Korean Honorifics and the Truth: Re-evaluating Grice's Pragmatics — Epoche B2
Not At Issue, But True or False: Korean Honorifics Against the Gricean Divide A Korean sentence cannot be uttered without settling how the speaker stands to the person addressed [1] . The verb ending does it, and there is no neutral option. The philosophical question is what that ending contributes. The standard answer, drawn from Grice [2] , is that it contributes nothing to truth conditions: the report is finished or it is not, whatever the ending. This essay argues that the standard answer confuses two things that Grice himself kept apart. Honorific morphology contributes content that is conventional, entailed and evaluable as correct or incorrect. What it lacks is not truth-evaluability but at-issueness — the standing to be the thing a reply denies. Once the two are distinguished, the claim that honorifics touch truth conditions is not extravagant; it is what the diagnostics deliver. The Gricean picture, and one thing it does not say In 'Logic and Conversation' (1975) Grice separates what a speaker says from what she implicates. What is said is the conventional meaning of the words, once ambiguity and reference are settled, and it is the bearer of truth conditions. What is implicated is everything else the speaker manages to get across. Conversational implicatures arise from the assumption that the speaker is cooperating and observing maxims of quality, quantity, relation and manner; they are calculable from the context and they can be cancelled, since a speaker may add a clause withdrawing them without contradicting herself. Grice also marks off a second kind, which is where the present question lives: In some cases the conventional meaning of the words used will determine what is implicated, besides helping to determine what is said. ('Logic and Conversation', in Cole and Morgan, 1975) The example he gives there is 'therefore'; 'but' is the other standard case, inherited from Frege. What such words add is fixed by the words themselves rather than worked out from context, and it cannot be cancelled. These are conventional implicatures. It is usually said that conventional implicatures are non-truth-conditional. That claim does not follow from anything in the passage, and it is worth seeing why. Grice's point is that this content is not part of what is said — not part of the proposition the speaker puts forward for assent or denial. Whether it is nonetheless entailed, and so capable of being false, is a separate question that his terminology does not settle. Christopher Potts's The Logic of Conventional Implicature (2005 [3] ) argues that it is: conventional implicatures are entailments, contributed on a second dimension, and an utterance can be right on one dimension and wrong on the other. Kent Bach reaches the same conclusion by a shorter route in 'The Myth of Conventional Implicature' (1999 [4] ): there is no such category, he argues, because what 'but' contributes is simply part of what is said, carried by a second proposition the sentence expresses. The two disagree about the architecture and agree about the point that matters here — the content is truth-evaluable. The Korean system Korean honorification is not a matter of vocabulary choice at the edges. It is grammatical, obligatory and organised into three subsystems, described in standard reference works such as Sohn Ho-min's The Korean Language (1999). Addressee honorification is carried by the speech level, marked on the verb ending. The polite informal 해요 style ends in -요, and the deferential formal style ends in -습니다 or -ㅂ니다. A junior employee reports to a manager in the second; the same employee tells a colleague the same news in the first. Subject honorification is carried by the pre-final suffix -시-, together with the honorific subject particle 께서 and a set of suppletive verbs — 계시다 for 'to be', 잡수시다 for 'to eat', 주무시다 for 'to sleep'. This marks the standing not of the addressee but of whoever the sentence is about. Object honorification humbles the speaker with respect to a third party: 드리다 rather than 주다 for giving, 뵙다 rather than 보다 for seeing. What matters philosophically is that these are three coordinated ways of encoding one kind of information — the position of a person in a social ordering relative to the speaker — and that the third-person cases make that information externally checkable in a way the addressee case does not. Why the content is evaluable Take subject honorification first, because it is the clearest. Suppose I say of my own small son that he 주무신다 — that he is sleeping, with the honorific suffix. Korean speakers do not report this as impolite; they report it as wrong. The same verdict falls on the widely criticised service-industry usage in which a coffee is said to have arrived with -시- attached, treating an inanimate object as a person entitled to deference. If the honorific were purely a matter of manners, over-politeness would be at worst excessive. It is instead treated as an error, and the natural description of the error is that the sentence presents its subject as occupying a position he does not occupy. Now the two diagnostics that separate this content from ordinary implicature. It cannot be cancelled. A conversational implicature can always be withdrawn: 'Some of the students passed — in fact all of them did' is not a contradiction. Nothing corresponding is available here. A speaker cannot use the deferential ending and then add that no deference is intended. The addition does not withdraw the content; it produces incoherence, which is the mark of something entailed rather than suggested. It projects. Put an honorific sentence under a conditional, a question, or a report of someone else's belief, and the social content does not go under with it. If I ask whether the manager 오셨습니까 — whether he has arrived — the deference is still mine, asserted now, and it is not part of what I am asking about. This is the behaviour Potts identifies as characteristic of conventional-implicature conten