Académique Documents
Professionnel Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
Oxford University Press and Analysis Committee are collaborating with JSTOR to digitize, preserve and extend access to
Analysis.
http://www.jstor.org
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
ANALYSIS 30.6 JUNE 1970
By ERNST TUGENDHAT
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
178 ANALYSIS
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
THE MEANING OF 'BEDEUTUNG' IN FREGE 179
object named,the significanceof the sentencemust also be thought of as
an object? Surelynot. And if we grant to the first party that sentences
are not names, do these philosophersnot agree that there is a functional
connection between the referenceof a name and the truth-valueof the
sentences of which the name may be a part and that so far they have
somethingin common? Let us call this again their significance. Now if
the significanceof a sentence cannot be thought of as an object named,
this party ought to present an alternativeaccount of significance. The
two partiescould thereforebe reconciledon the basis of a new account
of significancewhich is not biased towards the name relation and yet
does full justice to Frege's discovery of the functional connection
between objects of proper names and truth-valuesof sentences.
Nothing would seemeasierthanto provide such an account. Modern
semantics is alreadyin possession of a technical term for significance
which is not biasedtowardsthe name relation,the term 'extension'. We
speak of the extension of names and of sentences (and of predicates)
without necessarilyimplying that the extension is, except in the case of
names, an object. So the solution to our problem seems readyat hand:
the significanceof an expression is its extension.' Although I believe
that this answer leads in the right direction, it is not satisfactoryas it
stands, because the term 'extension' is defined in a differentway for
names and for sentences(and again for predicates). Two sentenceshave
the same extension if and only if they have the same truth-value,and
two nameshave the same extension if and only if they refer to the same
object. The term 'extension'is used in both cases for the samereasonas
Frege used the term 'significance'in both cases, but whilst Frege gave
an answer to what it is they have in common, the term 'extension'does
not give an alternativeanswer, it simply leaves the matter open. Our
question can now be reformulatedthus: what is it that the extension of
names and of sentences have in common? Can we find a unitary
definitionof 'extension'which is not biased towardsthe name-relation?
Let us take as our point of departureFrege's own introductionof the
term 'significance'for sentences. He writes (SB 32f):
Does a sentenceas a whole perhapshave only a senseand no signi-
ficance?It might indeedbe expectedthat such sentencesoccur,just as
there are parts of sentenceshaving sense but no significance. And
sentenceswhichcontainpropernameswithoutsignificance will be of this
kind. The sentence'Ulysseswas set ashoreat Ithacawhilesoundasleep'
clearlyhasa sense. But sinceit is doubtfulwhetherthe name'Ulysses'in
that sentencehas a significance,it is also doubtfulwhetherthe sentence
as a whole has one. But what is certainis that anyonewho seriously
takes the sentenceto be true or false ascribesa significance,and not
merelya sense,to the name'Ulysses'... Thatwe concernourselveswith
1 This is
Carnap's answer, given in his book Meaningand Necessity, to which the present
paper owes much.
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
180 ANALYSIS
the significanceof partof the sentenceat allis a signthatas a rulewe also
acknowledge,and require,the sentenceitself to havea significance....
Butnow whyis it we requireeverypropernameto havenot onlya sense,
but a significanceas well? Why are we not satisfiedwith the thought
alone? Because,and insofaras, we are concernedwith its truth-value.
. . It is the strivingfor truth,then, that urgesus in all casesto press
forwardfromthe senseto the significance.
The conclusion which Frege draws from this line of reasoning is:
'Thus we are impelled to accept the truth-value of a sentence as its
significance'. But the passage suggests a furtherconclusion. Frege says
that we are interestedin the significanceof any part of a sentence only
insofar as we are interestedin the truth-valueof the sentence. Is this
not to say that the significanceof the partsof sentences,and in particular
of names, consists in their contribution to the truth-value of the
sentences into which they may enter? In this case we should have to
take the significanceof sentencesas primary.Insteadof transferringthe
characteristicsof the significanceof namesto that of sentences,we should
reversethe orderand try to definethe significanceof names by meansof
the concept with which the significanceof sentencesis defined.
In order to do this I propose the technicalterm 'truth-valuepoten-
tial'. As a first step, this term can be definedfor names in the following
way: two names 'a' and 'b' have the same truth-valuepotential if and
only if, whenever each is completed by the same expression to form a
sentence,the two sentenceshave the same truth-value. This, of course,
is only a cumbersomeway of expressing the well-known definition of
extensional equivalence: a=b =Def. (P)Pa=Pb, which is Leibniz'
Principle of the Identity of Indiscernibles, to which Frege himself
explicitlyappeals(SB 35). But now a furtherstep suggests itself. With
slight modification, the definition can be converted into a general
definition for the truth-valuepotential of an expression,whether name
or sentence or predicate:two expressions 0 and0bhavethesametruth-value
potentialif andonlyif, whenever is
each comfileted bythesameexpression toformt
a sentence,thetwosentences havethesame truth-value. If we substitute names
for 0 and 0, this definitionbecomes identical with the first definition. I
shall returnlaterto the case of predicates. If we substitute sentences for
0 and f, we obtain the following statement:two sentences 'p' and 'q'
have the same truth-valuepotential if and only if, whenever each is
completedby the same expressionto form a sentence,the two sentences
have the same truth-value. Now 'p' and 'q' are alreadysentences;they
are not susceptibleto being completedas sentencesby a furtherexpres-
sion. Therefore, the addition 'whenever each is completed . . . ' is
superfluousin this case, and the definitionis reducedto the simpleform:
two sentences'p' and 'q' have the sametruth-valuepotentialif and only
if they have the same truth-value.
It is obvious that the two definitions for the sameness of truth-value
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
THE MEANING OF 'BEDEUTUNG' IN FREGE 181
potential of names and of sentences are identical with the well-known
definitions for the sameness of extension of names and of sentences.
But they now appear as applications of a single general definition of
sameness of extension. The concept of truth-value potential, thus
defined, can thereforebe consideredas an alternativeaccount of signi-
ficance to that given by Frege. Let us compare the merits of the two
accounts. I shall first show that the new account is more valuableeven
in understandingFrege's own exposition.
Frege's reason for calling both the object of a proper name and the
truth-valueof a sentencetheir 'significance'lies in what has been called
by Carnap the 'principle of interchangeability':if in a sentence we
replace one part by another with the same significancebut a different
sense, then the sense, but not the significance, of the sentence may
change (cf. SB 32). Frege did not give a justificationfor this principle,
and his interpretershave been puzzled over its precise status. Frege
enunciatesthe principlebefore saying what the significanceof sentences
consists in, and he clearly uses it precisely as an instrument for dis-
covering what the significanceof a sentence is. The principletherefore
does not seem to be a propositionthat can be true or false, but functions
ratheras a definitionfor introducingthe term 'significanceof a sentence'.
We shall,Frege seems to say, call the significanceof a sentencewhatever
it is which remainsunchangedwhen we replacea name in the sentence
by another name with the same significance. However, critics have
observed that this is unsatisfactory,since the truth-valueis not the only
thing that remainsunchangedwhen a name is replacedby anothername
of the same object. We can, for example, think of all the sentences
which have the same predicate and whose subject-termsrefer to the
same object as belonging to the same object-class. Then the object-
class of a sentence evidently meets the requirementof the principle of
interchangeabilityjust as the truth-valueof the sentencedoes.
The difficultyis solved if we go about it the other way and take the
function of the principle of interchangeabilityto consist in the intro-
duction of the significance,not of sentences, but of names. We then
start from the truth-valueof sentences, call this their significance,and
proceed to say that whateverpropertyof names remainsthe same when
we exchange them in otherwise identical sentences without changing
their truth-valueshall be called the significanceof the names. And this
simply amounts to saying that we shall call the truth-valuepotential of
names their significance. Proceedingin this direction,it then turns out,
instead of being assumed, that the significanceis, in the case of names,
the object referredto.x
xAs has been pointed out to me by Mr. Dummett, it is not strictly correct to say that the
truth-value potential is the object referred to. All we can claim is that two names that refer
to the same object have the same truth-value potential. Consequently, it would be preferable
to say that the truth-value potential of a name is, rather than the object referred to, its refer-
ence to that object.
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
182 ANALYSIS
One meritthereforeof the accountof significance hereadvancedis
thatit permitsa betterunderstanding of Frege'sown exposition.The
principleof interchangeability,in the interpretation
heregiven, simply
the of of
represents Principle the Identity Indiscernibles,and Frege
himselfrefersto it in this formwhenhe proceedsto the finaltest of his
proposal(SB 35). The conclusionsdrawnfromthe previousconsider-
ations he calls mere 'conjecture'('Vermutung'). Nonetheless,Frege
thoughteven then that he was provingby meansof the principleof
interchangeabilitythatthe truth-values of sentencescorrespondto the
objectsof names,whereasin factthe principleof interchangeability can
only provethatthe objectsof namescorrespondto the truth-values of
sentences.
That Frege proceeds in this reverse direction-from names to
sentences--must have been the main reasonwhy he appliedthe termin-
ology of the name-relationto the significanceof sentences. There was,
of course, an additionalreason,and that was his well-known distinction
between completeand incompleteexpressionsand his doctrinethat both
sentences and names are complete expressions.' But this doctrine in
itself is not enough to accountfor Frege's conception of the significance
of a sentenceas an object, becauseeven if it be conceded that names and
sentencesform one class of expressionsin contrastto predicates,it does
not follow that this classhas no essentialsubdivisionsin turn. And why
the name-relationshould be transferredfrom the sub-classof names to
the sub-classof sentencesFrege never explainedon any grounds other
than the principle of interchangeability.Hence we must conclude that
Frege'sapplicationof the terminologyof 'name'and 'object'to sentences
and their significance is due, in the last analysis, exclusively to the
traditionaldoctrine that the prototype of a complete ("categorematic")
expressionis the name. And yet it was Frege himself who had opened a
new approachwith the famous dictum in his Grundlagen derArithmetik
(?60): 'Only in the context of a sentence does a word signify anything'.
It is this statementwhich points to the conception of significanceas
truth-valuepotential.
So far I have been concerned to show the advantagesof the new
conception of significancefor the interpretationof Frege's own text.
Let us now comparethe two conceptions in their own right.
There exists a connection between names and sentences, expressed
by the principleof interchangeability.This, and this alone, is the fact to
be accountedfor. What the principleof interchangeabilityexpressesis
not the symptom of anything else, of some deeper propertythat names
and sentences might have in common. If we wish thereforeto charac-
terize both names and sentences by one and the same property, which
we can call the property of having a significance, this property must not
1 Cf. the essay 'FunktionundBegriff',especially p. 18.
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
THE MEANING OF 'BEDEUTUNG' IN FREGE 183
consist in anything over and above the functional connection which is
expressedby the principleof interchangeability. This requirementis
met by the conceptionof the significanceof an expressionas its truth-
valuepotentialandis not metby theconceptionof the significance of an
expression as the objectnamed by it.
But the argumentdecisivefor the adequacyof the accounthere
proposed,andfor theinadequacy of Frege'sown account,is the follow-
ing. Granted on the one hand that namesand sentencesform two
differentsemanticcagetoriesandon the otherthattheyhavesomething
in common,we mustrequireof an adequateaccountof whattheyhave
in commonthatit shouldnot obliteratetheirdifferences.This require-
ment is only met by the presentaccount. Why is this so? Why is it
thatif we interpretthe significanceof sentencessettingout fromnames,
we cannothelp assimilatingsentencesto names,whilstnamesare not
assimilatedto sentenceswhen we interpretthe significanceof names
settingout fromsentences? Thereasonis thatwe havehereaninstance
of a functionalconnectionbetweenpartandwhole. In anysuchinstance,
for examplea tool, machine,or organism,the partcan only be defined
by its relationto the functionof the wholeandnot viceversa.Sincethe
relationof partto wholeis functional,the referenceto the wholein the
definitionof thepartdoesnot resultin the assimilation of theproperties
of the part to the propertiesof the whole. On the other hand, any
attemptto definethe wholeby meansof its partsis boundto resultin a
non-functional accountof the wholewhicheitherassimilates its proper-
ties to the propertiesof the partor definesit as a mereconglomeration
of its parts,or both.
Thus the fact that the interpretation of significanceas truth-value
potentialis adequatewhile its interpretation as referenceis inadequate
shedslight on the natureof sentencesandtheircomposition:it canbe
used as evidencefor the claimthat the primarysemanticunit is the
sentenceand it can also be used to protectthis claimfrom misunder-
standing.The contentionthatthe sentenceis theprimaryunitof mean-
ing does not excludeits divisibilityinto meaningfulparts; it only
claims that the significance,and consequentlythe sense, of words
cannotbe understoodin isolation,but ratherconsistsin theircontri-
butionto the significanceor senseof sentences,respectively.Wheels,
cranksandpistonscanexercisetheirfunctiononlyaspartsof a machine;
but this is not to say that they cannotbe takenapartand usedin the
constructionof a new machine.
WhatI havesaidso far canbe summarized as follows. The correct
accountof Frege'sterm'significance' would seemto be to understand
it as truth-value potential;andsincethis accountalsoagreesbetterwith
some parts of Frege's exposition thanhis own account does, we can also
say that this was what Frege himself really meant and that he was only
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
184 ANALYSIS
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
THE MEANING OF 'BEDEUTUNG' IN FREGE 185
means is that, although we should not say that the significanceof a
concept-word is the extension, 'two concept-words have the same
significanceif and only if their extensions coincide'(p. 133). So here we
have a case where Frege refusesto considerthe significanceas any kind
of object whatsoever and is content to say under what condition two
predicateshave the same significance. And althoughFrege does not put
it in these words, this condition clearlyconsistsin having the sametruth-
value potential.
What Frege says in this manuscriptabout 'having the samesignific-
ance' corresponds to the characterizationof 'having a significance'
which he gives in the Grundgeset!e ?29. A predicate,he there says, has a
significance if, when to
applied any propername that has a significance,
it yields a sentencethat itself has a significance.1
These explanationsare furtherconfirmedby what Frege says about
what it means for a predicate to have no significance: 'When we are
concerned with truth, . . . we have to reject concept-words whose
demarcationis indistinct. Of every objectit must be determinedwhether
it falls under the concept or not; a concept-wordwhich does not satisfy
this requirement has no significance.'2 This passage is particularly
illuminating, for Frege is here pointing to the specific kind of truth-
value potentialof predicates,which is essentiallydifferentfrom the truth-
value potentialof names. This distinctionis all too easily blurredby the
usualaccount,accordingto which both objectsand conceptsareregarded
by Frege as "entities". Frege did not use such a term, and his so-called
"realism"appearsto be overemphasizedin the literature. Although we
may say that a concept-word "stands for" a concept, just as a name
standsfor an object, what this meansin the case of a concept-wordis that
it provides a demarcationfor the distinction of objects.
Looking back now at names, sentences and predicates, we can
conclude that what Frege discovered was not, as is often said, that
names have, besides a reference,a sense and that sentences and predi-
cates have, besides a sense, a reference,but that all these expressions
normally have, besides a sense, a significance in terms of truth and
falsity:3sentencesare significant('bedeutungsvoll') insofar as they are true
or false; predicatesare significantinsofaras they aretrue of some objects
property of predicates thus defined has been labelled by Montgomery Furth, in his
1 The
article 'Two types of denotation' (Studies in Logical Theory,ed. N. Rescher, Oxford 1968,
pp. 9-45) the 'property Z'. Furth thinks that 'the problem is: is possessing the property Z
anything like having denotation?' (p. 31). But Frege did not use the word 'denotation', and
the word which he did use does not commit us to asking any such further question. The
analogies which Furth proceeds to point out between what he calls two kinds of denotation
are precisely the analogies which obtain between names and predicates insofar as both have
a truth-value potential.
2 NachgelasseneSchriften,p. 133.
3 Here the non-technical meaning of in the sense of "significance" comes to
"Bedeulung"
the surface.
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
186 ANALYSIS
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
THE MEANING OF 'BEDEUTUNG' IN FREGE 187
of the complex sentence depends in one case on the normal significance
of the component sentences and in the other on their oblique signific-
ance, in every case thereforeon their significance.
This theory is usually considered to be somewhat artificial and
invented solely in order to save the universalityof the principleof inter-
changeability. Besides, Frege is accusedof having let himself be misled
by the close connection between significanceand the name-relation;it
is true that a sentence may function as a name, but how could the
significanceof a sentence, in Frege's technical sense of the word 'signi-
ficance',ever be anythingbut a truth-value? Is it not, then, more satis-
factory to throw overboard the universalityof the principle of inter-
changeabilityand remain content with saying, as one normally does in
contemporarysemantics,that there are extensional contexts and inten-
sional contexts?
I think not. The suggestion just madethat a sentencecould have no
other significance, in Frege's technical sense, than a truth-value, is
mistaken. A significance,in Frege's technical sense, can be anything
which may be considered a truth-value potential of any kind. Now
when a sentence is nominalized and a predicateattachedto it, or as is
common in intensional contexts a two-place predicateis attachedto it
and another name, the sentence, Frege says, merely expressespart of a
"thought", and only togetherwith a predicatecan itforma sentence,that
is to say, express a thought; such a sentence cannot stand by itself (cf.
SB 36f). And, we may add by way of elucidation, the truth-value
potential of such a sentencewhich cannot stand by itself cannot consist
in a truth-value. It can only consist in the contributionwhich it makes
to the truth-valueof sentencesinto which it may enter as a part. Apply-
ing our generaldefinitionof truth-valuepotentialto this case, we obtain
the statementthat two nominalizedsentenceshave the same truth-value
potentialif and only if, whenever each is completedby the same expres-
sion to form a sentence, the two sentences have the same truth-value.
And, since this is only the case when the sentences which have been
nominalized have the same sense, Frege's contention is fully justified
that what he calls the oblique significance,that is to say the significance
of the sentence in its new role as subject of a second-ordersentence, is
the sense which the sentence has when it functions independently.
Nominalization,then, is not just an accidentalgrammaticalfeature. In
whateverway it may be grammaticallyexpressed,a sentenceassumesthe
role of a name, when it is so used that its truth-valuepotential standsin
need of supplementationby a predicativetruth-valuepotential to yield
a truth-value. This result shows once more that Frege's concept of
significanceis functional:the significanceof one and the sameexpression
differsaccordingto whether it expresses a self-sufficientsemanticwhole,
a "thought", or only part of one.
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
188 ANALYSIS
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions
THE MEANING OF 'BEDEUTUNG' IN FREGE 189
must be converted into names, and the larger sentences then turn out
to be themselves subject-predicatesentences. The universality of the
principle of interchangeabilityis, then, a consequence of this primacyof
the subject-predicatesentence.
I have passed over sentences with 1st and 2nd order quantifiers.
But their significance is also defined in terms of the significance of
subject-predicatesentences. There are, however, kinds of sentence
which seem to resist this account, in particularcausal sentences and
contrary-to-fact-conditionals.But such sentencespresent difficultieson
any account. Frege tried, in the last part of his essay, to explicatesome
of these more recalcitranttypes of complex sentence,in particularcausal
sentences,but he did not attemptto relatehis explicationof these types
to the assumeduniversalityof the principleof interchangeability.
University
of Heidelberg
This content downloaded from 132.174.255.116 on Mon, 22 Feb 2016 23:18:17 UTC
All use subject to JSTOR Terms and Conditions