THE SEAS OF
LANGUAGE
Michael Dummett
CLARENDON PRESS » OXFORD
1993,Acknowledgements
“The author is grateful to the following for permission to reprint copyright
material
‘Oxford University Press for Essays 1 and 2 in this collection;
Stockholms Universitet for Essay 3;
Kluwer Academie Publishers for Essays 4, 11, and 13;
Pergamon Press Ltd for Essay 5;
Basil Blackwell Publishers for Essays 7,15, and 1!
‘Cambridge University Press for Essay 8;
Macmillan India Ltd. for Essay 12;
Macmillan London Lid. for Essay 16.
What is a Theory of Meaning?
@
According to one well-known view, the best method of formulating the
philosophical problems surrounding the concept of meaning and related
notions is by asking what form that should be taken by what iscalled a theory
‘of meaning’ for any one entice language; that i, a detailed specification ofthe
‘meanings ofall the words and sentence-forming operations ofthe language,
yielding a specification ofthe meaning of every expression and sentence ofthe
language. Its not thatthe construction ofa theory of meaning, in this sense,
for any one language is viewed as a practical project; but it is thought that,
when once we can enunciate the general principles in accordance with which
such a construction could be carried out, we all have arrived ata solution of
the problems concerning meaning by which philosophers are perplexed.
share the belief that ths is the most fruitful approach to the problems
within this area of philosophy, although I should not feel capable of giving a
‘demonstration that this was soto someone who deniedit: but wecan see some
reasons for it if we contrast certain other cases, So far a8 I know, no one has
‘ever suggested a parallel approach to the problems of epistemology: no one
has proposed that the right way to go about tackling the philosophical
problems relating to the concept of knowledge would be by considering how
fone might construct a theory of knowledge in the sense of a detailed
specification of everything that any one individual, or community, can be said
to know. The reason is, I think, that our grasp on the concept of knowledge is
rather more secure than our graspon the concept of meaning. Wearein doubt
about what ought to count as knowledge; we are even more in doubt about
how to formulate the principles we tacitly apply for deciding whether or not
‘something is o count as knowledge; we also have some uncertainty about the
semantic analysis of a sentence altibuting knowledge of something to
somebody: but at least we are quite certain which are the sentences whose
logical form and whose truth-conditions we are seeking to analyse. By
‘contrast, while most of us, myself included, would agree that the concept of
‘meaning isa fundamental and indispensable one, we are unclear even about
the surface structure of statements involving that concept. What kind of
sentence, of natural language, should be taken as the characteristic form for an
Fit publ in Samuel Gutenplan (of), Mind and Language, Woon Lctares, 1974
(ontre)1975,2 What is a Theory of Meaning? (1)
attibtion of particular meaning to sme given word or expression? Not
nly do we not know te answer to hit: we donot even know whether tthe
right question to ask, Perhaps it isimporibl, in generat state the meaning
ofan expression: perhaps we ought, rater, to enguie by what linguist
means, or poly even nonlingusie means, its possible To contey the
meaning fon expression, otherwise than by exalt statngit Or perhaps
ten hats wrong: perhaps the question shouldbe, not how we express hata
particular exprstion hata clan meaning, but how we should analyse
Sree wh ae te one of aig ae eres way.
ely because i this ate of pllosophy, we know eve ss What ts that
Weare taking about than we don othe areas, tha he proposal o approach
bur problems by considering how we might attompt to specify the meanings of
the expressions ofanenie language doesnot appear the waste of time that an
nalogous proposal would scem tobe within epstemotog.
Asis wellknown, some, pre-eminenly Quine, have pretered to circumvent
this by ivesigating th principles underying the construction not of
2 theory of meaning fra language, bu ofa ansltion manual fom i ato
some known language. The advange i that we know exacly what form a
teanalation mana has o take, namely an fective set of ules for mapping
sentences of he translated language nto sentences of the language ato which
the tranlation is being made! we ean therlore concentrate ently upon the
auestoeshow wesretosrrive al saysem of ransationasembodie ineuch a
mnanval, and what conditions must be satifed for nich a system to be
compile The disadvantage ls tha, wile the iteret of such an enquiry
Imus iin the light hows o the concept of meaning, we are unable tobe
Certain what consequences te results ofthe engi into tanlation do have
forthe notion of meaning jst because they ae stated without ect ppeal ‘0
that notion. To grasp the meaning of an expression isto understand tole in
the language’ a complete theory of meaning fora language therefore, a
complet theory of how the language functions asa language. Our interest in
‘meaning st «general concept, thus, aninteresin how language work: a
direct dseripion ofthe way 8 lsnguage works--o al that someone has to
lear todo when be leans the language—-would, according) resolve ove
Perplsits in away in which an indtet acount, by means of transation,
annot. tw uie rightly be sak that the interest the enquiry inte
translation attaches, not to the tasltion ise, butt the entra proposed
for judging the accepablty ofa scheme of translation, and that these must,
relate to what canbe observed ofthe working of thelanguage ob Wanslaed
Tnded, it might plausibly be maintained that nothing shot of a complete
sheory of meaning orth language —acomplete account of the way works
could bean adequate basis for judging the correctness oa proposed scheme of
{tanslation. shallot atempitoadjdiate the soundnes of tisclaan It
sound, then the apparent advantage ofthe approach vi translation, rather
What isa Theory of Meaning? (1) 3
than by asking outright what form a theory of meaning for the language
should take, is wholly illusory. If itis unsound—and certainly the actual
procedure of the principal practitioner of the approach via translation
suggests that he takesit as unsound—then it follows that there is no immediate
inference from results concerning translation to conclusions concerning
‘meaning.
have said that iis the job ofa theory of meaning fora language to give an
‘account of how that language works, that is, of how its speakers communicate
bby means of it: here “communicate” has no more precise signification than "do
Whatever may be done by the utterance of one or more sentences of the
language”. And here I will repeat what I have maintained elsewhere, that
theory of meaning is a theory of understanding; that is, what a theory of
‘meaning has to give an account of is what itis that someone knows when he
knows the language, that is, when he knows the meanings ofthe expressions
‘and sentences of the language. One question about the form which a theory of
‘meaning should take is whether it should issue in direct ascriptions of
‘meaning, that is, in propositions of the form “The meening of the
word/sentence X's...” or of the form “The word/sentence X means..." Ifthe
answer to this question is afirmative, then it may seem that such a theory of
‘meaning will have no need to advert explicitly to the notion of knowledge: if
the theory allows us to say that the meaning ofa given word or sentence i
something or other, say Q, then, presumably, we shal likewise want to say that
someone knows the meaning ofthat word or sentence if he knows that Q is
‘hat that word or sentence means. We shall later sce reason to doubt this; but
for the present let us suspend judgement. Ifthe theory of meaning allows us to
derive such direct ascriptions of meaning, and if these direct ascriptions are
such as to lea inthis simple way toa characterization of what tis to know the
‘meaning ofeach word or sentence the language, then, indeed, my claim that
8 theory of meaning must be a theoty of understanding is not intended in $0
strong. sense as to rule out such a theory, merely on the round that it did not
itsef employ the notion of knowledge: it would be proper to accept such a
theory as being a theory of understanding. If,on the other hand, although the
theory of meaning allows the derivation of direct ascriptions of meaning, these
ascriptions are so framed as not to permit an immediate characterization of
what itis thata person knows when he knows the meaning of given word oF
Sentence, then, by hypothesis, the theory is inadequate to account for one
extremely important type of context in which we ae disposed to use the word
“meaning”. If, however, the theory of meaning does not issue in such direct
scriptions of meaning at all; and if futher, it does not contain within ise
any overt account of what someone has to know in order to know or grasp the
‘meaning of each expression of the language, but merely provides an
explanation of other contexts in which we use the word “meaning”, such as".
‘means the same as Y*or"/has a meaning": then itappears tome, it willagain