Theories of Meaning

Rather too many years ago I spent a considerable amount of time reading and thinking about “The Theory of Meaning”, a somewhat esoteric part of Analytic Philosophy. I recently returned to this for no reason that I can easily identify (although that in itself is of some significance – see below), and it has given rise to a multiplicity of thoughts, some directly related to the topic and some not.

The directly-related thoughts are about philosophical argument and how sometimes a picture really is worth a thousand – indeed, several thousand – words; the indirectly-related thoughts (which I think are much more interesting) are about (a) what possible justification there can be for some of the smartest people on the planet engaging in this kind of ultra-abstruse discussion; (b) the answer to this question framed in terms of cultural holism; (c) a reaffirmation of fundamental principle that nothing is what it seems, things are far more complex than we ever imagine, and only mining the immensely obscure and difficult depths of human thought can hope to rescue us from this plight, if indeed anything can; (d) the social connectedness presupposed by holism – cf. the reference below to Michael Dummett’s essay “The Social Character of Meaning” – implies that we may find ourselves engaged in something seemingly utterly irrelevant and esoteric, even trivial and banal, but there may be strong unknown reasons that explain and justify such activity we cannot possibly determine in advance of doing it, and perhaps not afterwards either.

As an example of the fourth of these inferences, (d), I could cite the sequence of events that brought about this essay: taking down from a shelf seemingly aimlessly and at random a copy of a book I have not looked at for years; reading pages that seemed to bear no relationship to anything I am doing or thinking or even interested in at the moment; and finding shortly afterwards that they have precipitated a deluge of far-reaching ideas, some of which are reproduced below.

One of the best places to start, and the book in question, although far from being the easiest, is Michael Dummett’s The Seas of Language, OUP, 1993, reprinted in paperback in 1997. This collection of essays and lectures begins with two called “What is a Theory of Meaning? (I)” and – go on, take a wild guess! – “What is a Theory of Meaning? (II)”.

Professor Sir Michael Dummett (1925 – 2011) combined an extraordinary distinction as a philosopher with a passionate loathing of racism and a profound concern for migrants and refugees, thus demonstrating his own practical answer to the charge levelled above in (a). His obituary in The Daily Telegraph includes this:

“But his commitment to truth had very practical applications, and ones which he pursued with vigour and personal courage. In particular, throughout his career he maintained a deep interest in the ethical and political issues concerning refugees and immigration, informed by what he described as ‘an especial loathing of racial prejudice and its social manifestations’.”

And then subsequently …

“Dummett saw the root of the problem as lying in the political system. In his book On Immigration and Refugees (2001), he argued that lurking behind the egalitarian veneer of democracy is the more manipulative principle of playing on people’s prejudices to gain votes. This, when applied to issues of immigration, has invariably led to a jingoistic policy – a policy founded, essentially, on racism. In Britain, according to Dummett, much of the blame rested with the Home Office, a department which he accused of “decades of hopeless indoctrination in hostility”, first against Commonwealth immigrants, and later against asylum seekers and refugees. “For the Home Office,” he once wrote, “the adjective ‘bogus’ goes as automatically with ‘asylum seeker’ as ‘green’ does with ‘grass’.”

http://www.telegraph.co.uk/news/obituaries/culture-obituaries/8981654/Professor-Sir-Michael-Dummett.html

Dummett is persuaded for some very good reasons that a theory of meaning is really a theory of understanding, and that a full-blown theory of meaning would consist in an explanation of what it is to understand a language. To substantiate and explicate this claim he first has to deal with a cut-down and unsatisfactory misconception about meaning related to translation. His example is what he calls an “M-sentence”: “‘La terra si muove’ means that the Earth moves”. This tells us that a sentence in one language means the same as a sentence in another language, but it does not actually tell us anything about what those sentences refer to, what knowledge they entail, or whether indeed they entail any knowledge at all. This becomes more clear when he says that translation cannot furnish knowledge any more than the similar M-sentence “‘The Earth moves’ means that the Earth moves” (p.7).

To my mind it is unfortunate that Dummett and others like Davidson and Kripke use real sentences to try to illustrate this point, because a sentence framed in terms of non-existent entities (entities that in this instance exist only within the confines of rooms where I am teaching a philosophy elective, where they materialise on command), makes it much more powerfully: “A shpringlehock is called a ‘shpringlehock'” means that a shpringlehock is called a ‘shpringlehock’. Or “‘Shpringlehocks are grue’ means that shrpinglehocks are grue. These sentences are undoubtedly true within a cut-down and insufficient theory of meaning, but inasmuch as they do not enable us – or even require us – to possess any knowledge, still less understanding, they cannot qualify as examples of a full-blown theory of meaning.

This, I take it, is reasonably easy to understand, but what comes next is far more controversial, and cuts through the entire world of analytic philosophy where there is no satisfactory resolution to the question it poses, and where I think even Dummett flounders. When we ask what would constitute a full-blown or “full-blooded” as Dummett calls it (p.5) theory of meaning, we immediately find ourselves confronted by the challenge posed by a holistic theory of language such as that proposed by Quine in “Two Dogmas of Empricism”: does any word or any sentence assume its full meaning, and therefore convey fully whatever knowledge it contains, and therefore require and entail all the understanding necessary to grasp it fully, unless the entire language has been grasped?

Does any word or any sentence assume its full meaning, and therefore convey fully whatever knowledge it contains, and therefore require and entail all the understanding necessary to grasp it fully, unless the entire language has been grasped?

Dummett is inclined (op cit. p.17f) to reject linguistic holism because he believes that it can give no sensible account of what it is to learn a language, little-by-little, or to grasp a language partially, as would, on reflection, be true not only of those learning the language word-by-word and sentence-by-sentence, but of absolutely all of us, the most knowledgeable and wise and clever native-speakers included.

In a much earlier essay he addresses a similar problem by asking in what a full-blown concept of “gold” would consist, observing that the word plays a part in all sorts of different parts of the English language: in chemistry; in literature; in economics; in mythology; and so forth. Would it be reasonable to deny that someone understood the word “gold” merely because they had failed to become fully conversant with its use in all these different language-games? Some think it would not be reasonable (cf., inter alia, for a brilliant and detailed discussion by Dummett, “The Social Character of Meaning” in Truth and Other Enigmas, OUP (1978), chapter 23, p.427 et al. where he distinguishes “gold” from “elm” in various ways in the context of an analysis of a thesis advanced by Hilary Putnam). I think they are mistaken: whether something is “reasonable” in this colloquial sense of the word is not a philosophical criterion. It is transparently and tautologously true that someone who only understands some of the uses of the word “gold” and therefore possesses only part of the concept “gold” does not have a full-blown understanding of the word ‘gold’.

Why is this relevant? Granted that “Horses are called ‘horses'” means that horses are called ‘horses’ will not do as a theory of meaning, we need instead to have a way to characterise what the meaning of singular terms such as ‘horse’ might be, which is to say what it is to understand a word such as ‘horse’. Given further that absolutely nobody can ever be familiar with every conceivable use of any word, since the set of such sentences would constitute an uncountably-infinite set for exactly the same reason that all conceivable uses of the number “1” would constitute an uncountably-infinite set (as George Cantor famously and definitively proved; a proof for a putatively exhaustive list of all possible sentences containing the word “gold” could as easily be constructed), we are bound to conclude that a holistic theory of language entails the inescapable conclusion that nobody actually fully understands the meaning of any word, and therefore the meaning of any sentence, and therefore anything conveyed by a sentence or set of sentences. In short, nobody fully understands anything at all.

Nobody fully understands anything at all.

Someone might object that it is not reasonable, still less useful, to advocate or espouse a theory of language so demanding that it leads to the conclusion that no user of a language ever understands what their language means, for that would presumably require acceptance of the inference that no sentence uttered or written by any language-user is either ever fully understood by that user or by the user’s hearers and readers, and therefore that nobody ever fully understands what they mean by what they say/write or what anyone else means by what they say/write. But this inference is not an objection to the thesis; it is a confirmation of it, and indeed the most important possible conclusion to be drawn from it.

No sentence uttered or written by any language-user is either ever fully understood by that user or by the user’s hearers and readers, and therefore nobody ever fully understands what they mean by what they say/write or what anyone else means by what they say/write.

One of the welcome inferences to be drawn from this realisation, one of its happiest consequences, is that the apparent distinction between the solid reality of language and the apparent vagueness of art, music, and drama, is dissolved. An author is no more capable of knowing what she means by what she says or writes than an artist by what she paints. The inescapability of vagueness embraces art as well as language, but that vagueness, far from being a weakness, is for both art and language an irreducible strength. As Michael Polanyi once put it,

“For just as, owing to the ultimately tacit character of all knowledge, we remain ever unable to say all that we know, so also, in view of the tacit character of meaning, we can never quite know what is implied in what we say.”

Michael Polanyi, Personal Knowledge, p.95.

For similar reasons the vagueness of linguistic meaning also embraces – one might almost say “engulfs” – science, since while scientists might wish to contain the meanings of the words they use within a restricted language that admits the reinvention of “facts”, in practice this is a temporary illusion typical of a passing phase in scientific understanding, almost a deception, for a thorough-going quantum-mechanical understanding of science would force us again to embrace the universality of vagueness and the unknown.

Now here you might think that our dissenters can legitimately refer, and with more logical justification, to what is reasonable in a strong sense: is it reasonable to use language to affirm that nobody fully understands language; does that not visit upon us a vicious circle inasmuch as to claim to understand the sentence “Nobody fully understands anything at all” in full would simultaneously negate it. We seem inadvertently to have constructed a quasi-Gödelian self-referential sentence: if we are able to know it to be true, it must be false, since that would entail knowing at least one thing fully. But the inverse does not work: if we are able to know it to be false, that does not entails that we know it to be true because we can know it to be false on the basis of some other sentence that we understand fully, even if not this one.

And in fact even the first inference fails within a vague theory of language, which might better or more unambiguously be framed as a theory of language that characterises all meaning in terms of vagueness: that the ability to use a language does not require or entail a full command or understanding of the meanings of words, but the skill of employing and deploying words whose meanings are inescapably vague in a way sufficient to effect communication with other language-users who are similarly placed.

The ability to use a language does not require or entail a full command or understanding of the meanings of words, but the skill of employing and deploying words whose meanings are inescapably vague in a way sufficient to effect communication with other language-users who are similarly placed.

To a question such as “Does anyone fully understand the meaning of a word such as ‘gold’?” we can then give a two-part answer: “No, nobody fully understands the meaning of a word such as ‘gold’; but fully understanding the meaning of a word is not necessary in order to use it effectively for someone sufficiently skilled in a particular language”.

And therefore it is no objection to the central conclusion “Nobody fully understands anything at all” that this entails that nobody understands what “Nobody fully understands anything at all” really means. At the very least we can assert that nobody fully understands what it means “to understand” anything, and so this sentence does no more, and no less, than point to the fact that everything is vague (including the fact that everything is vague).

Everything is vague including the fact that everything is vague.

To the question “Does that mean that you don’t really understand what you mean when you assert that everything is vague?” we can therefore give an affirmative answer, and to the recidivist quasi-Gödelian riposte “But that suggests that you at least understand that you don’t really understand” we reply “No, it means that we don’t really even understand what it is we lack when we lack understanding, or even what this assertion means“. And one can immediately see how a meta-recidivist quasi-Gödelian can carry on like this ad nauseam or ad infinitum, whichever comes first (a joke I owe to a lecturer in combinatorial theory at Oxford whose name temporarily escapes me).

It is worth digressing to point out that under this reconceptualisation of language in terms of vagueness there are, strictly speaking, no such things as facts, and therefore the secondary premise with which Wittgenstein began the Tractatus – that “§1.1 the world is the totality of facts, not of things” – crumbles, not because we have reinstated the priority of “things”, which remain as completely beyond us as Wittgenstein supposed, but because we have dissolved the notion of “facts” within the seas of vagueness. Of course, Wittgenstein also repudiated the Tractatus later, and much of analytic philosophy was spawned by his Philosophical Investigations, which takes an entirely different position.

To cut short this Sisyphean (or perhaps it is a Tarskian) process we observe that in the sentence quoted above every word is vague: “everything”; “is”; “vague”; “including”; “the”; and “fact”.

But this excursus into the self-referential brings us back to an important realisation that cannot but apply to any attempt to construct a theory of meaning: that to state in language how language relates to what it means is to presuppose that we have solved the problem of how the explanatory terms we employ in such statements relate to the terms of the language, which is the philosophical equivalent of solving the problem of how you catch a lion “by catching two and letting one go”.

And we are brought back not to Michael Dummett, but to Wittgenstein: we cannot state in language how we use language or what language means; we can only show that we understand by using language in a community of like-language-users who deem our utterances sufficient for communication. We may not wish to reaffirm that “the meaning is the use”, but we certainly wish to convey the fact that we demonstrate our understanding of meaning by the skills we deploy in our use of a langauge; we convey our understanding of the concepts referred to by the words of the language by the way we deploy the skills required to communicate with them to the satisfaction of life-minded language-users. Dummett and others may want more, but it is not reasonable to want what one cannot possibly have.

[To be continued.]