow is white, but you don’t believe it.” The
inconsistency aris not from what you are asrting but from the fact that you are asrting it. That’s what makes it a pragmatic contradiction.
By way of asking what would we mean in uttering ntences containing such terms as good, true, try, and appears, ordinary language philosophers came up with pragmatics-laden accounts of their meanings. They tried to supplant ntences’ truth-conditions with conditions for their appropriate (especially non-misleading) u and to equate what a speaker does in uttering a ntence with the mantic content of the ntence itlf. However, a ntence has its content independently of being uttered.2 Understanding it is of a piece with hearing it and parsing it. A ntence’s mantic content is a projection of its syntactic structure, as a function of the mantic contents of its constituents, and
is something a competent hearer grasps by virtue of understanding the language. The speaker’s act of uttering that ntence, with that content, provides or invokes additional information the hearer is to u in understanding an utterance of the ntence – there is no intermediate level of meaning between the mantic content of the ntence and the speaker’s communicative intention in uttering it (e §III.4 below).
校园散文As illustrated by generalized conversational implicatures (Grice 1975/1989: 37-8), there are regularities of u that, despite being systematic, should not be confud with linguistic meanings. Such pragmatic regularities can be explained by combining facts about the mantic contents of ntences people utter with generalizations about people’s acts of uttering them. To explain the regularities we do not have to resort to fanciful stories about the meanings of expressions bad on obrvations about the conditions for their typical or appropriate u. Instead, we can apply what Robert Stalnaker has aptly described as “the classic Gricean strategy: to try to u simple truisms about conversation or discour to explain regularities that em complex and unmotivated when they are assumed to be facts about the mantics of the relevant expressions”(1999: 8).
Executing this strategy requires taking into account three key elements of communication. First, there is the distinctive nature of a communicative intention. As Grice discovered, such an intention is
“reflexive”: the speaker intends his utterance “to produce some effect in an audience by means of the recognition of this intention” (Grice (1957/1989: 220). Specifically, although Grice did not identify it as such, this effect is recognition of the attitude the speaker is expressing (whether or not he actually has that attitude is another matter).So,an act of communication succeeds by way of recognition of the very intention with which it is performed. Compare this with, for example, the act of getting an audience to laugh, say by telling a joke. They can recognize your intention to get them to laugh without actually laughing. Normally that would require that they find your joke funny (of cour, they could find it funny that you are trying to get them to laugh, but that’s not what you intend them to laugh at). In contrast, all it takes for an act of communication to succeed is for the audience to recognize the speaker’s intention in performing it. In other words, a communicative intention has this distinctive property: its fulfillment consists in its recognition (by the audience). A communicative intention includes, as part of its content, that the audience recognize this very intention by taking into account the fact that they are intended to recognize it. Correlatively, in figuring out what the speaker’s communicative intention is, the hearer takes into account, at least implicitly, that he is intended to it figure out.3 To understand an utterance is to recognize the intention with which it is made.
2 Of cour, in denying that meaning is u, I am not denying the platitude that linguistic meaning is ultimately grounded in u. I do deny that individual us (other than effective stipulative definitions) endow expressions with new meanings.
3 Grice anticipates that this may em “to involve a reflexive paradox,” but insists that “it does not really do so” (1957/1989: 219). The audience does not have to already know what the speaker’s intention is in order to figure what it is, but merely that he is intended to figure this out.
Second, there is the fact that the speaker said what he said rather than something el. Perhaps he could have said something more informative, more relevant, or more appropriate. So the fact that he said what he said contributes to the explanation of why he said it, hence to the recognition of the intention with which he said it. Perhaps he doesn’t know more, perhaps he doesn’t want you to know more, perhaps it’s obvious what he could have added. Also, the fact that he said what he said in the way he said it, using tho words and with that intonation, rather than in some other way, may contribute further to the explanation, as with Grice’s manner implicatures (1975/1989: 35-37).
Third, although what the speaker says, the mantic content of the words he utters, provides the primary input to the audience’s inference, the audience also takes into account what is looly called
“context.”This is the mutually salient contextual information that the audience is intended to u to ascertain the speaker’s communicative intention, partly on the basis that they are intended to do so. Here, it must be stresd, context does not literally determine, in the n of constituting, what the speaker means. Rather, at least when communication succeeds, it provides the audience with the basis for determining, in the n of ascertaining, what the speaker means. What the speaker does mean is a matter of his communicative intention, not context, although what he could reasonably mean depends on what information is mutually salient.
Taking such information into account goes beyond mantics, for what a speaker means need not be the same as what the uttered ntence means. It is generally though not universally acknowledged that explaining how a speaker can say one thing and manage to convey something el requires something like Grice’s theory of conversational implicature, according to which the hearer relies on certain maxims, or presumptions (Bach and Harnish 1979: 62-65), to figure out what the speaker means. However, it is commonly overlooked that the maxims or presumptions are operative even when the speaker means exactly what he says. They don’t kick in just when something is implicated.4 After all, it is not part of the meaning of a ntence that it must be ud literally, strictly in accordance with its mantic content. Accordingly, it is a mistake to suppo that “
猪脚煲汤pragmatic content is what the speaker communicates over and above the mantic content of the ntence” (King and Stanley 2005: 117). Pragmatics doesn’t just fill the gap between mantic and conveyed content. It operates even when there is no gap. So it is misleading to speak of the border or, the so-called “interface” between mantics and pragmatics. This mistakenly suggests that pragmatics somehow takes over when mantics leaves off. It is one thing for a ntence to have the content that it has and another thing for a speech act of uttering the ntence to have the content it has. Even when the content of the speech act is the same as that of the ntence, that is a pragmatic fact, something that the speaker has to intend and the hearer has to figure out.
There are various ways in which what a speaker means can go beyond or otherwi be distinct from the mantic content of the ntence he utters. One familiar ca is speaking in a nonliteral way, by using metaphor, irony, metonymy, or some other figure of speech, whereby one says one thing and means something el. For example, ‘You are the ribbon around my life’and ‘That was the cleverest metaphor I’ve ever heard’ are ntences likely ud to convey something different from what their mantics predicts. In the other familiar ca, of conversational implicature (or indirection generally), the speaker typically means not only what he says but also something el, as in Grice’s example, “Jones has beautiful handwriting and. his English is grammatical.”
4 So it is not a “standard Gricean assumption … that any material derived via conversational principles constitutes an implicature” (Carston 2002: 100). For one thing, the maxims come into play in resolving ambiguities and determining references. Indeed, they can even bear on figuring out what the speaker uttered, as when one doesn’t hear the utterance clearly.
Less familiar but no less common is what I call conversational “impliciture” (Bach 1994), where what the speaker means is not made fully explicit and is an enrichment of what he says. This can occur either becau the ntence he utters express a “minimal” proposition or becau it is mantically incomplete, expressing no proposition at all, even relative to the context. This phenomenon is widely thought to undermine any dichotomous conception of the mantic/pragmatic distinction or at least to po special challenges in accounting for the relationship between what a speaker says and what he means in saying it. Such worries are illustrated by the nine different strategies to be discusd later for treating cas of impliciture. In my view, however, the doubts are unwarranted, and the strategies they have inspired actually confu meaning and u. After giving (§II) an assortment of examples of impliciture, I will explain (§III) why I regard the nine strategies as regressions in pragmatics.
II. Examples
It is generally recognized that most ntences people utter in everyday life have mantic contents that are either too variable or too skimpy to compri what people mean in uttering them, even when all of their constituents are ud literally. Indexicals who mantic content are a function of context are the source of variable mantic contents, but here I’ll focus on ntences who mantic contents are too skimpy. The mantic content of a ntence can be too skimpy, relative to a speaker’s likely communicative purpos in uttering it, either becau the proposition it express lacks elements that are part of what the speaker means or becau what it express, its mantic content, falls short of comprising a proposition (presumably the things people mean in uttering ntences are propositions). Sentences of the first sort express so-called “minimal”propositions, and ntences of the cond sort are said to be mantically incomplete.5 When a speaker utters a ntence with minimal propositional content, what he means is an expansion of that. When a speaker utters a mantically incomplete ntence what he means is a completion of its incomplete propositional content.
The first t of examples to follow contains ntences that express minimal propositions.6 In each ca, what the speaker is likely to mean is expressible by an expanded version of the ntence he utters, perhaps one containing the italicized material in brackets.
Implicit quantifier restriction
Everyone [in my family] went to the wedding.
Lola had nothing [appropriate] to wear.
The cupboard [in this hou] is bare.
Only Bill [among tho prent] knows the answer.
I have always [since adulthood] liked spinach.
Implicit qualification
I will be there [at the agreed time].
I haven’t had a coffee break [this morning].
5 The term ‘minimal proposition’ was introduced by François Recanati (1989: 304), but the term has been ud recently by Cappelen and Lepore (2005) slightly differently, for what they take to be the propositions expresd by ntences that almost everyone el takes to be mantically incomplete.
6To keep matters relatively simple, I will limit the discussion to declarative ntences, the ones that are generally assumed to express propositions.
