Ad hoc concepts paper with Mark Textor

It has become a staple of truth-conditional pragmatics to assume that the meaning contributed by the use of a word to the proposition expressed by a speaker in making an utterance need not be the fixed lexical meaning of that word. For example, in saying:

John’s a saint.

I might be talking about my neighbour, John, who has not been canonised, and be expressing a thought that might be partially glossed thus: John is extremely kind – to the point of self-sacrifice. According to truth-conditional pragmatic theories, when I do this I am predicating of John an ad hoc concept SAINT*, different from the lexicalised concept.

Evidence that this ad hoc concept gets into the proposition expressed comes from negation. Consider replying like this:

No, he’s not a saint. He just does what he would prefer and makes it look like a big sacrifice.

The speaker of the reply seems to be denying that John is a helpful etc. individual, not that he is a literal saint. It looks, in other words, as though the negation takes scope over  the proposition that John is a SAINT*, and this suggests that the proposition that was expressed by the original claim was that John is a SAINT*.

In the new paper we take all of this for granted and look in some detail at what kind of thing such an ‘ad hoc concept’ is (or would be).

We argue that ad hoc concepts are not really concepts; rather they are clusters of information ready to be used in inference. By cluster, we mean a collection which, in addition to having definite members and non-members, may have some borderline member/non-members, like a cluster of points on a graph. See below for the abstract: the paper is here.

Allott, N. & Textor, M. (2012). Lexical pragmatic adjustment and ad hoc concepts. International Review of Pragmatics, 4(2), 185–208

It’s in a special issue of the International Review of Pragmatics edited by Anton Benz, Katja Jasinskaja and Uli Sauerland. Many thanks to them for putting this together – also for organising the conference that most of the papers were from.

Thanks are also due to CSMN and the Norwegian Research Council for funding my research, and for making possible an extended visit by Mark Textor to Oslo last year for us to write the paper.

--


Lexical pragmatic adjustment and ad hoc concepts
Nicholas Allott and Mark Textor

Abstract

According to truth-conditional pragmatics, a word may contribute an ad hoc concept to the proposition expressed, that is, something that differs from the concept the word encodes (the lexicalized concept). In relevance-theoretic lexical pragmatics, ad hoc concepts are treated like a species of concepts proper. Concepts as well as ad hoc concepts are taken to be atomic. Lexical pragmatic adjustment is conceived as the formation of an ad hoc concept that is narrower or broader in extension (or both) than the lexicalized concept involved. We argue that difference in extension should not be taken as the crucial feature of lexical pragmatics, since ad hoc concepts can have the same extension as the lexicalized concept. In contrast, we propose a positive view of ad hoc concepts as clusters of information poised to be used in inference. (Surprisingly, ad hoc concepts turn out not to be concepts at all.) The cluster account drops the assumption that ad hoc concepts are atomic and can therefore provide a satisfactory explanation of lexical pragmatic adjustment.

Looks like I might be one of those LINGUIST linguists after all

... because I may have come up with a property of contrastive reduplication that isn't already in the literature.

Here's an example of contrastive reduplication:

I don't drink that herbal stuff. Haven't you got some TEA tea?
(where the caps on TEA indicate stress.)

There are several labels for this in the literature: contrastive reduplication (Ghomeshi et al. 2004), lexical cloning (Horn 2006) and identical constituent compounding (Hohenhaus 2004).

I noticed the phenomenon in my own speech a few years ago (I found myself saying 'TEA tea') – but hadn't thought about it much until I saw an excellent talk by Ewa Waƚaszewska last week at the relevance theory conference in Poland. (The references to the literature are from her handout.)

The conveyed meaning is a stereotypical or prototypical category based on the lexical category: e.g. 'tea' covers many different types of infusion, but 'TEA tea' (as used above) only covers infusions made with material from the tea plant (camellia sinensis). More about meaning in a minute.

Contrastive reduplication (CR) is not just repetition of a word. As Ewa pointed out, the prosody is different (more equal stress, and a pause before the repetition of the word) in e.g.:

Give me something to drink, but not coffee. I want tea, tea!

Also what is conveyed is quite different. There is no necessary narrowing of the category here, just emphasis, and this is typical for repetition.

It occurred to me that another test is what happens with plurals. My intuition is that this is good:

Those people wondering around Warsaw are actually academics here for a conference. They aren't TOURIST tourists.

And this is bad:

Those people wondering around Warsaw are actually academics here for a conference. They aren't *TOURISTS tourists.

Whereas for the first word in repetition the plural is correct and the singular is impossible:

Warsaw is so busy these days. The centre is crawling with tourists, tourists!

Warsaw is so busy these days. The centre is crawling with *tourist, tourists! 

I assumed that this would be old hat, but Ewa and Diane Blakemore (who has a student currently working on CR) both said that they didn't think it was in the literature. So perhaps I am a LINGUIST linguist. (Despite working almost exclusively on stuff outside the language faculty, I mean.)

On the meaning of CR: I said above that it induces prototypical 'narrowing'*. If I understood (and remember) Ewa's talk properly, though, she was presenting (not necessarily endorsing) an argument that this is not right because the narrowings can be to different, sometimes largely disjoint, categories. e.g. LINGUIST linguist might on one occasion be narrowed to linguist concerned with language proper (rather than, say pragmatics), on another to linguist in the generativist school, and on another, perhaps to person who knows and can use many languages (i.e. polyglot).

I'm not sure if Ewa endorses this argument, but it seems to miss the mark. We know that prototypes are context-sensitive (since the work of the psychologist Laurence Barsalou). But then the fact that CR can narrow to different categories is to be expected, just as long as the result is always communicated to be prototypical. And as far as I can see, it is. TEA tea can be used as above, or it can be used as I found myself using it, to mean (roughly) British-style blended 'breakfast' tea, crucially served with milk (i.e. prototypical tea for the average Brit).

CR means something like this: starting from the lexical category, find a more specific category, which the speaker is presenting as prototypical.

Anyway, I'm not claiming that this analysis of the (procedural) meaning of CR is novel. It may be that something along these lines is what Ewa had in mind.

---
*The scare quotes on 'narrowing' are because I think that in ad hoc concept formation the correct formulation is to talk about clusters, not extensions.
See this post and our paper on the subject: Allott, N. & Textor, M. (2012). Lexical pragmatic adjustment and ad hoc concepts. International Review of Pragmatics, 4(2), 185–208
(I've edited this note to link to the new post and the paper now that the paper is out).

The cooperative listener?

In a comment on this post on Language Log, Spell Me Jeff wrote:

I've been doing some admittedly shallow searching on topics like Grice and cooperative principle, and all I'm finding focuses on the role of the speaker. Surely there is work describing a "cooperative listener"?

I think that there are several areas of research that may be relevant, including work on:

1. what expectations hearers have,
2. whether the Cooperative Principle is real/operative,
3. charity in interpretation (perhaps the hearer-side counterpart of cooperation), and
4. feedback that hearers provide to the speaker (also a candidate for being the hearer-side counterpart of cooperation).

There has been a lot of work on what expectations hearers have (and what they are justified in having). For example neo-Griceans such as Larry Horn and Stephen Levinson try to boil Grice's maxims down to two or three. According to Levinson, each of them can be expressed as a guideline for the speaker and a rule of thumb for the hearer.
e.g. Speaker side: Say as much as you can.
Hearer side: What isn't said, isn't. (Levinson 2000, p. 31)

A classic paper on hearer expectations (and speaker maxims) is Wilson and Sperber 2002, which argues that there is no maxim of truthfulness and, correspondingly, hearers do not expect speakers to say only things that are literally true:

"hearers expect to be provided with true information. But there is an infinite supply of true information which is not worth attending to. Actual expectations are of relevant information, which (because it is information) is also true. However, we have argued that there just is no expectation that the true information communicated by an utterance should be literally or conventionally expressed..." (pp. 627-8)

On the Cooperative Principle more specifically, a number of theorists have argued that talk-exchanges need not be cooperative, and that the fundamental explanation for speakers tailoring their utterances to their audience is simply that they want to be understood (Kasher 1976, Sperber and Wilson 1986, inter alia; but see Grice's reply to some of these criticisms: 1989, around p. 369). If this is right then hearers can't generally expect cooperation, but they can expect that utterances will have been sufficiently tailored for them to be understandable and worth processing. Relevance theory can be seen as an attempt to spell out what this amounts to, and how it could be enough for communication to work:

The goal of inferential pragmatics is to explain how the hearer infers the speaker’s meaning on the basis of the evidence provided. The relevance-theoretic account is based on another of Grice’s central claims: that utterances automatically create expectations which guide the hearer towards the speaker’s meaning. Grice described these expectations in terms of a Co-operative Principle and maxims of Quality (truthfulness), Quantity (informativeness), Relation (relevance) and Manner (clarity) which speakers are expected to observe (Grice 1961; 1989: 368-72): the interpretation a rational hearer should choose is the one that best satisfies those expectations. Relevance theorists share Grice’s intuition that utterances raise expectations of relevance, but question several other aspects of his account, including the need for a Co-operative Principle and maxims, … The central claim of relevance theory is that the expectations of relevance raised by an utterance are precise enough, and predictable enough, to guide the hearer towards the speaker’s meaning. (Wilson and Sperber, 2004, p. 607)

An interesting question is what the hearer counterpart of cooperation is. It may be charitable interpretation, as perhaps manifested in interpretations of misspeakings, such as taking someone who says 'I was feeding the penguins in the park' to have been expressing the proposition (or trying to) that she was feeding the pigeons.

There is a huge literature on the 'principle of charity' in the philosophy of language, most famously in the work of Quine and Davidson, e.g.: "Assertions startingly false on the face of them are likely to turn on hidden differences of languages." (Quine, 1960, p. 59) Note that this principle, if applied rigorously (and perhaps simplemindedly), gets the pigeon/penguin example wrong and makes a mess of irony e.g. my saying 'It's beautiful weather again' in a downpour.

Formulations of the principle of charity all seem to run into some trouble or other. Whether it is assuming the speaker is speaking the literal truth, or assuming the speaker is expressing coherent beliefs, or assuming the speaker is like the hearer in her attitudes -- all appear vulnerable to counterexamples, that is, cases where operating with one of these expectations, the hearer would fail to correctly understand the speaker. One can argue that these 'counterexamples' are just exceptions to a general rule, of course. But the failures seem rather systematic.

One way in which hearers generally are rather cooperative is in their provision of feedback to the speaker to indicate various things: that they are following (or not), that they agree (or not), that they would like to hear more (or to cut in). Herb Clark has done interesting work on this (e.g. Clark and Krych 2006). There is also work on this sort of thing, particularly the way that speakers and hearers 'negotiate' taking turns in conversation, in Conversation Analysis (e.g. Sacks et al 1992).


Clark, H. H. & Krych, M. A. (2004). Speaking while monitoring addressees for understanding. Journal of Memory and Language, 50(1), 62-81.
Grice, P. (1989). Studies in the Way of Words. Cambridge, Mass: Harvard University Press.
Kasher, A. (1976). Conversational maxims and rationality. In A. Kasher (Ed.), Language in Focus: Foundations, Methods and Systems. (pp. 197–216). Dordrecht, Holland: Reidel Publishing Company.
Levinson, S. C. (2000). Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, Mass: MIT Press.
Quine, W. V. O. (1960) Word and Object. Cambridge, Mass: MIT Press.
Sacks, H., Jefferson, G. & Schlegloff, E. (1992). Lectures on Conversation. Oxford: Blackwell.
Sperber, D. & Wilson, D. (1986). Relevance: Communication and Cognition (2nd Ed. 1995). Oxford: Blackwell.
Wilson, D. & Sperber, D. (2002). Truthfulness and relevance. Mind, 111(443), 583–632.
Wilson, D. & Sperber, D. (2004). Relevance theory. In L. R. Horn & G. L. Ward (Eds.), The Handbook of Pragmatics. (pp. 607–632). Malden, Mass: Blackwell.

Honourable Mr Elephant: Translating honorifics

Reading an interesting review of Shipwreck (or Shipwrecks) (破船 ha sen) by Yoshimura Akira (吉村昭) got me thinking about the meaning of honorifics and about translation.

The reviewer doesn’t say this (and nor do the other reviews available online), but ‘fune’ in ‘o-fune-sama’ is surely ‘boat’ or ‘ship’: ‘舟’ presumably (although there are other characters that mean ‘ship’ and are pronounced ‘fune’). The honorifics ‘o’ and ‘sama’ indicate something like auspiciousness here, I think, so ‘o-fune-sama’ might be translated as ‘the blessing of a boat’, in the sense of a boat received by the village as a kind of gift from higher forces: from the gods, or fate, say. (Cf. ‘In her ninetieth year, Sarah was surprised and delighted to receive the blessing of a child.’)

Honorifics are notoriously difficult to translate. The translator of Shipwrecks obviously judged that with ‘o-fune-sama’ it was best not to try.

Even when a good translation is possible, it often leaves out the honorific dimension. In most contexts v- and t-forms (e.g. ‘vous’ and ‘tu’) are best just translated as ‘you’, as is Spanish ‘usted’ -- a third-person pronoun used to refer to the addressee -- although in very limited contexts (at a barber’s, a tailor’s, or in a PG Wodehouse story) sir might prefer a more literal translation, hmmn? (Of course, English has other, role-specific, third person-ish honorifics: ‘Your Grace’, ‘Mr. President’.)

Returning to Japanese, ‘o-kane’ is just money, not ‘honourable money’ (while ‘kane’ can sound a bit rough and in some contexts might be best rendered into slang: something like ‘dosh’ or ‘the readies’).This demonstrates that the prefix ‘o-’ is sometimes required for polite speech, particularly for those (like women) who are not thought capable of speaking colloquially and politely at the same time.

But Japanese honorifics are particularly context-sensitive in their effects; they don’t always raise the tone. I suspect everyone has to say ‘kami-sama’ or ‘o-kami-sama’, not bare ‘kami’ (god) – or sound mildly blasphemous. But in other cases honorifics can indicate closeness as much as respect. My favourite example, ‘(o-)zou-san’ (lit. (honourable) Mr elephant) is a phrase used by children, those who speak to children, and those who speak to elephants while children are listening. A friend who speaks much more Japanese than me once suggested that ‘o-furo’ (the normal way of saying ‘bath’, but literally ‘honourable [warm] bath’) should be translated ‘lovely hot bath’.

What is the point of all these observations? Partly reinforcing my preexisting beliefs (prejudices) about translation: it is an art, or an applied science, and not a domain which could ever have its own theory. (One of the main contentions of Ernst-August Gutt’s dissertation on relevance theory and translation: published by Blackwell’s – review in the Journal of Linguistics.) It must be a lot of fun.

But also what might be a novel thought. Resistance to translation (and to paraphrase) is supposed to be typical of procedural meaning, exemplified by so-called discourse connectives like ‘so’, ‘then’ and ‘alors’. And it is easy to see how procedural meaning need not contribute to truth conditions, and it is clear that honorifics usually do not: I might have been wrong to tutoyer you, but that doesn’t mean I didn’t refer to you; and ‘o-kane’ and ‘kane’ surely have the same denotation. Perhaps then the semantics of honorifics is procedural. Surely someone must have said this. Relevance theory and the study of polite speech are both popular with people who study pragmatics in Japan. But I can’t find it on the obvious search terms: relevance theory, procedural meaning, honorifics, keigo.

In any case, I’m not convinced. Some questions of politeness and word choice obviously have nothing to do with word meaning in a narrow sense. That a taboo word is forbidden (and in which contexts, to whom, etc.) is a fact about the social status of that word, not something to explain in terms of its semantics. (Very much pace recently fashionable – but in my view daft – accounts of swear words in terms of ‘conventional implicature’, or ‘presupposition’.) Perhaps the same goes for honorifics. It seems to make sense not to postulate any word meaning for honorifics – given that politeness is not necessarily part of speaker meaning, as Mark Jary pointed out some years ago. I wrote in the entry on politeness in my Key Terms book,

“[It is not clear] whether politeness is communicated. Is a speaker who is being polite necessarily, or even usually expressing politeness as part of her speaker meaning? One view is that in being polite a speaker is mainly trying to avoid any implicatures to do with the speaker-hearer’s relationship, by staying within certain parameters of socially acceptable behaviour.”

Compare this with what Mark proposes: a “Relevance Theoretic account of polite verbal behaviour” which:

“distinguishes cases where politeness is communicated from those where it is not, [and] distinguishes the strategic manipulation of expectations of politeness from cases where politeness emerges from the speaker crafting her utterances in such a way as to avoid making manifest assumptions likely to have a detrimental effect on her long term social aims.”

(Last year when I wrote this I hadn’t read Mark’s paper. I should have known about it, though, and I would have put his paper in the list of further reading at the end of the book. It's a good deal more interesting, in my view, than the references on politeness that I did include.)