Philosophical Review

Wittgenstein in Transition: A Review of the Philosophical Grammar

Author(s): Richard W. Miller
Source: The Philosophical Review, Vol. 86, No. 4 (Oct., 1977), pp. 520-544
Published by: Duke University Press on behalf of Philosophical Review
Stable URL: http://www.jstor.org/stable/2184566 .
Accessed: 28/06/2014 17:28

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.

Duke University Press and Philosophical Review are collaborating with JSTOR to digitize, preserve and extend
access to The Philosophical Review.

http://www.jstor.org

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 Review,LXXXVI, No. 4 (October1977).
ThePhilosophical

DISCUSSION
WITTGENSTEIN IN TRANSITION
A ReviewofthePhilosophical
Grammar'
RichardW. Miller

TN 1929, Wittgenstein returnedto philosophy,having abandoned

it ten years previously,after the publication of the Tractatus.
He immediately began theprocessofrevising, combatting,and supple-
mentingthe ideas of the Tractatusperiod,a processwhich ultimately
produced,among otherworks,the Philosophical (written
Investigations
between1936and 1949),theRemarks on theFoundations of Mathematics
(1937-1944),and thenotebookslaterpublishedas On Certainty (1950-
1951).
The PhilosophicalGrammarlargelyconsistsof a manuscriptcom-
pletedin 1933,incorporating writingbegunas earlyas thesummerof
1930. It is in two parts,each almost as long as the Investigations,
"The Propositionand Its Sense" and "On Logic and Mathematics."
I have emphasizedthecronologyof thePhilosophicalGrammarpartly
becauseof its strikingly transitionalcharacter.On the one hand, the
Grammar is extremely forward-looking. More than in the Philosophical
Remarks(1930 and much more than in "Some Remarkson Logical
Form" (1929), Wittgenstein attacks, explicitly and in detail, funda-
mentalideas of his earlierperiod: the notion that a particularspecies
of psychologicalfactsgive sentences their meaning; the requirement
thateveryproposition should have a single complete analysis;theeffort
to provide a single,completely general characterization of the essential
features of every possible language. With the exception of topics con-
cerning certainty and other minds, he is already occupied with the
problems that dominate his laterwritings, such as, "What givessounds
or markstheirlife as expressionsof a thought?","How does the
significance of acts of language-teaching and of expressionsof under-
standingdepend on the circumstances which theyoccur?","How
in
is the meaningof a mathematicalpropositionrelated to it proof?"
Finally,many of the proceduresof the Grammerare frequentand
fundamental in the Investigations,forexample,the replacementof the
searchforthecommonelement which all possibleinstancesof a given
'Ludwig Wittgenstein, PhilosophicalGrammar(Rush Rhees, ed.), translated
byAnthonyKenny.University ofCaliforniaPress,Berkeleyand Los Angeles,
1974.495 pp. $18.00.Citationsfromthe Grammarwill be givenusingunaccom-
panied page numbersor sectionnumbers(the latterin Roman numerals).

520

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

concept must share by the description of an array of instances which

brings out both their "family resemblance" and their differences,
and the criticism of traditional philosophy for exaggerating the
similaritiesamong diverse phenomena, under the influence of super-
ficial grammatical analogies. On the other hand, there is a crucial
that is not found in the earlier book, the idea
aspect of the Investigations
that words could not have their meanings if they did not typically
functionas tools, to help others,to obtain help, to master nature, or in
similarways. While thereis an early use of the phrase "language game"
in thePhilosophicalGrammar (p. 43), the mature concept is not yet present,
and the instrumental aspects of language are generally assigned a
marginalsignificance.In the course of this review,I shall argue that this
differenceconstitutes a limitation of the earlier book, reducing the
plausibility and effectivenessof the discussions of meaning in the
Grammar.
The PhilosophicalGrammar,like all of Wittgenstein's writings, is
the opposite of a textbook. His remarks are frequentlyobscure. They
oftenseem pointlessor paradoxical. In my subsequent surveyof what I
take to be some main ideas in the Grammar,I shall often be interpreting
Wittgenstein'swords in ways that are open to dispute. Ideally, my
interpretationswould be accompanied by extensive quotations from
the text,but considerationsof space preclude this. I hope the interested
readers will compare my interpretationswith the passages I cite.

"Part I: The Proposition and Its Sense"

The firstpart of the PhilosophicalGrammar concerns topics of broader

interestthan the second, and, in my view, sheds more light on what it
examines. I shall only have space to consider Wittgenstein'sreflections
on one general question, in Part I. But it is a question to which he
constantlyreturns,here as in the Investigations: "What can distinguish
something with a particular meaning from a mere occurrence of
sounds, marks, or experiences which is semantically dead?" Thus,
Wittgensteininquires as to what makes a thought the thought of
something(sectionsV and VII), a picturea pictureofsomething(sections
V and IX), and what makes an arrangement of marks a word with
a meaning or a sentence expressing a certain proposition passimm).
These questions lead him to consider what the understanding of
a word or rule, as against behavior reflectingno such understanding,
consists in passimm,but especially sections I, II, III, V, VII, and X).
To take one particular question, crucial for all the rest: what can
make certainnoises intoan utteranceused to expressa given proposition,

521

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

ratherthan another proposition or none at all? In responding to this

question, Wittgenstein repeatedly makes a negative point familiar
to readers of the Investigations: What makes the utterance of sounds
into the use of words to express a proposition is not the simultaneous
occurrenceof a mental state or process (see pp. 85, 103, 148; see also
pp. 45, 50, 71, 74, 80, 82, 99, 100, 106, 107).
Here, Wittgensteinis, in part, combatting his own formerview, in the
Tractatusperiod, of what gives signs their semantic life. In his earlier
view, a physical fact functionsas a sentence describing a possible state
of affairsif its elements are projected onto objects, to which they then
refer.The method of projection is to think the thought which is the
sense of the sentence.2A thought,as Wittgensteinexplained to Russell,
is an arrangementof "psychical constituents,"the empirical properties
of which "it would be a matter of psychology to find out".3
The general view that a sign receives its sense through the occurrence
of a psychic fact was extremely common in the early twentieth-
century.Moore and Russell had shared it with the early Wittgenstein,
and so, in essence, had Frege.4 The episodes which were seen as giving
utterancestheir meanings were regarded as observable, in one's own
case, by some innersense, ratheras writtensentences are observable by
sight and spoken sentences by hearing. Moore, Russell, and Frege
were explicit on this point, and it is implicit in the previously cited
letterof Wittgenstein'sto Russell. I shall call the view that such episodes
make utterances expressions of propositions "the mental episodes
theory."
In the PhilosophicalGrammar, Wittgenstein'sdirect arguments against
the mental episodes theoryare of two main kinds. In the firstplace, he
distinguishesbetween the logical consequences of the fact that some-
one has used a sentence to express a proposition, and the logical
consequences of the fact that someone has made certain sounds while a
mental episode has occurred. In particular, Wittgenstein emphasizes
that the expression of a proposition requires the possession of general
linguisticabilitiesby the speaker,including an implicit grasp of the logic
and grammar of the language in question. Thus, he says, "To under-
stand a proposition is to understand a language. A proposition is
a sign in a systemof signs" (p. 131, see also pp. 63, 130). And he denies
that using a sentence to express a proposition could entail, as it must,
a masteryof syntax, if such a use simply consisted of the occurrence
3.11-3.12,3.2-3.203.
2Tractatus,
'Notebooks,1914-1916, pp. 129-130.
'See, for example, G. E. Moore, Some Main Problemsof Philosophy,pp. 57f.,
B. Russell, The Problemsof Philosophy,p. 51, G. Frege, "The Thought", Mind,
v. 65, pp. 308 and 310.

522

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

of a mental episode simultaneous with the utterance of the sentence

(see pp. 71, 132f.).These considerations might be illustrated as follows:
Someone cannot use the English words "The cow is brown" to express
the thought that a cow is brown, if he has not mastered a grammar
rulingout otherchains of words (forexample, "The brown is cow," "The
cow is brown all over and is not brown.") But the occurrence of a
mental episode along with the utterance of "The cow is brown" would
be compatible with the absence of such masteryof grammar. In general,
the use of words with a sense entails implicit masteryof grammar and
logic such as a simultaneous mental episode cannot guarantee. So
the use of words with a sense cannot be identical with the occurrence
of a mental episode.
Wittgenstein also criticizes mental episodes theories by arguing
that the states and processes of which they speak would, taken by
themselves,be as semantically dead as sounds or marks, taken in
isolation. In a mental episodes theory, the occurrence of an intro-
spectible state or process is supposed to make an utterance into the
expressionof a proposition.But, Wittgensteinargues, the occurrence of
such an episode along with an utterance does not constitute the expres-
sion of a certain proposition. It could be an episode in the expression
ofanotherproposition,or a mental event dissociated fromthe expression
of any proposition(see pp. 40, 70, 96, 143f., 152). Wittgenstein's point
can be illustratedas follows.Consider any mental episode the occurrence
of which might be taken to give the utterance of the words, "The cow
is brown," its meaning in English (forexample, having a mental image
of a brown cow.) That episode will turn out to be one which might be
associated with the expression of another proposition (for example,
the proposition that a cow is standing) or none at all (for example, a
mental image might simply have come before one's mind while no
proposition is being expressed.)
In his reflectionson the question "What gives a sign its life?", some
of Wittgenstein'snegative claims go beyond the rejection of the mental
episodes theory,and shed considerable light on his intentions in the
Thus, Wittgensteinappears to reject not merely the idea
Investigations.
that introspectiblestates and processes give meaning to signs, but the
idea that anystate or process does so:
The psychologicalprocess of understandingis in the same case as the
arithmeticalobjectThree.The word"process"in the one case, and the word
"object"in theotherproducea falsegrammaticalattitudeto theword,[p. 85].
We want to say "Meaning is surelyessentiallya mental process,a process
ofconsciousnessand life,notofdead matter."Butwhatwillgivesucha thingthe
specificcharacterofwhatgoeson?-so longas we speak of it as a process.And

523

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

nowit seemsto us as ifintendingcould not be any processat all, of any kind

whatever.-Forwhat we are dissatisfiedwith here is the grammarof process,
not thespecifickindof process.-Itcould be said: we should call any process
'dead' in thissense.[p. 148. See also pp. 45, 50, 80ff.,90f., 106f.,144].
Taken togetherwith similar remarks in the Investigation (for example,
remarks 307, 290, 471), these denials suggest that Wittgenstein saw
himselfas combatting a general misconception about meaning, which
might be expressed in the identificationof meaning with any state or
process,either in the speaker's mind or in the speaker's body.
Wittgenstein'sgeneral attitude toward state and process theories of
meaningcan, I think,be elaborated as follows.When philosophersspeak
ofsomeone's use of a termto mean somethingas the occurrenceof a state
or process, their use of the latter terms is an extension of ordinary
usage. To understandthe significanceof these philosophical characteri-
zations, we need to discover what analogies these philosophers assume
to hold between such expressions as "his meaning something" and
expressions that are normally said to be concerned with states or
processes,for example, "how he felt," "how his condition worsened."
Obviously there are many relatively superficial analogies, here. For
example, the fact that someone meant the color brown, the fact
that someone felt queasy, the fact that someone's pain grew worse
are all temporal. He might,on a subsequent occasion, mean blue, or feel
better.But the inadequate theories of meaning to which some philos-
ophers are attracted result from their having been led by the uncon-
troversialanalogies to assume yet deeper ones, where none is to be
found. In particular, philosophers talking about meaning sometimes
assume that one could, in principle, give a satisfyinganswer to the
question,"Why was his utteranceof thesewords a case of expressingthis
proposition?"by describingor indicatingsomethingthat has happened. 5
While this assumption can legitimatelybe made in the cases in which
"state" and "process" are ordinarily used ("His queasiness consisted of
his feeling dizzy and nauseous, at once", "His pain's growing worse
consisted of its spreading over his whole arm"), Wittgensteinwants to
deny that any satisfying answer to the philosophical question of
what gave a sign its life can consist of such a description of what
happened. In short,Wittgensteinholds that using words to express a
propositionis not a stateor process,forthe reason that a satisfyinganswer
to the question, "What did his using these words to mean this propo-
sitionconsistof?" cannot be a descriptionor indication ofsomethingthat
TForexamplesof such analyses,see the previouslycited passages fromthe
earlierWittgenstein, Russell,Moore,and Frege.A similaranalysisis suggested
byHume's theoryofideas and byJames'descriptions ofthenatureof thought
I, pp. 185ff.).
(see Principlesof Psychology,

524

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

has happened. The view that expressing a proposition is a state or

process is encouraged by similarities between the way we talk about
expressinga propositionand the ways we talk about ordinary processes
and states.If, guided by those similarities,one asks, "What occurrences
make utterances the expression of propositions?", one's answers will
be those theoriesof meaning that Wittgensteinattacks in the Grammar
and in later writings.
Because Wittgensteinrejects the notion that any state or process,
in a mind orin a body, gives an utterance its semantic life,the Grammar
contains attacks on the physicalist analogue of the mental episodes
theory,the idea that the occurrenceof a state or process which a physiol-
ogist might observe is what makes the utterance of sounds the ex-
pression of a given proposition. His underlying objections are the
same as those directed against mentalistic theories: any physiological
state or process could, in principle, occur in the body of an uttererof
sounds who lacks mastery of the grammar of the language in which
those sounds occur, or who is using those sounds to express a different
proposition from the one in question, or none at all. Hence, the use
of an utterance to express a proposition cannot be identical with the
occurrenceof a physiological state or process at the same time as the
utterance.[See pp. 144, 148, 189.]
Some readers of Wittgensteinhave regarded his attacks on mental
episodes theories of language as intended to prepare the way for a
behavioristtheory.In particular, the language-games at the beginning
of the Investigationsand the remark usually misquoted as "The meaning
of a word is its use"6 have suggested to some that Wittgensteinregarded
the use of an utterance to mean something as the production of an
utterance with appropriate intentions to influence the behavior of
others.Presumably, this analysis is understood as a firststep in a be-
haviorist project eventually defining intentions in a behaviorist
fashion.Otherwise,it is hard to imagine what the gains of the analysis
would be, since,as Wittgensteinwas well aware, philosophical problems
regardingthe meaning of words all have analogues in problems con-
cerning the content of intentions.
In the Grammar, Wittgensteinrejectsbehavioristaccounts of language,
and criticizes at length the equation of meaningful utterance with
behavior intended to influence behavior. Again, the appeal to a

much more qualified statementis: "For a large class of

6Wittgenstein's
cases-thoughnot forall-in whichwe employthe word"meaning" it can be
defined[the German,"erkidren",means also, more vaguely,"explain"] thus:
the meaningof a word is its use in the language" (Philosophical Investigations,
remark43. PassagesfromtheInvestigations be cited usingremark
will hereafter
numbersand, wherenecessary,the abbreviation"P.1.").

525

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

requirementof mastery of grammar and other specifically linguistic

rules is crucial. Someone who produces sounds to influence the
behaviorof othersmay produce utteranceswhich succeed in influencing
the behavior of othersin desired ways. But successfulutterance need not
be correctuses of words, any more than a rooster's successful call to
the hens is. In particular, although some combinations of sounds pro-
duced by a person may have the effectdesired and some may never
do so, the distinctionbetween meaningful and nonsensical utterances
need not apply.
Suppose I now made the discoverythat someone would bringme sugar at
a signplus thecry"Su", and would bringme milkat a signand thecry"Mi",
and would notdo so in responseto otherwords.Should I say that thisshows
that"Su" is the correct(the only correct)sign forsugar, "Mi" the correct
signformilk?
Well,ifI say that,I am not usingtheexpression"sign forsugar" in the way
it is ordinarilyused or in the way I intendedto use it.
I do notuse"thatis thesignforsugar"in thesamewayas thesentence"ifI press
thisbutton,I get a piece of sugar". [p. 188]
Correct/incorrect,meaningful/nonsensical are not always applicable
distinctionswhere one can speak of the success or failure of the use
of utterances to influence the behavior of others. And, Wittgenstein
proposes, nothing counts as a use of words to express a proposition
unlessit is correctand meaningfulas against incorrectand nonsensical.
(See pp. 187ff.,69f.)
In the remarks discussed so far, Wittgenstein's intentions are,
broadly speaking, anti-reductionist.As he repeatedly says, "Language
must speak for itself" (see pp. 40, 63; see also pp. 67, 97). Since the
publication of the Grammarin the original German, I have heard it
said, on several occasions, that Wittgensteinmeant, in this remark, to
deny that descriptions of what words refer to describe connections
betweenlanguage and extra-linguisticreality. But there is no reason to
suppose that any such linguisticsolipsism is intended. Rather, Wittgen-
steinis rejectingthe philosophical tendencyto seek a reductive answer to
such questions as, "What makes this utterance an instance of the
expression of a proposition?", that is, an answer consisting of terms
which do not, when taken in isolation, speak of meaning, referring,
intention, or the like. Mental-image and behaviorist theories of
language are examples of such attempts to describe linguistic
phenomena without any reliance on intrinsicallysemantic concepts.
In Wittgenstein'sview, philosophers are drawn to the construction of
such theories because they feel that the possibility of something's

526

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

expressinga propositionwould be remarkable,even mysterious,if

thereduction ofthesemanticto thenonsemantic could notbe achieved.'
So far,I havediscussednegativeremarkson thetopic,"What givesa
sign its meaning?",remarksin which Wittgensteinattacks various
ways of answeringor understandingthe question. But at times,he
also sketchespositionanswersto this question. His intentionis not
to prepareforsome reductionof semanticconceptsto nonsemantic
ones.To thecontrary, his answersto thequestion,"What givesa sign
itsmeaning?"are meantto servetwoanti-reductionist purposes.They
aremeantto bringto theforeaspectsoflanguagewhichmentalepisode
theoriesand otherreductionist viewsare bound to neglect.And they
are meantto mitigatethe feelingthat meaningis mysteriousunless
it can be reducedto some combinationof nonsemanticphenomena.
In the Grammar, Wittgenstein's positiveaccount of the workingsof
languageis,I think,inadequateand one-sided.Here,in contrastto the
he reliestoo exclusivelyon the requirementof guidance
Investigations,
bylinguistic rules.At thesame time,he neglectsan aspectof language
whichhe was laterto regardas crucial,the use of wordsas tools.
At severalpointsin the Grammar, Wittgenstein suggeststhat sounds
or marksthat would otherwiseseem "dead" expressa proposition
becausetheirproductionis governedby a systemof ruleslikethe rules
definingthe class of meaningfulsentencesin English. He says, for
example,
The appearance of the awkwardnessof the sign in getting its meaning
across,like a dumb person who uses all sorts of suggestivegestures-this
disappearswhenwe remember thatthesigndoes itsjob onlyin a grammatical
system.
7The notion that language and thought are mysterious if such a reduction
cannot be accomplished is vividly expressed by Frege in the following passage:
The lifelessadded to the lifelessjust producessomethinglifeless.An image [ Vorstellung]
linkedto an imagejust producesan image,and no amountofingenuityand complexity
concerning associationscan change that.And if feelingsand attitudesare piled on to
all this,it changesnothing.. .'Yes, but the graspingof. . [a proposition,
e.g. the law of
gravitation] is a mentalprocess!Right.But it is a processwhichlieson the borderof the
mental,and yetwhichcannot be understoodfroma purelypsychologicalstandpoint,
since there is somethinginvolved which is not strictlymental: the proposition
[derGedanke].This processis, perhaps,the mostmysterious of all."
(Schriftenzur Logik undSprachphilosophie: Aus demNachlass, G. Gabriel (ed.), pp.
63f.). Frege, refusingto reduce the semantic to the nonsemantic, feltthis refusal
obliged him to regard the semantic as a mysteriousrealm. Other philosophers,
beforeand since, have been less tolerant of mystery,and have chosen psychol-
ogistic or behavioristic theories as necessary means of escaping the "mystery"
Frege embraced. For Wittgenstein, both Frege and his opponents share a
common delusion, that the semantic is mysterious unless it is reducible to the
nonsemantic.

527

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

You mightas it werelocate (look up) all of the connectionsin the grammar
of thelanguage.There you can see the whole networkto which the sentence
belongs.(p. 149).
'I arrivein Viennaon the24thofDecember.'Theyaren'tmerewords!Of course
not: when I read themvariousthingshappen inside me in addition to the
perceptionof the words:maybe I feeljoy, I have images,and so on.-But I
don'tjust mean thatvariousmoreor less inessentialconcomitantphenomena
occurin conjunctionwiththesentence;I mean thatthesentencehas a definite
senseand I perceiveit. But thenwhat is thisdefinitesense?Well, that this
particularperson,whomI know,arrivesat suchand sucha place etc. Precisely:
whenyou are givingthe sense,you are movingaround in the grammatical
backgroundof the sentence.You're lookingat the various transformations
and consequencesof the sentenceas laid out in advance; and so they are,
in so faras theyare embodied in a grammar.(You are simplylooking at
thesentenceas a move in a givengame.) [p. 153].
Part of what Wittgensteinis saying in such passages is surely right.
If one wonders how the human utterance of words can constitute
the expression of a proposition while a rooster's calling to the hens
does not or a baby's crying does not, one good answer is that the
utteranceof the relevant sequences of words reflectsmasteryof a syntax
permittingsome combinations of elements and excluding others. By the
same token,awareness that a simultaneous mental episode would not
make an utterance into the expression of a proposition is, in part, a
responseto the factthat such a state or processcould occur on the part of
someone who has not mastered a syntax.
Still, someone might not be using sounds to express propositions,
even thoughhis production of sounds is the resultof masteryof a system
of rules excluding some combinations and permitting others, such
as the rules of English. Suppose we teach someone who has no initial
idea of the meanings of English words to play a game in which he is
to produce stringsof sounds constructibleby rules R, a set of rules in-
cluding rules R', describing permissible substitutions. The strings of
sounds constructibleusing rules R are, in fact,the meaningful sentences
of English,and stringsinterchangeable according to R' are synonymous
in English. If the student learns to play the game, need he use words
to express propositions in the game? Not necessarily. Suppose the
student is also taught that the rules he is given are the rules of gram-
maticality and synonymy in a language. Need his mastery of the
rules then result in his use of words to express meanings? Again, the
answer seems to be "no." The pupil may be prepared to reject "This is
red all over and also blue all over," but he may still need to be told
what red is.
At times, in the Grammar, Wittgenstein seems to take such phrases
as "grammatical system" to include rules less formal than the ones to

528

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

which I have alluded, for example, the rule that "red" stands for the
color, red. If "grammatical system" is understood this broadly, then an
utterance is the expression of a proposition if it reflects mastery of
a grammatical systemsuch as that governing English and other natural
languages. But the breadth of this appeal to grammatical rules severely
limitsitsusefulnessin combatting mental episodes theoriesof language.
For the question, "What makes an utterance of 'red' into an utterance
guided by the rule 'Use "red" to stand for the color red'?" remains
troublingand unresolved.Unless more is said about the use of a word to
stand for something, one might feel that what makes an utterance
intoa descriptionof a state of affairs,ratherthan a mere chain of sounds
produced in accordance with formal rules, is the occurrence of certain
mental episodes connecting words with their referents.Wittgenstein
held such a view at the timeof the Tractatus.8Writingunder the influence
of the Tractatus,Schlick proposed that mental episodes are responsible
for signs' having content, as against mere grammatical form, and
argued for the conclusion that the content of the propositions we
express,as against their form,is uncommunicable.9
Occasionally, Wittgensteinalso proposed that someone uses words
to expressa proposition ifhe knows how to go about verifyingthe truth
of the propositiontheyexpress. But such connections between language
and verificationare at once too restrictiveand too underdefined to
remove the feeling that language is mysterious. On the one hand,
words are sometimes used to express propositions in situations in
which no method of verifyingthe proposition expressed is in question.
As Wittgensteinlater emphasized, this is characteristicof assertions that
one is in pain. The "methods of verification"that philosophers speak of,
here,are, in his later view, empty ceremonies (P.!., 246, 288-290, 311).
On the other hand, where the connection between meaning and
verificationis less controversial, it is unhelpful. For mental episodes
theoriesof language can arise, as they do for Schlick, from the effort
to distinguishverification,that is, the recognition that a state of affairs
makes a certain sentence true, from the recognition that the logical
"We use theperceptiblesignof a proposition(spokenor written,etc.) as a
projectionof a possiblesituation.The methodof projectionis to thinkof the
senseoftheproposition" ( Tractatus,3.1 1)."... I don'tknowwhattheconstituents
ofa thoughtarebutI knowthatitmusthavesuchconstituents whichcorrespond
to thewordsofLanguage.Again the kindofrelationof theconstituents of the
thought and ofthepicturedfactis irrelevant. It wouldbe a matterofpsychology
to findout . . . 'Does a Gedankeconsistof words?'No. But of psychicalcon-
stituentsthat have the same sort of relationto realityas words" (letter to
Russell,8/19/19,in Notebooks,1914-1916, pp. 129f.).
9"Formand Content"in M. Schlick.GesammelteAufsaetze,Gerald, Vienna,
1938.

529

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

formof a state of affairsis the same as the grammatical form of the

sentence.Indeed, ifwe had an adequate grasp of the differencebetween
verifyinga proposition, on the one hand, and simply experiencing
certain feelingsof satisfactionand utteringcertain sounds in the face
of certain facts, the distorted views of language which Wittgenstein
seeks to combat would not arise in the firstplace.
In Wittgenstein's later discussion of meaning in the Investi-
gations,there was to be an important shift in emphasis. In the
later work, Wittengenstein still regards it as essential to an ut-
terance's expressing a proposition that the utterance reflect
mastery of a system of signs with a syntactic complexity similar
to that of English, German, and other actual languages (see, for
example, P.1. 493 f.). But his appeals to this fact are relatively
rare and scattered. Rather, his repeated emphasis is on the use of
words as tools, to help others, to obtain help from others, to express
feelings, to gain mastery over nature, and in similar ways. It
seems to be Wittgenstein'slater view that one uses words to express
a proposition if one's utterance reflects mastery of a system of
signs with an appropriate formal structure and if one produces
these and the other utterances governed by the system of signs for
appropriate ends and purposes. (See P.!., 244, 270, 291, 432, 569).
No general specification of the ends and purposes appropriate to
language is or can be given, though the everyday practices of
actual users of language is taken as providing the relevant
paradigms (see 108, 494). Thus, to use a previous example,
someone's utterance, "The cow is brown," is an expression of a
proposition if he uses this utterance in determining whether a
prospective buyer's wants are satisfied, if his utterance reflects
mastery of the syntax of English, and if in making utterances
governed by this syntax he is typically using these utterances for
purposes to which English-speakerscommonly put them.
This later view of language might seem to be incompatible with
Wittgenstein's notion, in the Grammar,of the autonomy of lan-
guage and his denial that meaningful usage is a process or state.
But there is no incompatibility here, for several reasons. For one
thing, one is not preparing to define away specifically linguistic
concepts, since one cannot define the kinds of ends and purposes
necessary for meaningful usage or the kind and degree of syntactic
complexity that is essential. Also, the concept of the use of a
word for a purpose involves reference to intention, and is a
semantic concept in the broad sense I sketched. A purpose
has a content (for example, "to get someone else to buy one's

530

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

cow"), and the possession of such content cannot, in Wittgenstein's

view, be defined in nonintensional, for instance, behavioral terms
(see P.!., 439-445). Finally, using a word for a purpose is not
a process or state. Just as Wittgenstein argues that subsequent
unexpected behavior might reveal the possibility that someone
was not using "cube" or "plus" with the usual meaning, despite
whatever process or state accompanied that use (P.I., 139-155), he
might have argued that subsequent unexpected behavior might
reveal the possibility that someone was not using words for the
purposes he seemed to, despite whatever process or state had
occurred.
Without a one-sided identification of meaning with syntax, the
instrumentalistview of meaning in the Investigations brings to the
fore aspects of language which are incompatible with mental
episodes theories and other reductivist accounts. Thus, if what
gives an utterance its content were a mental episode, the expression
of this content could not consist of the use of words as tools in
the way I previously sketched. For anything that happens when the
utterance is produced need not happen in the mind of someone
who uses words as tools in the relevant ways. For example, no
amount of description of the role of the utterance "I am in pain"
as a replacement for crying will require a mental episode of
meaning pain to take place when someone refers to pain, in
making that utterance (P.I., 244, 271). Similarly, the role of
numeralsin buying,selling,record-keeping,and so on can be described
withoutreferring, even implicitly,to introspectibleepisodes of meaning
in enumerators'minds (see P.I., 1).
In the Grammar, Wittgenstein rarely refersto the use of words as
tools,except when he is attacking theorieswhich regard such use as the
sole requirementformeaningful utterance. He does sometimes speak of
the "role in human life" of a word. But in his most explicit
discussion of the relevance of this role to questions of meaning, he
says that this role only determines the meaning of a word "in the
sense in which one speaks of 'the meaning of an event for our lives',"
and denies that the role of a word in our lives is important for the
questions of meaning he intends to examine (see p. 68). This
differencebetween the Grammarand the Investigationsis especially
poignant in those cases in which remarks in the earlier book resemble
remarksin the later one, but differprecisely in the attitude expressed
toward the role of words in human life. Consider the following near-
parallels:
Grammar:It is only in a language that somethingis a proposition.To
understanda propositionis to understanda language.
531

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

A propositionis a sign in a systemof signs. It is one combination of

signsamonga numberof possibleones,and as opposed to otherpossibleones.
As it were onepositionof an indicatoras opposed to other possible ones.
[p. 1311
Investigations:
To understanda sentencemeans to understanda language.
To understanda languagemeans to be masterof a technique.[?199]
Grammar:The use of a word in the language is its meaning.
Grammardescribestheuse ofwordsin thelanguage.So it has somewhatthe
same relationto the language as the descriptionof a game, the rules of a
game,have to the game. [p. 601
Investigations:
Fora largeclass ofcases-thoughnot forall-in whichwe employ
theword"meaning"it can be definedthus: the meaningof a word is its use
in the language.[No equation of use withgrammarfollows.][?1991
Grammar.Augustinedoes describea calculus of our language,only not every-
thingthatwe call language is thiscalculus. [p. 571
Investigations:
[withreferenceto thesame passage fromAugustine]:Augustine,
wemightsay,doesdescribea systemofcommunication; onlynoteverything
that
we call languageis thissytem.[p. 31
A contrastbetween the Grammar and the Investigations
can be drawn as
follows.It is compatible with everythingsaid in the earlier work that a
being who never uses words as tools, to communicate, to cope with
nature, to solve problems, or in any other way, could nonetheless
use words to express propositions. Wittgenstein'slater view that "Lan-
guage is an instrument. Its concepts are instruments" (P.I., 569)
excludes thispossibility.Many readers will probably feel,as I do, that in
this respect the Investigations
marked an important advance.

"Part II: On Logic and Mathematics"

In thesecond partofthePhilosophicalGrammar, Wittgensteindefendsan

approach to mathematicssharplyat variance fromthe traditional,"Pla-
tonist" view that mathematical propositions describe, truly or falsely,
the relations between extra-human, specifically mathematical objects.
"Mathematics," he says,"is a calculus, and hence isn't really about any-
thing (p. 290) . . . You can't get behind the rules [of logical
inference] because there isn't any 'behind'. (p. 244) . . . rules
[of inferencein logic and mathematics] must be laid down arbitrarily,
that is,are not to be read offfromrealitylike a description (p. 246)." In
context,theseand similarbold assertionsseem to amount to the following
view of mathematical truth:mathematical signs express a true proposi-
tion only in that one's acceptance of them is a logical consequence
of one's commitment to accept the results of certain rules for con-
structingformulae. One's acceptance of such rules is not validated or
532

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

invalidated by the correspondenceor lack of it between these rules and

a furtherreality.
The approach to mathematics I have just sketched is incomplete in
importantways. What is the nature of "logical consequence"? What is
thedifferencebetweena logical consequence of commitment to a system
of rules and a step that a person or people committed to the system
happen to findpsychologicallycompelling? Why would it be more than
a potentiallydangerous idiosyncracyif an otherwisenormal speaker of
English were to teach his child, "Twice two is five"?-Some positive
answers to such questions are necessary,if this anti-Platonist approach
is to be satisfying.In the Grammar, as we shall see, Wittgensteinsuggests
that the answers are to be found by examining the roles and the uses
that mathematical operations have in our lives. But he does not present
the sketches of these roles and uses that occur in the Remarkson the
Foundationof Mathematics.In this respect, Part II, like Part I of the
Grammar,is transitional and preliminary. Yet, again like Part I, it
contains some arguments and assertions which supplement and illu-
minate Wittgenstein'slater work.
For convenience's sake, I shall referto Wittgenstein'sview of mathe-
maticsas "Wittgenstein'sconventionalism." Though his view differs,for
example, fromtheconventionalism of Carnap, thereis a common nega-
tivethrust,the denial that mathematical propositionsare made true by
the relationsand propertiesamong extra-human objects, as botanical
propositionsare made true by relations and properties among plants.
It is also convenient to have a general term for propositions
which are accepted, and properly so, even though they are not vali-
dated (or falsified)by relations and propertiesof objects to which they
refer.Following an occasional usage of Wittgenstein's,I shall call such
propositions"stipulations."
Most people who have reflected on the nature of knowledge and
understanding in mathematics have felt some attraction to Wittgen-
stein's conventionalism. If the truth of mathematical propositions
depends on facts concerning mathematical objects, the existence and
interrelationsofwhich are independent of human activities,it is a puzzle
how humans can have the mathematical knowledge they are usually
supposed to possess. Traditional appeals to self-evidence seem inade-
quate to explain how one can have knowledge of mathematical truths.
Self-evidence,afterall, is a feeling.Why should it be more than an acci-
dent if the feelingsof human beings concerning hypotheses about an
extra-human mathematical reality should be a reliable guide to the
truth?On the other hand, the appeal to past experience as justifying
mathematical beliefs is inadequate to account for the immunity from
empirical disconfirmationgenerally accorded mathematical proposi-
533

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

tions. For example, our past successes in applying the principle that
two plus two are four could not explain the special absurdity of
claiming to "test" this principle by counting a new pair of pairs of
apples (see p. 307). Finally, the possibility of merely understanding
importantmathematical propositions can seem mysteriouson a non-
conventionalistview. Thus, Cantor, on the basis of a famous applica-
tionof the "diagonal method", assertsthat thereare more real numbers
than there are rational numbers. Understood as a proposition about
entitieswhich would exist apart fromhuman activities, this conclusion
seems to require that one infinitetotality of objects be greater than
another. But can we understand the notion of infinite totalities dif-
feringin extent? It is sometimes supposed that Cantor showed us the
way to such understanding by showing how, given an infinite list
of real numbers, a number can be constructed which is not on the
list. But the proposed explanation is only as intelligibleas the'notion of
a completed infiniteconstruction which proceeds in finite steps. It
has seemed unclear to many that we can grasp such a notion.
If Wittgenstein'sconventionalism is accepted, these problems lose
much of theirforce.Mathematical propositionscannot be disconfirmed
or falsifiedby a deceptive or surprisingreality,not because we have an
intuitiveor experimentalguarantee of theirtruth,but because they are
stipulationsto which no furtherrealityneed correspond,any more than
the rules of chess need correspond to the actual rules of war (see p.
291). The assertionthat thereare infinitelymany numbers of a certain
kind is simplytheassertionthat the correspondingsystemof stipulations
excludesthestatementthatsomethingis the last or the highestnumber of
that kind (see pp. 282, 321, 466f.). Comparisons of infinitetotalities as
greateror smaller than one another express, in somewhat misleading
form,comparisons between differentsystemsof propositions we have
adopted. For example, Canter's diagonal method shows that in the
systemsof propositionsdefiningour concept of a real number, as against
the systemsdefiningour concept of a rational number, there is no need
to suppose that some method of generating numbers will generate any
given number in a finite number of steps. We are thus free (though
not compelled) to stipulate as Cantor did, that there are reals that are
not accessible in the latter fashion (see pp. 287, 464, and Remarkson the
Foundations ofMathematics, Part I, appendix 2).
IfWittgenstein'sconventionalismseems to have distinctadvantages in
dealing with knowledge and understanding in mathematics, it seems to
createspecial problemsconcerningthe existenceof truthin mathematics.
We speak of truemathematical propositions(for instance, that 25 X 25
equals 625, against the falsehood that 25 X 25 equals 645). Wittgenstein
accepts this way of speaking. And yet how can a proposition be true
534

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

if it is not madetrue by relations and propertiesof objects to which it

refers?Most of the interest of Part II of the Grammarresides in
Wittgenstein'seffortsto clarifyand to overcome worriesconcerning the
status of mathematical truth within his conventionalism.
One obvious objection to Wittgenstein'sconception of mathematics is
that it reduces mathematics to a mere game of sign-manipulation,
deprivingus of the rightto speak of mathematical signs as expressing
propositionsat all (and, a fortiori,of the rightto speak of theirexpressing
truepropositions).Wittgensteindevotes a section of Part II (section I 1)
to "the comparison betweenMathematics and a game." He insistson the
importantsimilaritiesbetween mathematics and highly rule-governed
games such as chess and between mathematical problems and such tasks
as chess problems. But at the same time, Wittgensteinagrees that the
analogy is incomplete. For mathematics requires not merely rule-
governedmanipulationsofsigns,but applications ofthesemanipulations
outsideof mathematics.A game withnumerals governed by arithmetical
rules such as our own need not be counted as arithmetic, unless
thisactivityhas furtherapplications of appropriate kinds (p. 292). By
"applications" Wittgensteinmeans practical applications which give an
activity"its importance forlife" (p. 294). Thus, in distinguishing him-
selffrom"formalists",who regard mathematics as nothing more than
a game of sign-manipulation,Wittgensteinmakes his closest approach
in the Grammarto his later insistence on the role of words as tools,
But even though there would be no mathematical
in the Investigations.
propositions mathematicsdid not have an appropriate kind of impor-
if
tance forour lives (in counting, measuring,keeping records,or the like),
a mathematical proposition,Wittgensteinsays, does not include, as part
of its meaning, a description of any part of the practical impor-
tance of mathematics. He regards the failure to note this distinction
as creatingmuch confusion in disputes over formalismin mathematics.
"People cannot separate the importance, the consequences, the applica-
tion of a factfromthe fact itself;theycan't separate the description of a
thing fromthe description of its importance" (p. 295).
Wittgenstein'sinsistenceon the connection between mathematics and
practical activities makes his conception of mathematical truth less
liberal than it first appears to be. If one of us English-speakers,
inspiredby an oracular "revelation," feltcompelled to say, "Two plus
two equals five,"his utterance might not be the expression of a mathe-
matical proposition,much less a true one. For it might lack appropriate
connectionswithpractical activities.Anotherconstraintsuggestedby the
discussionof mathematics and games is structural: a sequence of signs
would expressa mathematical proposition only if it is produced as part
of a proceduregoverned by rules with sufficientformal resemblance to
535

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

thosegoverning our actual mathematicalactivities,thatis, our actual

proceduresof calculation, mathematicalproof,and mathematical
problem-solving.In the Grammar, the discussionof theseconstraints
is vague and fragmentary.But it is clear thatWittgensteinwould use
themto deny that a varietyof alleged counterexamplesdo, in fact,
expresstruemathematicalpropositions.
In making the meaning of mathematical signs depend, in part, on
Wittgenstein
theirpracticalsignificance, isbreakingwitha fundamental
assumptionof his earlierphilosophy,namely,the assumptionthat the
existenceof a propositionwitha particularsensedoes not depend on
factsabout theworld(see Tractatus,2.021If.) If the worldwerein flux
in sucha way thatcountingand measuringcould play no role in our
lives,then,the remarksin the Grammar imply,we could not express
mathematical propositions."Mathematics"would,at best,be a game.
Thus,evenifcontingent factscannotfalsifywhatwesayin mathematics,
theycoulddepriveour utterances oftheirmeaning.This breakwiththe
Tractatus, (see remarks
whichisemphaticand explicitin theInvestigations
142,240-2), is onlyimplicit,however,in the Grammar. Here, then,is a
further characterof the latterwork.
signof the transitional
Anotherobjectionthat mightbe raisedagainst Wittgenstein's con-
ceptionofmathematics concernstheexistenceof mathematicalpropo-
sitionswhich are undecidable in the sense that they are neither
entailednorexcludedbycurrently acceptedsystems ofmathematics. It is
oftensupposedthattherecan be undecidablepropositionswhichare,
nonetheless,eithertrueor false.This positionis especiallyattractive
giventhe statusof the problemswhich are the subjectsof the most
creativemathematicalresearch.In such research,mathematicians
pursuethe question,"Is it the case that p?", where "p" expresses
Fermat's"Theorem,"Goldbach'sConjecture,or someotherproposition
whichmight,forall thatiscurrently known,turnout to be undecidable.
Yet itseemsthatforsuchresearchto be theinvestigation ofa questionor
a hypothesis,mathematicians mustalreadyknowthatthereis somefact
of thematteras to the truthor falsehoodof the relevantproposition,
a factwhichalreadyexistsand is waitingto be discovered.In this
respect,open questions in mathematics can seem like open questions
typical of the natural sciences, for example, "Has there been life on
Mars?", "Do viruses cause any human cancers?" This possibility that
mathematical whileundecidable,should be eithertrueor
propositions,
conventionalism.
withWittgenstein's
false,isincompatible Forsuchtruth
orfalsehoodwouldnotbe basedon ouracceptanceofa systemofpropo-
sitions.
likemathematicians
Wittgenstein, school,regards
ofthe"intuitionist"
the admissionof undecidable mathematicalpropositionswhich are
536

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

eithertrue or false as a mistake, an unwarranted extension of the law

of the excluded middle into inappropriate terrain (see pp. 367f.). He
sees that our ways of speaking about creative research in mathematics
encourage the admission of the possibilities he seeks to exclude. His
defense of the exclusion of undecidable mathematical truths largely
consistsof a discussion of these ways of speaking, a discussion which
is meant to reduce our inclination to suppose that a question or con-
jecture formathematical investigationmustconcern a proposition which
is either true or false.
To divorce the most creative mathematical research from the ascer-
tainmentof whethera proposition or its negation is true, Wittgenstein
needs to explain how our talk of "questions" and "hypotheses" can
be appropriate, where the object of our research may be undecidable
and, hence, cannot be assumed to be either true or false. On the one
hand, Wittgensteinaccepts that numerous similaritiesbetween research
in mathematics and research in the natural sciences justify our using
many of the same terms to describe the two enterprises.Thus, where
"p" may turnout to expressan undecidable mathematical proposition,
"Is it the case that p?" can still be viewed as a question in that it
serves as a stimulus for research (p. 371). In creative mathematics,
as in natural science, calling something an hypothesis "trains your
thoughtson a particular object-I mean a particular region-and we
might say 'we shall surely discover something interestingabout these
things'" (p. 359). The social framework in which "open questions of
mathematics"are pursued also resemblesthat in which open questions in
natural science are investigated. Thus, reminding us of one rather
mundane parallel between research into Fermat's "theorem" and, say,
researchinto the existenceof lifeon other planets, Wittgensteinreminds
us that prizes might be offeredforthe solution of Fermat's problem (p.
362).
These parallels between mathematical and scientificresearch make it
temptingto furtherassimilate the two realms of investigation and to
suppose that any open question in mathematics, like a typical open
question in the natural sciences,must have a trueor false answer,waiting
to be discovered.But these parallels, while justifyingthe common appli-
cation of "question," "hypothesis,"and similar usages, do not forceus to
this furtherconclusion. And, if Wittgenstein'sconventionalism is right,
the requirementthat every question investigated in mathematics have
a right answer can only lead to confusion. For an open question in
mathematicswould have the same logical status as the question "How
can white win in twenty moves?" asked of a game which is still in
the process of being invented, in which relevant rules forwinning may

537

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

not have been laid down (p. 363). Both questions can, in an appro-
priatesetting,stimulate research,but neitherneed have a rightanswer.
An opponent of Wittgenstein'sconventionalism will, of course, accept
that the mathematician is often engaged in working out the con-
sequences of prior commitment to a system of mathematical propo-
sitions. The most acute disagreements between Wittgenstein and the
"Platonist" will concern the interpretationof those special junctures
in the acquisition of mathematics at which people come to accept new
propositions not entailed by those they have previously accepted.
According to the Platonist, if a mathematical deviant were to reject
thenovel propositionsothersaccept, eitherthedeviant or the othersmust
have a false beliefabout the world of mathematics. From Wittgenstein's
perspective,there need be no fact of the matter as to who is wrong.
There are many different junctures at which all or most mathemati-
cians do, in fact, accept mathematical propositions not entailed by
propositionsthey have previously accepted. In many such cases, it is
quite tempting to say that either the majority or the deviant would
have to be wrong, in case of dissent. At least for representative
examples of such junctures, Wittgensteinneeds arguments to show that
this Platonist demand is invalid. However, since these junctures are
extremelydiverse (consider: "two plus two are four," the least upper
bound axiom, the existence of nondenumerable cardinalities), there is
no reason to require a single convincing Wittgensteinian argument,
applicable, mutatis on all occasions. In the Grammar,
mutandis, Wittgenstein
concentrateson defendinghis nonobjectivistview of one such juncture,
the acceptance of proof by mathematical induction as a valid proof
procedure.
If people progressfar enough in the standard curriculum in mathe-
matics, they learn the technique of proof by mathematical induction,
or,as Wittgensteinusually calls it,"recursive proof." Roughly speaking,
in such proofs one establishes that a proposition is valid for all
cardinal numbers by establishing that it is valid for the number one,
and that, if valid for a given number n, it is valid for the next
largernumber. Recursive proof is essentially differentfrom techniques
of proof that one learns earlier (see p. 360). In particular, the validity
of recursiveproof will not, typically,be entailed by the mathematical
rulesand propositionsone has previouslyaccepted. Thus, if a refusal to
accept recursive proof as valid would be a failure to grasp a
mathematical truth,such truth cannot, as Wittgenstein proposes, be
determined solely by the mathematical stipulations one has already
accepted.
In section VI of part II of the Grammar,"Inductive Proofs and

538

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

Periodicity", Wittgensteinargues at length that a refusal to accept

recursiveproofsas valid need not reveal inability to grasp mathematical
truths.Of course,even the nonrecursivist,as I shall call the person who
rejectsrecursiveproofwhile accepting parts of mathematics which are
traditionallyencountered earlier, must accept that a recursive "proof"
is an interestingdiscovery.It establishes interestingconnections between
algebraic propositionsand propositionssolely concerned with particular
numbers (p. 434). For example, a recursive "proof" of an algebraic
equation providesa means of generating a seriesof proofsof numerical
instances of the equation, a series which cannot have a last possible
member (see p. 399). But the nonrecursivistis not being unperceptive,
even if he is being unusual, in refusing to take the further step
fromthisclaim to theconclusion that the "proof" establishes the validity
of the equation for all cardinal numbers. Of course, "arguments" are
given to get us to accept recursive proof as valid. We are asked to
think of a process reaching out to all numbers successively. (My
veryperspicacioushighschool math teacher,Margaret Cotter,persuaded
us with a story about an infinite row of tin soldiers successively
knockingeach otherover.) But thereis nothingunreasonable in denying
that a process can, in fact, continue until it encompasses an
infinite totality. As Wittgenstein puts it, "that one can run the
number series through the rule is a form that is given . . . Running
the stream of numbers through is not something which I can say I
can prove" (p. 434). Thus, the acceptance of recursive "proof"
as actually establishing algebraic laws as true for all numbers is
a stipulation which one may make or decline to make (p. 449). The
recursivistand the nonrecursivisthave two concepts of proof, neither
of which restson a mistake (pp. 417, 419f.).
Aside from making more plausible Wittgenstein's conception of
mathematical truth, Part II of the Grammarsheds light on a
claim concerning meaning in mathematics which has puzzled many
readers of the Remarkson theFoundations of Mathematics-theclaim that
the proof of a mathematical proposition determines its meaning (see
RFM, pp. 52, 76, 164f.). In the passages I have surveyed, Wittgenstein
makes this point repeatedly, saying, for example, "the proof belongs
to thesenseof theproved proposition,i.e., determinesitssense" (p.375; cf.
pp. 350, 366, 373f., 378, 453, 458).
In the Grammar,when one reads these remarks in context, their
meaning turns out to be less crude and their motivation less simple-
minded than one might otherwise have thought. In the firstplace,
Wittgensteinturns out to be using "sense," "meaning" and related
expressionsin a relatively broad fashion. At one point, he says that

539

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

"the word 'attempt' has quite differentmeanings" depending on

whetherone is speakingof an attemptto do somethingby a certain
means,as in attempting to lifta weightwitha hand,or speakingof an
attempt to accomplish something in whichone beginsby searchingfor
themeansto employ,as in attemptingto wiggleone's ears forthe first
time(pp. 393f.).In comparingthe personagesI have identifiedas the
nonrecursivist and the recursivist,he says that they use "proof'
withdifferent senses(pp. 417 and 419). Clearly, the differences of
meaningand senseWittgenstein has in mindare not of the same sort
as one indicatesin remindingsomeonethat"bank" has two different
meanings.
Also (thoughhe vacillateshere),Wittgenstein does not,in the final
analysis,appeal to a general verificationist thesis identifyingthe
meaningand the methodof verification of everyproposition,as his
basis forconnectingthe proofof a mathematicalpropositionwith
its meaning.Indeed, in several passages, he says that propositions
outsideof mathematics, forexample,"This man died two hoursago"
(p. 370) and "There is no one in the next room" (p. 462), can be
verifiedin essentially differentways withoutany change of sense.He
explicitlydistinguishes thesecases in which difference in verification
requiresno difference in meaning fromthe case of propositionsin
mathematics.
The remarkthattheproofofa mathematicalpropositiondetermines
itssensemightbe elaboratedas follows:to knowwhat a statementin
mathematics meansis to knowhowto proveitor howtocheckit(thatis,
howto proveeitheritor itsnegation)or how to use it as an axiom in a
procedureforprovingotherpropositions.An explanation that does
not enable one to provea mathematicalproposition(or to check it
or to use it as an axiom-alternativeswhichshould,fromnow on, be
understood) does not teachone its meaning.There is no awarenessof
the meaningof a mathematicalpropositionthat someone can have
apartfromtheknowledgeofhowto proveit.Once someonehas learned
how to prove a mathematicalproposition,there is no furtherfact
relevantto its meaningwhichneeds to be conveyed.
As Wittgenstein pointsout on severaloccasions,reflections on our
actualpracticein teachingarithmetic makehisconceptionoftheroleof
proofin graspingmeaninga naturalone to adopt. We teach children
what"5 plus7 equals 12" meansbyteachingthemhowtoworkoutsums.
Thereis no evidentway of separatingthisprocessinto phases of two
differentkinds,in one of whichwe conveythe meaningsof sums,in
theotherof whichwe conveythe abilityto do sums."If you want to
knowwhat2 + 2 = 4 means,you have to ask howwe workitout.. . In

540

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

everyday life we don't work out 2 + 2 = 4 ... But of course we could

. . . and children in fact do so by counting off.Given the sequence of
numbers 12345 the calculation is 1 2 1 2" (p. 333). "The content of
1 234
5 + 7 = 12 is precisely what children find difficultwhen they are
learningthis proposition in arithmetic lessons" (p. 347). What children
find difficultis, of course, learning how to work out sums and check
them in the systemof calculations which has "5 + 7 = 12" as a part.
Apartfromthesereflectionson language learning,the main motivation
forWittgenstein'sconception of meaning in mathematics is, I think,his
rejectionof Platonism, on the one hand, and instrumentalist, on the
other.The rejectionof Platonism excludes accounts of how the meaning
of mathematical signs is learned which makes such learning depend on
some formof contact between the learner and mathematical objects. As
we have seen, Wittgensteinalso rejecs instrumentalist, that is, the view
that learning the meaning of mathematical signs consists of learning
how to apply them in practical activities through which nonmathe-
matical objects are altered. While mathematical signs would lose their
meaning if mathematics were to lose all of its practical significance,
grasping the meaning of a mathematical sign is differentfrom under-
standing its practical applications.
From this anti-Platonist, anti-instrumentaliststandpoint, it is hard
to see what knowledge of the meaning of mathematical propositions
could consist of, if it does not consist of knowledge of how to prove the
proposition.No plausible alternative seems to be available. By the same
token,a Platonist will be inclined to deny that knowledge of meanings is
knowledge of proofs,and to claim that grasping the sense of a mathe-
matical propositionsometimesdepends on perceivingwhat the referents
of the component terms are like, in a process resembling the under-
standing of the referent,say, of "red" as a resultof ostensive definition,
but depending on a nonvisual formof perception of a special kind. (See
K. Goedel, "Russell's Mathematical Logic" and "What is Cantor's Con-
tinuum Problem?" reprinted in H. Putnam and P. Benacerraf feds.],
Philosophy ofMathematics.)
If my account of Wittgenstein'sreasons for connecting proof with
meaning is correct,his motivations were quite specific to mathematics.
Unlike a verificationist,he is committed to no corresponding theoryof
meaning outside of mathematics. For example, outside of mathematics,
wherenonconventionalistconceptions of truthare appropriate, explana-
tion of meaning by ostensive definition, as against the teaching of
"proofs" that is, methods of verification),can play an irreducible and
crucial role in language learning.

541

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

Wittgensteinis well aware of one consequence of his association of

meaning with proof which many would find disturbing: The same
formulawith the self-samesense cannot be proved in radically different
ways,that is, using proof-proceduresso differentthat the one cannot be
used to validate the other. Thus, while an equation relating sines to
cosines mightbe thoughtto be established in radically differentways in
elementarytrigonometryand in the calculus, Wittgensteindenies that
the same equation occurs with the same sense in the two branches of
mathematics(p. 378). He similarlydistinguishesbetween the sense of a
geometric formula established experimentally, via physical measure-
ment, and the same formula established deductively, in axiomatic
systemssuch as Euclid's (p. 319f.). In the section on recursiveproof, he
distinguishesbetween the sense of an algebraic formula established
through"recursiveproof" and the sense of the same formula when it is
accepted without proof as a fundamental axiom of algebra.
These distinctionsof differencesin sense seem well motivated provided
one accepts Wittgenstein's"conventionalism" concerning mathematical
truth.Suppose, to returnto a previous example, that one mathematician
accepts a "recursive proof" as establishing the validity of an algebraic
formula,F, while the other does not. The latter suffersfrom no defect
in mathematical technique. He is a nonrecursivist,who simply accepts
the mostfundamentallaws of algebra as axioms. According to Wittgen-
stein,neitherthe recursivistnor the nonrecursivistis wrong when they
disagree as to whether"recursive proof" establishes the validity of the
formulain question. An attractive explanation of such a differencein
which neitherside is to be condemned is that each side attaches a dif-
ferentmeaning to the sentence (for example, "F can be established
recursively")which one side affirmsand the other negates. In particular,
"F can be established recursively"and "F cannot be established recur-
sively"can both be true,as said by the respectiveparties, ifeach means
something differentby F. In much the same way, one botanist can
classifya microbeas a plant because it contains chlorophyll and another
can deny it is a plant, noting that it lacks a celluloid cell-wall, without
either's being wrong, since "plant" has differentmeanings, in their
differentrespectivetechnical usages.
Because of the association of proof with meaning, Wittgenstein is
never forced to choose between the acceptance of both sides of a sub-
stantive disagreement in mathematics and the denial that truth exists
in a branch of mathematics. Suppose that two mathematicians, each
being faithfulto the consequences of his stipulations, answer a third
mathematician's question in differentways. That difference would
reflecta differencein the proof-proceduresthey accept. On this basis
Wittgensteincould argue that the apparent disagreement is merely

542

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 RICHARD W. MILLER

verbal, the initial question was relevantlyambiguous, and both

answersare true.
Like much else in Part Two of the Grammar, the discussion of
meaningand proofraisespressingquestionsthat do not receive the
attentiontheydeserve.One concernsthe existenceof purely nota-
tionalor stylisticdifferences in modesof proof,such as the difference
betweendoing calculus in Newtonianor Leibnizian notation. The
self-samemathematicalpropositioncan be the outcome of proof-
proceduresdiffering in merelynotationalor stylisticways. How can
these peripheralvariationsbe distinguishedfromthose that create
differences in meaning,withouta reversionto mathematicalPlaton-
ism? Wittgenstein does not confrontthis question in the Grammar.
If I have correctlyinterpreted the second part of the Grammar, it
cruciallydependson a sharpdistinctionbetweentruth,as it applies to
mathematical propositions,and truth,as it applies to propositions con-
cerningphysicalobjects.That distinction requires,in turn,a distinction
of reference as it occurs in mathematicsfromreferenceto physical
objects,thelatterbeingthetraditionalparadigmofreference. Here is an
importantparallel betweenthe philosophyof mathematicsof the
Grammar and Wittgenstein's criticism,in theInvestigations,
of "the gram-
marof 'objectand designation'"adopted in traditionalphilosophyof
mind(see P.I. 293,compare304,308). Though I cannotarguethepoint
here,thisrejection ofreference to physicalobjectsas theparadigmforall
reference strikesmeas one ofthemostimportantresemblancesbetween
the two books.
Philosophers withan interestin Wittgenstein'sphilosophicaldevelop-
mentclearlyneedtoreadtheGrammar. Butwhyshouldthosewitha more
casual interest in Wittgenstein read thisbook,the themesof whichare
elaborated, supplemented, and refined inlaterwritings, alreadyavailable
fordecades?Primarily, I think,becauseofWittgenstein's discussionsof
thequestion, "Whatgivessignstheirlife?",discussions whichsupplement
theInvestigationsin twoways.Themeswhichare foundin thelaterwork
are discussedat greaterlengthin the earlierone, or with usefuldif-
ferencesin emphasis.I have in mind,in particular,the parts of the
Grammar in whichWittgenstein insistson theimportanceofmasteryofa
syntaxin theuseoflanguage,attackstheviewthatlanguageis simplya
causal mechanismservingcertain goals, or denies that what gives
meaningto wordsis a processor state.Also,theGrammar has a charac-
teristicvirtueof transitional works.Reading Wittgenstein's reflections
on questionsas to whatgivessignstheirlife,one seeswhyhe foundthese
questionsveryhardand veryimportant. Many readersofWittgenstein,
and somewhowriteabout him, either regardthequestionsconcerning

543

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions
 WITTGENSTEIN IN TRANSITION

meaning which Wittgenstein discussed as easily solved, aside from

mattersof detail, or find his worries artificial and self-imposed. The
discussionsin Part I of the PhilosophicalGrammar (especially section VII)
are a good cure for this easy optimism.
The recentlypublished English translationof the Grammar was done by
AnthonyKenny. The translationis accurate and untendentious.Kenny's
styleis somewhat closer to word-for-wordtranslation that Anscombe's,
in her renderingof the Investigations.
Partly as a result,his translation is
less gracefulthan hers. (Anscombe, of course, was also working with a
more polished original.) But aside fromawkward phrases (for example,
"Similarly,you can to order see the figurebelow. . . " [p. 394]), Kenny's
prose is quite readable. And, with rare exceptions, it is as clear as the
original text. One exception occurs in the middle of page 143 where
"What the proposition really means is the image of something red"
should be "What the proposition really refersto [meint] is the image of
somethingred." Also, on page 466, Kenny writes,somewhat obscurely,
m = 2n allows the possibilityof correlating every time with another"
where the original permitsthe clearer translation "m = 2n allows the
possibilityof correlating every number with another." But perhaps a
phrase has been leftout, a victim of the plague of typographical errors
to which I now will turn.
The firstprinting of this translation is strewn with typographical
errors.In nearly everycase, the mistake and its correction are obvious.
But those who do not know German will be puzzled by the "W" that
should be a "T" in the truthtable on page 55. The nonsensical "a com-
mand such signals" on page 195 should be, surprisingly,"one could
learn such signals." On page 275, the final comma in "F(a,b,c,d,e,)"
should be omitted. At the top of page 292, there should be a negation
sign in frontof "(21 x 8 = 148)". At the bottom of page 311, "writing
for"should be "waiting for."At the bottom of page 378, thereshould be
an ellipsis (". . .") after "sin x = x - x3". The passage is pointless
3!
withoutthiscorrection.Joining in a familiar reviewer'slament I would
say that one expects better for $18.00. But I no longer do.

CornellUniversity

544

This content downloaded from 193.142.30.51 on Sat, 28 Jun 2014 17:28:50 PM

All use subject to JSTOR Terms and Conditions