ltury Philosophy
The Story of Analytic
Philosophy
Plot and Heroes
Edited by Anat Biletzki and Anat Matar
1998
London and New York
8
Frege, seDlantics, and the
double definition stroke!
Juliet Floyd
I
In 1964, in his introduction to his partial translation of Frege's
Grundgeset<:l, Montgomery Furth wrote:
There is ... rCllson IClr
importance today to Frcgc's treatment
of [his primitive assertions and his sound rules of inlcrenccJ: that is, Frcgc's
explanation of the primitive basis of his system of logic, and particularly of the
primitive symbolism, is undertoJrm in terms of a deeply tlwug/U-out semantUal interpretation, which in tum embodies an entirt philosophy rtf language. The influence of the
latter upon 1M semantUal structure and even the syntax of the language developed
makes iudf fdt steadily throughout the discussion. This phiWsophy of language is
very profound, and possesses great interest quite independent of its origin as
the handmaiden of "logicism". It tIJo is not gmerally weU rmtlmtood even toti4y.
(Frege 1893: vi-vii, my emphases)
In a quite different vein,Jean van Heijenoort, in his 1967 paper "Logic
as Calculus and Logic as Language," wrote:
From Frege's writings a certain picture of IogU emerges, a conception that is
perhaps not discussed explidtly but nevertheless constandy guides Frege. In referof logic.
ring to this conception I shall speak of the
. . . The universality of logic expresses itself in an important feature of
Frege's system. In that system the quantifiers binding individual variables
range over all objects ... .Frege's unWmt consists rtf aU that tnne is, and it isfixed.
. . . Another important consequence of the universality of logic is thaJ
nothing can be, or has to be, said outside rtf the !JIstem. And, in fact, Frtge never raises
any mdasystematU question (consistency, independence of axioms, completeness). Frege is indeed fully aware that any formal system requires rules that
are not expressed in the system; but these rules are void of any intuitive
logic; they are "rules for the use of our signs" (note: Beg §13). In such a
manipulation of signs,.from which any argumentative logic has been squee;:.ed Oltt,
Frege sees precisely the advantage of a formal system.
Since logic is a language, that language has to be learned. Like many
languages in many circumstances, the language has to be learned by suggestions and clues. Frtge repeatedly states, when introducing his !JIstem, that he is giving
142 }uliet Floyd
"hints" to 1M I'UklH; lllat IIu
fl.
Semantics and the doubll
hal to mill
and iMUId not b'Itut/fl him
" IIIIm rif ''grJoi will", The. problem p to bring the reader to "catch on": he
has to get into the language. (note: Here 1M i1ifhunc1 of I+ege on Witfgtnstein is
obvious. Also, Frege's rifusal to entertain mtla.lYstematic question.r explains perhaps
why he was not too disl urbccl by
sl alcllIcnt
WIll:Cpt /lorse is not a
conccpt". The paradox arises from the fact that, since concepts, being limetions, are not objects, we cannot name them, hence we are unable to talk
about them, Some statements that are (apparently) about concepts can easily
be translated into the system; thus, "the concept ttx) is realized" becomes
"(& )ttx )". 17u statements that resist such a translation are, upon examination, melaf)lslematic; for example, "there arc functions" cannot be translated into the
system, but we see, once we have "caught on", that there are function signs
among the signs of the system, hence that there are functions.)
(Van Heijenoort 1985: 12 13, my emphases)
Van Heijenoort insists that for Frege, logic is universal, it cannot pick or
choose different universes of discourse, and the formula language
(BegriffischtiftJ that is the formalization of this logic is not subject to differing
2
interpretations. Moreover, when correctly formulated this formula
language lays bare the framework within which all thought must take place,
it lays bare the limits of rational discourse, the limit.r of sense. AllndinR to the
Kerry paradox, van Heijclloort treats the issues raised by "the concept horse"
problem as paradigmatic for Frege's attitude toward most, if not all "metasystematic" questions. And thus he holds that in the (German) presentation
of his formula language Frege takes himself to rely wholly, or at least in large
part, upon "suggestions," "clues," and "hints" to help the reader see or
"catch on" to "rules ... from which any argumentative logic has been
squeezed out." In this way van Heijenoort argues that no "metasystematic
questions (consistency, independence of axioms, completeness)" can arise for
Frege about his formal system, his formula language.
This reading sharply questions Furth's claim that in Frege we fmd "a
deeply thought out semantical interpretation" or "semantical structure"
embodying "an entire philosophy of language." For although van
Heijenoort wrote in a subsequent paper that for Frege the formula
language elicits the latent aruJ determinate content of ordinary languagt?-a picture
many would caJl "semantical"4----neverthdcss van Heijenoort insisted that
Frege conceived himself to be uncovering this latent structure of language,
namely logic, from within the one and only universaJ language, namely,
logic itself. Hence, unlike Furth, van
denied that
r:uuld
have adopted any sort of "metatheoretical" or model theoretic stance in so
proceeding. (For van Heijenoort, "semantics" always meant model theoretic or set theoretic semantics.) The formula language was, van
Heijcnoort claimed, intended by Fregc to totally supplant rather than simply
to reflect or to aid natural language. This claim explicitly appears in a later
stroh
143
pan of "Logic Ai Calculul and Logic au Language," where van HeljenQort
buttrelles his reading of Frege by appealing to what he says are parallel
lines of thought in Russell:
QI.Iestiolls ahout the system are as absent from Principia matlumatica III they are
from Frege's work. Smumtic notions are unknown. "I-" is read as " ... is true", and
Russell could hardly have come to add to the notion of provability a notion of
validity based on naive set theory.... If the question of the semantic completeness of quantification theory did not "at once" arise, it is because of the
universality-in the sense that I tried to extricate---of Frege's and Russell's logic.
The universalformal language supplants the natural language, and to preserve, outside rif the
.rystem, a notion rif validity based on intuilive set theory, does not seem tofit into the scientific
reconstruction rif the language. 17u on!>, question rif completeness that ""!y arise ir, to use an
expression rif Herbrand's, an experimentlJi question. As many theorems as possible art derived in
1M .rystem. Can we exhaust the intuitive modes of reasoning actually used in
science? To answer this question is the purpose of the Frege-Russell enterprise.
(van Heijenoort 1985: 14, my emphases)
In his 1973 Frege: Philosophy of lAnguage Michael Dummett massively
articulated the reading of Frege called for by Furth. Hence, in particular,
though without explicitly mentioning van Heijenoort, Dummett strongly
with vall Hcijelloort's denial of the pORRihility of metasystematic
considerations in Frege:
Although Frege did not expressly defme the two notions-semantic and
syntactic-of logical consequence, they lit rea4J to hand in his work: for there, on
the one hand, is the formal .rystem, with its precisely stated formation rules,
axioms and rules of inference; and there, on the other, are the semantic exp/anQ.tions of the sentences rif the formalized language, set out, clearly separated from the
formal development, in German in the accompanying text.... The sentential
fragment of Frege's formalization of logic is complete, and likewise the flrstorder fragment constitutes the first complete formalization of fIrst-order
predicate logic with identity. It was left to Frege's successors to prove this
completeness, as also to establish the incompleteness of his, or of any of the
effective, formalizations of higher-order logic. Frege had it to hand to raise these
questions: but he did not do so ...
(Dummett 1973, 1981: 82, my emphases)
Dummett's reading greatly enhanced the wowing interest in Frege by
securing Frege's relevance to contemporary philosophy of language and
theory of meaning. It has become the dominant reading..
However, Dummett has not gone unchallenged. In his 1979 "Logic in
the Twenties: The Nature of the Quantifier," Warren Goldfarb explicitly
criticized Dummett5 and agreed with van Heijenoort about the absence of
any "metasystematic" standpoint in Frege. Goldfarb based his argument
on a detailed historical assessment of the development of quantification
theory and concluded:
1« Jr;liIt Floyd
&mtmtits and the double definition Jtroke
[The] lack of intelligibility [of Independence questions in logic] may be
intrinsic to the logicist program. If the system constitutes the universal logical
language, then there can be no external standpoint from which one may view
and discuss the system. MefaJystematic considtrations are ilkgiJirruJte rather tJuzn simplY
undesirahk. This is what Harry Sheffer called "the logocentric predicament"
(Sheffer, "Review of Whitehead and RusseU, Principia Mathematica", Isis 8:
226-231), and forms a large part, I think, of the motivations behind
Wittgenstein's Tractalus. (Fn: A similar "Iogocentricity" may underlie Frege's
curious claim that "Only true thoughts can be premises of inferences" (Frege
1971: 425).)
(Goldfarb 1979: 353, my emphasis)
Ij.!
j)"
;
;J
Goldfarb too (1979: 353) buttressed his claim about Frege's "logocentricity" by drawing a parallel between Frege and Russell, pointing out that
in §17 of The Principles of Mathematics (1903) Russell had written:
1IIi
1
indemonstrables there must be; and some propositions, such as the syllogism, must be of the number, since no demonstration is possible without them.
But concerning others, it may be doubted whether they are indemonstrable or
merely undemonstrated; and it should be obseIVCd that the method !!f supposing an
if
i[
in such cases as the axiom !!f parallels, is here not universal!Y availahk. For all our axioms
are principles of deduction; and if they are true, the consequences which
appear to follow from the employment of an opposite principle will not really
1,;1
t;
.i,
t
i';
follow, so that arguments from the supposition of the falsity of an axiom are
here subject to speeial fallacies.
(Russell 1938: IS, my emphasis)
;,1
Thus, though Russell in one sense certainly does raise a question about
independence, he seems, as Goldfarb says, "not even to see the intelligibility of stepping outside the system to use an intuitive logic in
metasystematic arguments" (1979: 353).
Since the 1970s several other philosophers interested in early analytic
philosophy have also explicidy differed with Dummett's semantical reading
of Frege by offering their own anti-semantical readings. 6 Gordon Baker
and Peter Hacker (1984), Cora Diamond (199Ia, 1991b), Burton Dreben
Jaakko and Merrill Hintikka (1979, 1986), Thomas Ricketts
(1985, 1986a, 1986b, 1996a, 1996b, 1997), Hans Sluga (1980, 1987), and
Joan Weiner (1982, 1984, 1990, forthcoming), while differing in various
ways with each other, all belong to this anti-semantical tradition. 7 Like van
Heijenoort they have argued that far from advocating a theory of meaning
or a semantical theory, Frege rejects as nonsense any attempt to account
systematically for the nature of logic, the nature of meaning, or the nature
of language. Of course, no one denies that Frege uses the notion of the
content of a judgment and, later, the notions of Sinn and &tkutung. What is
questioned is the idea that in using these notions Frege is theorizing about
I,
il'r "
It,
7
! ,
;1
I
* • UP
$
¥
•
;;.•-.. ,
, - ...,.-...... -
145
language or meaning in general. In order to question this idea, these
philosophers emphasize as crucial Frege's use of what he calls "elucidations," "Erlauterungen" (Sluga 1980: 180-2; Weiner 1990: ch. 6; Weiner
forthcoming). Frege tells us that elucidations serve to set forth the way we
are to understand certain of his undefmable, i.e. logically primitive,
notions (1984: 300, 302; 1979: 207, 214, 235).8 In one way or another
Frege's anti-semantical readers all seem to agree with van Heijenoort's
focus on Frege's explicit uses of elucidation, hints, clues, and metaphors in
presenting his formula language. 9 Indeed, Weiner goes so far as to maintain that Frege must deny both that our numerals in everyday arithmetic
may be said to have &dtutungen, and that our ordinary, pre-analyzed
propositions of arithmetic have determinate truth-values (Weiner 1984:
78). While not all the anti-semantical readers go this far, all emphasize the
extent to which Frege explicidy emphasizes that at times he is forced to fall
back upon hints, metaphor, and indirect suggestion, relying on his reader
for cooperation, good will, and guesswork. 1O Such figurative and colorful
language indicates, they argue, that Frege has no antecedent or independent theoretical account of semantical notions. Thus elucidations are to
be sharply distinguished from judgments, from theory; they do not and
are not intended to play anything like the role of theoretical premises in
deductive arguments. Furthermore, on the anti-semantical reading all
Frege's judgments, Le. all his genuine theoretical claims, are expressible in
the formula language, where proofs are set forth by means of assertions,
and assertions are recorded by means of the judgment stroke. As van
Heijenoort insisted, Frege does not and cannot so transcribe certain
apparendy crucial claims about his basic philosophical notions (e.g.
jUnction, concept, object). It is Frege's conception of logic, these philosophers
argue, which leads him to treat such apparendy metasystematic and
semantical talk as mere pedagogical rhetoric. Thus all such talk must be
sharply distinguished from a theory of sense, a theory of meaning, or even
a theory of how the truth-values of sentences are determined by semantic
features of their constituents. tl
For example, in "What Does a Concept-Script Do?" (1983), Cora
Diamond emphasized, as had van Heijenoort, Frege's treatment of "the
concept HOTJe":
What Frege thinks is that through an inadequacy of ordinary language, we can
form sentences in it which are acceptable according to its rules but which are
not the expression of any thought. It is possible to become clear about what
has happened, if we are led to see how thoughts are expressed in a language
more nearly adequate by the standards of logic. In grasping the significance of
the distinctions embodied in that language, we do not grasp any ineffable
troths. A truth is a truth about something; a true thought (that is) is about
146 ]uliLt Floyd
whatever logic may construe it as being about. But the distinctions embodied in
the concept-script are not what any thought can be about.
(Diamond 1991a: 140--1)
Ricketts bases hi'l argument for an anti-semantical reading of Frege on
what he takes to be Frege's underlying conception of judgment and inference. t2 This conception, he argues, lies behind the way Frege introduces
his formula language and is intimately connected with Frege's arguments
that truth is indefmable. t3 At best, holds Ricketts, Frege's post-l 890 talk of
Sinn and Bedeutung is in "deep tension" with his underlying conception of
the universality of logic (Ricketts 1985: 3; 1986a: 172; 1986b: 66-7).
Such readings are almost always buttressed by an appeal to the influence
of Frege upon Wittgenstein. In 'Tractatus 4.112 Wittgenstein wrote that:
The aim of philosophy is the lowcal clarilication of thOllghts.
Philosophy is lIot a theory Lut all activity.
A philosophical work consists essentially of elucidations
The result of philosophy is not a number of "philosophical
propositions," but to make propositions clear.
(Wittgenstein 1922: 4.112, c( 6.54)
The line of descent from Frege appears to be explicit; the anti-semantical
tradition sees in Wittgenstein the heir to Frege, and the 'Tractatus show/say
distinction is said to have emerged directly from Frege's conception of
logic as universal. t4
Dummett remains unmoved by such claims. Indeed, in his 1981 The
InJtrpretatUm of Frege's Phiiosop/rl he explicitly based his semantical reading of
Frege on what he insisted was Frege's "universal logic":
It is, of course, true that Frege did not attempt to explain his logical notation by
giving rules of translation from his formulas into sentences of
Rather, when you grasp the SelL'IeS of the primitive symbols, you thereby grasp
you afe thl'lI ahle 10
the thoughts l'xpres.'Icd by the formula.'I, whkh
express in any lauguaRe known to you and (:apahle or cxpn'ssinR tlwlIl. IIl1t it is
plain that he did not think that, in the logical notation of
he had
devised a language in which thoughts could be expressed that could not be
expressed in any other way; the logic of
was intended as a univl'TJal
logic, not one pcculi,lr to a spedallanRwlRc III1(,OI1I1('cll'<1 wilh Ih(' thollRhts WI'
are ordinarily concerned to communicate to one another. The meaning-tJuoretical
notions used in Part I of GrutuJgeset;;;e are therifore not to be considered as app{ying exclusive{y
to Prep's.frmnoJ -!,sttm. They are to be taken as applicable both to it and to
natural language, and perhaps as requh'ed for any intelligible language; and the
theory of meaning tIuJJ embodies them according{y serves both as a tJuory of sensefor natural
and as afoundationfor tlIeJormallngic ...
If these conclusions are correct, there is a substantial body of Frege's
Smumtit:.1 and tlze doublt definition strokL
147
theory-precise{y tluJt comprising hisphilosOP/rl of Ionguage·-of which no definitive
exposition, comparable to Grwu!geset;;;e, or even carrying an authority equal to
that of GrwulIogm, exists.... We can do no more than surmise the reason
why he never achieved a formulation of his general thMwy of philosophicallogie.
(Dummett 1981: 18-19, my emphases)
Dummett admits that Part I of Grundgesetze, where Frcge explains to the
reader how to use his system, contains "no definitive exposition" of a
Frege's apparatus of philosemalltical theory. Yet, Dummett argues,
sophical notions such as jUnction, object, sense, riference, and so on are
intended to apply to natural language as well as to language regimented in
the formula language, there is in Frege's writings a semantica1 theory
which goes far beyond "hints" and "clues" to help the reader "catch on"
to the formula language. Dummett takes Frege's conception of the universality of lowe to commit Frege to a theory of meaning for his formula
a tltt'my whirh silllllhatWOlIsly applies to sentenc.es of natural
language. Thus we have Dummett's implicit challenge to the anli-semantical tradition: What is the point of forcing a reading on Frege which
precludes him from adopting anything like a contemporary metasystematic
perspective on his formula language? That is, if we were to imagine Frege
living a bit longer than he had, and learning of model-theoretic results, on
what grounds and for what purposes can we hold that Frege's conception
of logic would have had to shift if he had come to accept those results as
illuminative of logic?
Dummett reiterated this challenge in his 1984 review of Baker and
Hacker's liJgi.cal Excavations. Baker and Hacker too had claimed that:
Begriffischrifl. contains the first complete axiomatization of the propositional
calculus. Of course, Frege did not prove its completeness or its consistency.
Indeed, the scorn that he poured on Hilbert's metalogical investigations of
axiomatizalion of geomelry he would have turned too against metalogical
proofs concerning his own logical system. The only proof of the consistency of
a
of axioms is a demonstration of their joint truth, the only demonstration
the compatibility of the denial of one with the joint
of their
asscl't ion of Ilw n'sl (On the /-1/1l11d(Itio/lJ iiI
and rormal 77,eOlieJ III
Arithmetic, 1:1[,104). Lacking any nwans of prccis(·ly dcmarcathlg logical laws
apart from their derivation as theorems within his axiomatic system (Cr
(Aundgeset;;;e der Arithmetik p. xvii), he was not even in a position to frame an exact
l1"cstioll aholll Ihr rOlnplclrncss of his axiomati7.ation.
(Uak('f and Hacker 191H: I H)
Hence Baker and Hacker concluded that there are "no grounds for
asserting that LFrege] advanced ... to any conception that the true busiof logicians is a science of language (semantics) ... The hypothesis
that [Fregc] intended to lay the foundation of logical semantics is implausible" (1984: 248---49).t5 Dummett responded:
148 Juliet HOJd
The word "semantics" is used in several different ways, but the references to
logic and logicians suggest that what [Baker and Hacker] have in mind is a
semantic theory for a formal language as conceived in contemporary model
theory. If so, their [above-quoted] assertion is very surprising, since Part I of
Gnmdgesetzt appears to contain a semantic theory for the formal language,
clearly separated from the account of its formation rules, axioms and rules of
inference: this theory is stated by stipulating what references the primitive
symhols are to have, and laying down how the reference of a complex expressioll is determined Irom cII<:
of iL' C()llstitU('lIls. III addition, Frege
gives a general framework for such a theory, namely an account of the various
possible logical types of expression, of their nature and how they are formed,
and of what it is to assign a reference to an expression of anyone such type;
this is likewise clearly separated from the specific stipulatiolls governing the
primitive symbols of the system.
One reason why Baker and Hacker do not see the maUer in this light is that
they conllate a semantic theory with a
definition of logical conscquellC!'. Thry arc quite right in saying that
I,leked the lalter notion. He
larknl it hrmuse he <lid IIUt ullI'mte wi. h I ht· mUff'pl iun of a rallKe 01" possihle
interpretations of a
... lj: hOlllel,"- lit. "nd.lormed iIIil "onception, lit
would Mve Iuld very liule more lIJork 10 do 10 arrive at the semantic notion qf
for the
blUkgrourui tirlory stated in Pari I l!!' Gruntlgesetzt would immediate!Y Milt yitkkd aformulmion qf wIult, in general, a'!>' one such interpretation should consist in. It is pruisely because
r!f 1M presence qf this badrgrowui theory, and its cl1m, though not complete, resemblallu 10 1hz
notion used !?y modern logieion.t r!f an interpretation r!f a formal longuage within classUal
two-valued Stma1ltics, iIIat Frege's work can beftuiyully compared wiJh tIuU 0/ laJer logieitms.
(Dummett 1981: 20 I, my emphasis)
,
it
:1
.,.,.
r!,
:'
·1
[1
i
iii.
,,.
Thus Dummetl emphasizes the ease with which Frege's work may be
appropriated from a contemporary model-theoretic perspective. And he
remains deeply dissatisfied with the notion that Frege's mature philosophy
of logic is in tension with model theoretical clarification of logical notions.
Dummett again gave voice to this dissatisfaction in his review of a 1987
essay of Dreben and van Heijenoort (Dummett 1987). In this essay, a
commentary on Godel's dissertation and subsequent publications on the
model-theoretic (semantical) completeness of first-order logic, Dreben and
van Heijenoort raise the question: Why was it that fifty years elapsed
between the publication of Frege's Begriifsschrffl (1879) and Godel's proof
of the completeness theorem (1929)? 16 Their answer is that in order to
frame the question of completeness coherently, the algebra of logic tradition-the tradition which denied the universality of logic and emphasized
the notion of varying universes of discourse-needed to be combined with
thc quitc diffcrclltlogieist tradition stemming from Frcgc and Russell:
fbr trege, and then for RusseU and Whitehead, logic was universal: within each
explicit formulation of logic all deductive reasoning, including all of classical
analysis and much of Cantorlan set theory, was to be formalized. Hence not
only was pure quantification theory never at the center of their attention, but
&mnnl.ics and ti,e doubli definition stmki
I '1
metasystematic questions as such, for example the qUl'!stion of completeness,
could not be meaningfully raised. We can give different formulations of logic,
formulations that differ with respect to what logical constants are taken as
primitive or what formulas are taken as formal axioms, but we have no vantage
point from which we can survey a given formalism as a whole, let alone look at
logic whole.
(Drehen and van Heijenoort 1986: 44)
Once again, this reading of Frege is huttressed by quoting from RUllseJl:
In the words of Whitehead and Russell 1910 (page 95, or 1925 page 91),
"It is to some extent optional what ideas we take as undefined in
mathematics.. , . We know no way of proving that such and such a
system of undefmed ideas contains as few as will give such and such
results. Hence we can only say that such and such ideas are undefined in such and such a system, not that they are indefinable."
We arc within IURic and ranllot!ook at it from outside. We arc sllbject to
wlmt Slwlfer railed "thr logorcntric prcdic:amC'lIt" ... Thr only way to
approaeh the problem of what a limna! system can do is to derive theorems,
Again to quote Russell and Whitehead, "ille
reasOR inftvor of a'!>' theory on /he prillciple 0/ mathmuJtics must always be inductive, i.e., it must lit in 1hzflUt that tirllhzory in
quetion eno.hlts us 10 tkdw:e ordinary matJumoJid' (191 0, page V, or 1925, page v)o (On
and
this point see van Heijenoort, "Logic as Calculus and Logic as
Goldfarb, "Logic in the Twenties: the nature of the quantifier" 0)
To raise the question of semantic completeness the
Frege--Russell-Whitehead view of logic as all embracing had to be abandoned, and Frege's notion of a formal system had to become itself an object
of mathematical inquiry and be subjected to the model-theoretic analyses of
the algebraists of logic.
(Dreben and van Heijenoort 1986: 44, my emphasis) 17
•
•
0
Dreben and van Heijenoort's use of the history of logic to develop an
anti-semantical reading of Frege did not convince Dummett:
Drehen and van Heijenoort ... discuss [a] question, one of the most interesting in conceptual history, why it took so long for the concept of completeness
to be framed.
The explanation given by Dreben and van Heijenoort is that,
in the tradition descending from Boole through Peirce and SchrOder, the very
notion of a formal system was lacking: in his paper of 1915, for instance,
Lowenheim worked with exclusively semantic notions. frege, on the other
hand, had bequeathed the notion of a formal system to the tradition that
stemmed from him; but, according to that tradition, logic is all-embracing:
since there is only one logic in accordance with which all reasoning must be
conducted, we cannot step outside logic in order to formulate theorems about
rather than within it.
However accurate a statement of the views of Russell and Whitehead
this may he, it appears to me an oversimplification as applied to Frege. True,
he never considered the sentential and first-order fragments of his logic as
significantly separable from it; and he had repudiated the traditional
0
••
.
.,
,:
I,
,n
- -, --.. _--
•
•
,a ••
4
Semantics and th£ double definition stroke
150 juliet Floyd
I
II
t;
l
I
il
conception of variable universes of discourse. But, in his Grundgeset;;e, he did
attempt a precise formulation of the semantics of his formal system, clearly
distinguished from the specification of its axioms and rules of inference.
Indeed, he attempted a proof that every term of the formal system had a
unique denotation (and every sentence a unique truth-value), which, if
correct, would have constituted a consistency-proof. Moreover, he did not
content himself with laying down specific interpretations for the primitive
symbols, but stipulated what form the interpretation of an expression of
each logical type must, in general, take. All tluJt he locked, therifore,for aformulation of the concepts of valUJi!)! and satiffio.bilitY was tJu conceptum of variable
interpretations (ironically so close at hand in Hilbert's Foundations of Geometry).
The mystery of the half-century that elapsed between the invention of
mathematica110gic and the formulation of its fundamental problem is thus
not fully dispelled.
(Dummett 1987: 573-4, my emphases)
In unpublished lectures Dreben has pointed out that in at least two places
In Begriffsschrift
Frege did raise the question of completeness
Frege wrote,
Since in view of the boundless multitude of laws that can be enunciated we
cannot list them all, we cannot achieve completeness
except by
searching out those that, by their jxJwer, contain all of them. Now it must be
admitted, certainly, that the way followed here is not the only one in which the
reduction can be done. That is why not all relations between the laws of
thought are set out clearly through the present mode of presentation. There is
perhaps another set of judwnents from which, when those contained in the
rules are added, all laws of thought could Jik('wisr he deduced. Still, with the
method of reduction presented hcre such a multitude of relations is exhibited
that any other derivation will be much facilitated thereby.
(Frege 1879: §13)
And in "On Mr. Peano's Conceptual Notation and My Own" (1897):
In order to test whether a list of axioms is complete (vollstiindig), we have to try
and derive from them all the proofs of the branch of learning to which they
relate. And in so doing it is imperative that we draw conclusions only in accordance with purely logical laws, for otherwise something might intrude
unobserved which should have been laid down as an axiom.
(FTege 1984: 235; 1967: 221)
However, Dreben argues that these passages conftrm the Dreben and van
Heijenoort reading of Frege. For in both passages Frege eschews any jump
to a metalevel; he operates with what Russell called "inductive" :reasoning
and what van Heijenoort, followinR Herhrand, called an "experimental"
approach: derive as many theorems as possible within the system in order to
answer the question of whether the system is complete.
Thus, those questions Dummett holds that Frege merelY didn't raise, but
"had it to hand to raise," others say he didn't raise because he couldn't raise,
151
"couldn't" in the sense thal he did not have room, within his conception of
logic, for the posing of these questions in their metamathematical sense.
In response to a 1992 lecture of Dreben's, Quine expressed scepticism
about the anti-semantical reading. IS Does Dreben mean, asked Quine,
that if Frege had seen Gode1's proof of the completeness theorem, Frege
wouldn't have been able to understand it? Dreben replied: the Frege of
the Grundgesetze would have understood it as a piece of mathematics, as
showing that a certain set-theoretically defmable class is recursively
enwnerable; but Frege would have questioned whether this set was a
proper specification of his notion of logical truth, of logical validity.19
Naturally, Quine, devoted as he is to Tarski, was not satisfted with this
response. Dreben's reply to Quine is also his reply to the suggestions of
Heck, Stanley, and Tappenden that since logic is universal, by the techniques of Godel and Tarski, many metasystematic questions, in particular
the completeness theorem, can be carried out within the system. 20
Presumably Heck, Stanley, and Tappenden would share Quine's dissatisfaction, as of course would Dummett.
II
My aim in what follows is modest: I wish to slightly shift the scope of the
interpretive debate about Frege and semantics. Rather than directly
addressing those elements of FreRe's conception which have been treated
at length in the recent literature, I
foeus on one especially puzzling
feature of I''rege's formal procedure, namely, his usc of the double stroke
(" 11-"), his symbol for explicit definition. Frege's use of this symbol-and
his attitude toward his use of it--seem to me to exemplify in an especially
perspicuous way how it is that Frege's writings can so easily lend themselves to such radically different interpretations and appropriations of his
philosophy. For the double stroke for definitions shows us, I claim, the
importance of the fact that at least until 1903 Frege never clearly articulated a distinction between "metalanguage" and "object language," or
between a "metasystematic level" and an "object level."21 It helps us to
see that even if we can quite naturally regiment Frege's work by means of
the distinction between meta- and object level without appearing to go
beyond what is written in his logical system, the deeper philosophical
questions about Frege's standpoint still remain. My discussion is not
intended to
or exhaustively characterize the interpretive questions I
have canvassed about l''rcgc and scmantics. I intcm], rathcr, to cmphasize
that these questions will not be resolved by any argument resting its case
on purely formal or mathematical features of Frege's system. If I am right
that there is no way to establish how to read uniquely Frege's use of the
152 Juliet Floyd
I III!
r
I
..,"
!:
:1
'\
I,
r,
r
i
r
.'
I:
,I
.'"
II
IIri
,
II
fI;
II I
double stroke, then the interpretative questions make themselves felt even
at the most basic level of Frege's system, in the part of his work one might
think is not opell to philosophical interpretation, viz., in his formal system.
This confrrms that the debate about how best to read Frege is at root a
philosophical one, not amenable to resolution by means of any particular
formal or mathematical distinction we may find in Frege.
In contemporary logic and philosophy of logic we may distinguish three
different issues surrounding identity statements and definitions. First, we
may give a syntactical specification of identity statements and definitions
in a particular formal language by making statements in a metalanguage
(whether metasyntactic or metasemantic). Sccond, we may discuss the
nature or purpose of definitions and the nature or purpose of identity
statements about the identity relation. Third, we may give an account of
the nature of the identity relation itself, the naturc of the truth asserted in
a true identity statement. 22
It is striking that in his 1894 review of Husserl's Philosophy of Arithmetic,
Frege seems to have conflated these issues. This apparent conflation seems
to indicate that Frege did not always sharply distinguish between statements and what statements are about. For in this review Frege forwards the
following argument, which purports to show that identity is undefmable:
"Since any defmition is an identification [Gleichung] , identity itself [Gleidlheit
selbst] cannot be defmed" (Frege 1984: 200; 1967: 184; 1980a: 80).23
The argument appears to be this: in order to present a defmition, one
must use the identity sign, and thus there is no way to present an explicit,
i.e. eliminative definition of the idcntity sign itsclf. 24 Thereforc, Frcge
reasons, identity must be taken as a logical primitive, an undefinedbecause undefmable, and hence logically simple-notion. 25
This 1894 argument for the indefinability of identity is simultaneously
an argument that the notions of difinability and of d¢nition cannot be
defined. 26 For in the Begriffsschrift and in the Orundgesetze, in order to present
a definition Frege must always use, not only the identity sign, but also his
special sign for definitions, the double stroke (U 11-"). Hence, by Frege's
1894 line of reasoning there is no way to c1iminate the sign for definition
by means of a defmition itself, for in order to frame a defmition one must
always use the double stroke. The plausibility of this reasoning and the
parallel argument about identity, thus appear to tllrn in part upon Frege's
not sharply distinguishing between an object level and a metaIevel, and in
. part upon the way in which he handles definitions formally within his
logical systems.
However, fourteen years earlier Frege had claimed to be able to defme
identity, to sho\v, that is, that it is not a 10gicaUy simple notion. Presumably
his idea was 10 use Leibniz's law of the icknlily of indiscernihles,
Semantics and the double d¢nition stroke
153
according to which a is identical to b if a and b share all their properties.
In "Boole's Logical Calculus and the Begrilfsschrift' (1880-81) Frege
argued that his own formula language is superior to Boole's partly because
Boole uses a greater number of signs. Indeed I too have an identity sign, but I
use it between contents of possible judgment almost exclusively to apply the
stipulation of sense of a new designation. Furthermore I now no longer regard
it as a primitive sign, but would defme [erkliirm] it by means of others. In that
case there would be one sign of mine to three of Boole's.
(Frege 1979: 35-6; 1969: 40)
This appears to indicate both that Frege previously
identity as a
primitive notion, and that he came, at least for a time, to change his mind
about this. 27 This may be connected with his doubts about the treatment
of identity he propounded in the Begriffsschrifl (1879). There Frege held
that identity is a relation between names and not objects, writing:
Identity of content differs from conditionality and negation in that it applies to
names and not to contents. Whereas in other contexts signs are merely representatives of their content, so that every combination into which they enter
expresses only a relation between their respective contents, they suddenly
display their own selves when they are combined by means of the sign for identity of content; for it expresses the circumstance that two names have the same
content. Hence the introduction of a sign for identity of content necessarily
produces a bifurcation in the meaning (Bedmt:uniJ of aU signs: they stand at
times for their content, at time for themselves.
(Frege 1879: §8)
Of course, Frege is aware that it will strike his readers as odd that names
come to change their BtJeutungen in contexts where the identity sign
appears between them. He even raises the spectre of an attitude toward
identity later pursued by Wittgenstein in the 'TractlltUS:
At frrst we have the impression that what we are dealing with pertains
merely to the expressiDn and not to the tJwught, that we do not need different
signs at aU for the same content and hence no sign whatsoever for identity
of content.
(Frege 1879: §8)
But Frege immediately calls this "an empty illusion" and argues that the
sign for identity of content is needed whenever we have two different ways
of "determining" a conceptual content, for, "that in a particular case two
WI!JIs of detnmining [the content} really yield the same result is the content of a
judgment' (1879: §8). He also points out that "a more extrinsic reason for
the introduction of a sign for identity of content" is our need for definitions, it being "at times expedient to introduce an abbreviation for a
lengthy expression" (1879: §8). But it does not seem that Frege conceives
of dther definitions or of the shift in the meaninK of the names which
j'
I
154 Juliet Floyd
I'
!;
takes place in identity contexts in terms of a clear distinction between
meta and object level. Rather, he treats identity contexts as shifting the
Bedeutungen of certain names within the very same language.
Frcgc came to surrender his Begriifsschrifl theory of identity. 28 By
1890-1 he had divided up his original notion of conceptual content into
two aspects, via the notions of Sinn and Bedeutung, and he used these
notions to explain the relation between names and an object in both informative and uninformative truths of identity. Whether or not this is
connected with his argument that identity is a logically primitive (undefmable) notion, after 1884 he never again claims that identity can be defmed.
In the Gruruigeset<,e identity is taken to be a logical primitive. 29
Despite his vacillation about the nature of the identity relation and the
nature of our statements about that relation, Frege seems never to have
./iJrmal!v. However, we
vacillated ahout how to handk
that just as there are competing ways of reading Frege's discussions of
truth, meaning, and sense, so there are competing ways of reading his use
of the double stroke for defmitions. I shall focus most closely on the use of
the double stroke in the Grundgesetze, which, I believe, should be the
primary testing ground for any interpretation of Frege.
By closely following Frege's textual divisions in volume I of the
Grundgeset;:;e, we may and should, at least prima facie, distinguish several
different ways in which Frege discusses-and uses-his formula language.
First, in his Introduction, he presents, entirely in German, a series of what
he calls philosophical arguments about the nature and scope of logic.
Second, in Part I, Chapter I he introduces his basic logical notions (such as
fimction, concept, obJect, vahle-range, and generality) and presents the primitive
signs, the Basic Laws and the formal rules of inference of the formula
language. Third, in Chapter 2, Frege discusses the nature of definitions,
introduces the double stroke, and then uses it to frame what he calls besondere, that is, "special" definitions. These definitions are applied later in
and Basic
proo[q through substitutions. Fregc also lists the Baqic
Rules and "adds a few supplementary points" (1893: §§47-8).
Part II of Volume I, where Frege's proofs occur, falls into two kinds of
sections (cf. 1893: §53). On the one hand there are the Z"legungen, or "analyses," where Fl'l'gc sketdu's lhe
e!l'Il11'lIts of his pmols ill a
mixture of German, signs of the formula language, and names of judgments, that is, (expressions of) judgments surrounded by quotation marks. 30
Each such Z"legung is set off in a separately numbered section, and is
immediately followed by another section presenting a precise construction,
or AtdfJau. An At.ifbau is used to construct proofs. It contains primitive signs of
the formula language, indices and markers to help us read the A'!fbau, and
signs that have been explicitly introduced by means of prior defmitions. It
Semantics and the double definition stroke
155
contains no words of German or of any other natural language. An A"!fbau
always consists of an annotated series of judgments, each prefaced by the
judgment stroke, and each one of which either (i) is an identity indexed with
a capital Greek letter referring back to an explicit defmition given in a prior
Zerlegung, or else (ii) follows by means of Frege's explicit inference rules from
Basic Law(s) or from truths previously asserted in earlier Atdhauten. By
contrast, the Zerlegungen contain no formal proofs. Indeed, the Zerlegungm
contain no judgments in Frege's special sense, no uses of the judgment
stroke to recordjudgments. The Zerlegungen contain on{y words f!f German, as weU
as names f!f jutigmmts and occurrences f!f identities preceded l!J the double strolce which
rm not surrounded l!J quotation marks. Writes Frege,
In connection with the following proofs I emphasize that the implementations
which I always mention under the heading "Zerlegung' serve only for the ease of
the reader; they could fail to appear without lessening the power of the proofs
in any way, for that
is to be found only under the heading "Acifbau".
(Frege: 1893: §53)
Now what, precise{y, does Frege say about the use he makes of his double
stroke for defmitions? We are interested in the bearing this question has on
the appropriateness of ascribing to Frege a sharp distinction between his
use and his mention of his formula language, between his procedures at
what we might naturally calI his object level and what we might naturally
calI his meta level. It is striking that Frege so sharply distinguishes between
contexts of formal proof, his AtdfJauten, and contexts in which he gives definitions and discusses his formula language, his Z"legungen. The
anti-semantical tradition would, presumably, deny that the Zerlegungm are
part of any substantive metatheory for the formula language. 3l I thus ask
the following interpretive questions:
I Is Frcge's double stroke a (primitive) sign of his formula language?
2 If it is not a sign used within Frege's formula language, but rather a sign
ahout the formula language, does Frege's use of the double stroke in his
defmitions function as elucidation of his formula language, or rather as
part of a metatheory of his formula language?
3 If the double stroke is part of a metatheory, is this a genuine semantical
tht'ory or just a mtla.fVntactic tht'ory?
The answers to these questions, as I shall now try to explain, are not as
readily available as one might think.
In both the Begriffischrifl and in the Grundgeset;:;e the double stroke indicates the operation of derming one sign in terms of others, and this
operation always has a double role. Frege's double stroke reflects what he
used to
calls the "double-sidedness of formulas" (Doppelseitigkeit tier
d
156 Juliet Floyd
j
I
I'
r
il
!
I
)
(i
,
i
express his defmitions (cC 1879: §24). From 1879 on, whenever Frege uses
the double stroke to frame a definition, he takes himself to be putting
fOlWard a stipulation, something that is not a judgment at all (1879: §24). Yet
any such stipulation can, he maintains, be "immediately transformed" into
a judgment of identity and used in logical constructions, in proofs within his
system. (In the 1879 Begriifsschrifl Frege calls these judgments "analytic"d 1879: §24.)32 And this is how Frege actually proceeds. He frames an
explicit definition with the double stroke and then peels away the first of
the two vertical strokes in order to assert a corresponding identity judgment in the context of an A,ybau.
Thus in its first role, a defmition stipulates that definiens and
defmiendum are to share the same content (according to Frege's 1879
view), and the same Sinn and Bedeutung (according to Frege's post-1890
view). On Frege's early account of identity,
I
II
I
"
!Ii
dt!
ijl
:'I.
II
L
,I
r
t
n
'I
in
LA definition is a proposition which] differs from the judgments considered up
to now in that it contains signs that have not becn delined before; it itself gives
the definition. It does not say "The right side of the equation has the same
C'flntent as lhc left," but "It is to have the same content." Hence this proposition
is 1I0t II jUtIWlIcllt, 1Il1l1 c'nllsrclw"lItly 'lIll a .!l',,'''dif ';rl/!A""ml rithrr. tn lise thl'
Kalltiall expression.
(Frege 1879: §24)
And on Frege's later account of identity,
LThe douhle stroke of definition J is uscd in place of the judgment-stroke where
something is to be, not judged, but abbreviated by definition. We introduce a
new name by means of a d!finitUm by stipulating that it is to have the same sense
LSinn] and the same meaning [Bedeutung] as some name composed of signs that
are familiar.
(Frege 1893: §27)
In its second role, a defmition is immediately "transformed" into ajudgment
in an Atifbau. This judgment says that the two signs do share the same
content--or, after 1890, the same Sinn and the same Bedeutung. Frege
conceives of the double stroke as literally picturing this dual aspect of defilil'st, vertical hal'
the stipulation; its
verlical har,
nitions:
which looks just like the judgment stroke, indicates the transition from the
stipulation into a judgment of identity. The judgment is only expressed, i.e.
asserted, in an A,ybau, in proofs where it is prefaced by the judgment
stroke. But the transition from a stipulation to a judgment is immediate
and unexplained. 33
How are we to interpret Frege's use of the double stroke? One suggestion is to build on our contemporary distinction between meta- and object
language in characterizing Frege's procedures. Then we might hold that
the double stroke is not a primilive sign of Frege's formula language. After
SemantUs and the double difinition stroke
157
all, it occurs only within the Zerlegungen 34 , and never within Atifbauten. Its
role, we might claim, is relegated to a place "outside" the workings of the
formula language itself. Furthermore, because an A,ybau conlains defined
terms which are not primitive signs of the formula language, we might
also hold that an ArdlJau rifers to the "pure" formula language itself, in
which formal derivations are written sole{y with the use of primitive signs
of the formula language. This I shall call the
reading of the
double stroke. 35
On this reading Frege is seen to treat the double stroke for defmitions
as a metasyntactic device: it is not part of his formal system. Every use of
Frege's double stroke in a Zerkgung is coded into a corresponding Atifbau
through its index, a capital Greek letter. This letter must be written in the
margin at every line of an Arifbau where the corresponding judgment of
identity appears. So, on the metasyntactic reading, the double stroke is an
explicit sign of an algorithm which "decodes" every ArdlJau of the
Grundgesetz:.e into a proof couched solely in primitive signs of the formula
language. (fhe ArdlJauten on this view are like "programs" which can be
alF;orithmic.ally carried into an "assembly Ir.vellanguage.") In the primitive
fbnllula lanKllaKc' il:wll' all "ddinitional"
ilklllily vanish.
This might be held to explain why Frege repeatedly remarks that Irolll the
perspective of the formula language, definitions are "mere abbreviations,"
convenient shorthand devices used to make logical constructions,
A,ybautm, perspicuous. Differently put, Frege's Zerlegungen, his analyses, use
definitions in order to present his logical constructions in such a way that
they can be taken in by us:
or
The need of defmitions never ceases to be apparent in any attempt of this
sort.... The definitions do not really create anything, and in my opinion may
not do so; they merely introduce abbreviated notations (names), which could be
dispensed with were it not that lengthiness would then make for insuperable
external difficulties.
(1893: Preface, vi, my emphasis)
In the Begriffsschrifl the same point appears to be made twice:
IiII' the iutrotluclillll of a
lor
of l:nnlent is
A ilion' exhinJic
that it is at times expedienl to introduce an ahhreviation for a lengthy
expression ...
(Frege 1879: §8, my emphasis)
[A defmition] is not a judgment, and consequendy not a
either,
to use the Kantian expression.... If [it] were a synthetic judgment, so would
be the propositions derived from it. But we can do without the notation introduced by
proposition and hence without the proposition itself as its
definition; nothing follows from the proposition that could not also be inferred
without it. Our sole purpose in introducing such definitions is to bring about an
extrinsic simplification by stipulating an abbreviation. They serve besides to
Semantics and the double definition stroke
158 Juliet Floyd
emphasize a particular combination of signs in the multitude of possible ones,
so that our faculty of representation can get a firmer grasp of it.
(Frege 1879: §24, my emphasis)
According to the metasyntactic reading, an ArdlJau of the GnmdgesetJ:J
which is not written exclusively in the primitive formula language represents
a genuine proof; that is, it presents a precise set of rules telling us how we
are to write down a particular proof in primitive signs of the formula
language. We can take it in as a representation because it uses abbreviations. But we know it successfully represents a genuine proof only insofar
as every definitional judgment of identity occurring in it can be eliminated in a mechanical way, with no appeal to elucidations, hints, or
guesswork.
Of course, this "metasyntactic" reading of the double stroke cannot
wholly sever the signs it speaks of from their Sinne and from their
&deutungen. When Frege speaks of a defmition as a "stipulation," he does
not mean that a defmition tells us to substitute one uninterpreted sign for
another uninterpreted series of signs. Rather, the stipulation appeals to the
content, to the Sinn and the Bedeutung, of the definiens. As he writes in
"Function and Concept," "In defmition it is always a matter of associating
with a sign a sense or a meaning. Where sense and meaning are missing,
we cannot properly speak either of a sign or of a definition."
Thus the metasyntactic reading inevitably spills over into some sort of
metasemantical reading. The double stroke may be said to function as a
paradigmatically semantical sign insofar as it tells the reader how to assign
Sinne and &deutungen to signs. Nevertheless, even though the assignment of
Sinne and &deutungen to newly introduced signs may take place in a metalanguage, the terms of this language have, at least so far, no more, no less
and no different semantical content than what is already gotten--directly
and immediate1y-out of the object language. Thus, the question as to
whether this metalanguage consists of mere elucidation, as opposed to an
articulate semantical or philosophical theory, remains unresolved-and
perhaps unresolvable-by this reading.
Moreover, in an unpublished manuscript which further develops his antisemantical reading of Frege, Ricketts explicitly denies that Frege's double
stroke for defmitions is in any sense a metasyntactic device. 36 According to
Ricketts, just as the judgment stroke uses the formula language to make
judgments-· something we do within IUIIRuagc -so the double stroke uses
the formula language to defme one sign in terms of another-something
else we do within language. On his view, the double stroke is part of Frege's
formula language, and defmitions take place within the universal formula
..
159
language. Thus does Ricketts interpret the immediacy of Frege's transitions from stipulations to judgments of identity.
If Ricketts is correct, then Frege's system has no precise, closed, vocabulary, and there is no set of well-formed formulas given by Frege once and
for all. Instead, through the use of the double stroke, the formula
language is systematically enlarged in a step-by-step way. This enlargement helps us to work with the formula language more easily, and this is
the point of Frege's definitions. On this reading there is no room to view
the Atifbauten as metasyntactic representations of the primitive formula
language, for there is no formula language with a fIXed vocabulary. Frege
had no closed formalism at all! Instead, the ArdlJauten function by means of
the formula language in the context of a carefully regulated, growing
vocabulary. The additions of new vocabulary in defmitions function like
special axioms of identity which are always conservative extensions of
Frege's original primitive system-though as Ricketts emphasizes, Frege
nowhere attempts to prove such a metatheoretic statement. On Ricketts'
reading, the uses of the double stroke, though not part of Frege's logical
constructions, are formalized in the Arifbauten by means of the judgments
of identity which are indexed to the explicit definitions given in the
Zerlegungen. Defmitions are set off from ArdlJauten because they are not
judgments, but, rather, stipulations. But they are still "part of the system"
in the sense of being framed within the universal formula language, they
are not stipulations or abbreviations made from a standpoint external to
the system or the language. Thus Ricketts holds that Frege did not have a
metasyntactical theory, but only rules for writing proofs down in his
formula language.37
This shows how deeply the debate between the semantical and the
anti-semantical readings of Frege may be said to penetrate our understanding of Frege's philosophy and logic. For we have been told--even by
van Heijenoort himself-that Frege was the first to frame precisely the
notion of a formal system (1985: 12). If Ricketts is right, then even this
understanding of Frege may have to be revised, even if only slighdy. My
point has been to emphasize the difficulties we face in attempting to
extract defmitive answers about Frege's overarching enterprise, and even
about the structure of his formula language. To read Frege we are
constantly thrown back on our own philosophical predilections---even at
38
those points where Frege's procedures may seem most clear.
NOTES
This chapter is slated to appear in French in Frege, Logique MathimaJique et
Philosophie, ed. Mathieu Marion and Alain Voizard (paris: I'Harmattan). I
160
2
3
4
5
6
7
8
9
}ulift Hr!vd
hereby give my permission for the paper to be printed there and for my rights
to it to be transferred to Routledge for the present volume.
I am indebted to the English translations of Frege cited, though I have
occasionally altered them when I felt it necessary. Unless otherwise specified,
page numbers refer to the original page or section number of Frege's works.
Van Heijenoort, "Logic as Calculus and Logic as Language," (1985: 12-13):
"Boole has his universe class, and De Morgan his universe of discourse,
denoted by '1'. But these have hardly any ontological import. They can be
changed at will. The universe of discourse comprehends only what we agree
to consider at a certain time, in a certain context. For Frege it cannot be a
question of changing universes. One could not even say that he restricts
himself to one universe. His universe is the universe."
Van Heijenoort, "Frege and Vagueness," (1985: 95). Van Heijenoort is
quoting from Quine's "Facts of the Matter" (1977), and holds, with Quine,
that these words accurately describe Frege's conception of his own enterprise.
Of course, the whole point of Quine's thought experiment about the indeterminacy of translation is to undercut the view that there is a "latent and
determinate content" to be elicited from ordinary language. Compare Burton
Dreben, "Pumam, Quine and the Facts" and Hilary Pumam's reply in Hill
(ed.) (1992).
Compare Michael Dummett, The Interpretation c!f Frege's Philosopf!y (1981: 13,
17-19).
Goldfarb wrote: "Frege has often been read as providing all the central
notions that constitute our current understanding of quantification. For
example, in his recent book on Frege (Frege: Philosophy c!f Language, ftrst edn),
Michael Dummett speaks of 'the semantics which [Frege] introduced for
interpretation of
formulas of the language of predicate logic.' That is,
such a formula ... is obained by assigning entities of suitable kinds to the
primitive nonlogical constants occurring in the formula ... [T]his procedure
is exactly the same as the modern semantic treatment of predicate logic'
(pp. 89-90). Indeed, 'Frege would therefore have had within his grasp the
conrepts necessary to frame the notion of the completeness of a formalization of logic as well as its soundness ... but he did not do so' (p. 82). This
common appraisal of Frege's work is, I think, quite misleading" (Goldfarb
1979: 351).
I use the term "anti-semantical" as shorthand for "anti-theory of meaning"
or "anti-theory of semantics." It should be clear to the reader from what has
already been quoted that in a loose or intuitive sense of the term, Frege lives
in semantics. My overview of this "anti-semantical" tradition is highly
schematic, and obviously cannot do justice to the specifics of each individual
member's view of Frege.
Others who have discussed and characterized this tradition and its relation to
Dummett include Richard Heck (forthcoming); Jason Stanley (forthcoming);
and Sanford Shieh (forthcoming). James Conant's article "The Search for
Logically Alien Thought: Descartes, Kant, Frege and the Tractatus" is replied
to by Hilary Putnam (1992). See also Kemp (1995).
Note that "Erlliuterung" is not systematically translated (Cf. 1969: 224,232,
254).
An especially subtle working out of the sorts of paradoxes Frege faces in
introducing his system is contained in de Rouilhan (1988).
Sm/antics and the double ddinition stroke
161
10 Passages where Frege calls attention to such ideas occur in "On Sense and
Reference" (1980a: 61; 1984: 161; 1967: 145); see also "Concept and
Object" (1980a: 54; 1984: 193; 1967: 177; 1980b: 37; 1976: 63; 1893, 1903:
Appendix 2, n I, 1969: 288ft).
II For example, in "What Nonsense Might Be" (1991a: 97) Cora Diamond
explicitly criticized Dummett's reading of Frege, emphasizing that: "On the
Frege-Wittgenstein view, if a sentence makes no sense, no part of it can be
said to mean what it does in some other sentence which does make sense-any more than a word can be said to mean something in isolation. If 'Caesar
is a prime number' is nonsense, then 'Caesar' does not mean what it does
when it is in use as a proper name, and the last four words do not mean what
they do in sentences which make sense" (Diamond 1991 a: 100). "I have
wanted
parallels (between Frege and Wittgenstein) to be suggested. I
have
wanted to suggest ... a distance between Frege's view and what
might be called a Tarskian view" (199Ia: 112).
12 All of Ricketts' work is relevant, but see especially 1985 and 1986b.
13 Frege's arguments that truth is indefmable are canvassed by Ricketts in
"Objectivity and Objecthood" (1 986b), as well as in "Logic and Truth in
Frege" (1996a).
14 See Diamond (199Ia: ch. 4), Sluga (1980: 182), Ricketts (1985), and Weiner
(forthcoming).
15 Like Dreben, Goldfarb and van Heijenoort, Baker and Hacker draw a
parallel between Frege and Russell: "It is salutary ... to remember how
modern the semantic conception of validity is. Although Russell acknowledged a profound debt to Frege 'in all questions of logical analysis', Principia
Mathemati£a did not formulate a clear distinction between logical truth and
provability within its axiom system, and hence it did not envisage the possibility of proofs of consistency, independence, and completeness" (1984: 373).
16 Compare van Heijenoort, "Logic as Calculus and Logic as Language" (1985:
14), and Goldfarb (1979).
17 Compare van Heijenoort, "Logic as Calculus and Logic as Language" (1985:
14). Notc that in a footnote on p. 95 of ft'incipia Mathtmatica (1910 ednj p. 91
1925 edn) Russell and Whitehead explicitly refer to the 1903 passage quoted
by Goldfarb (see p.l44 above). It might be added that even as late as Russell's
Introduction to Mathmuzti£al Philosopf!y (1919) we ftnd Russell maintaining that
"The theory of deduction ... and the laws for propositions involving 'all'
and 'some,' are of the very texture of mathematical reasoning: without
them, or something like them, we should not merely not obtain the same
results, but we should not obtain any results at all. We cannot use them as
hypotheses, and deduce hypothetical consequences, for they are rules of
deduction as well as premisses. They must be absolutely true, or else what we
deduce according to them does not even follow from the premisses" (Russell
1919,1920: 191).
18 Dreben, "Frege on FOundations," 4/6/92 lecture to the Boston University
Colloquium for the Philosophy of Science. Quine responded as a member of
the audience.
19 See, for example, Thomas Ricketts (1996a). Ricketts argues that for Frege
truth is not a genuine property, and truth not a genuine concept. Hence Frege
would not have been able to accept, without significant modification of his
standpoint, the Tarskian criterion of material adequacy as relevant to any
lUI
r'"
........
".I
• . '"
... . , - - - - . - - - . - ..
*
e....
IS.
20
21
22
23
24
25
26
27
28
29
...........-.-
...
Semantics and the double definition stroke
162 Juliet Floyd
truth definiJion. A fortiori, from Frege's point of view the Godel completeness
theorem, formalized in the &griffischrijl, could not give any explanation or
account of the notion of "logical truth." Compare Ricketts' "Objectivity and
Objecthood" (1986b: 76, especially footnote 18).
For an investigation of a similar response, namely Wittgenstein's response to
Godel's first incompleteness result, see my "On Saying What You Really Want
to Say: Wittgenstein, (j()del and the Trisection of the Angle" (Floyd 1995).
In the very late "Logical Generality" (1923 or later) Frege began to articulate
a distinction between what he calls Hilfssprache and Darlegungssprache, which
could be taken to anticipate Tarski's later distinction between metalanguage
and object language. See Frege 1979: 260lf., 1969: 287lf.
We can also discuss, in set theory, the capacity of a formal theory to define,
i.e., to contain, formulas model-theoretically satisfiable by certain relations
and
Such a discussion would I Ihink be quilc fon'iRJl 10 Frege, but I shall
nol argue Ihis point here.
I am grateful to J. Weiner for having pointed
passage out 10 me in her
and the OriRins of Analytic Philosophy" talk.
(1!)/llla: /I(); 1%7: 1/11).
holds that
Scc H'cgc's revicw of
Frege's argument in the Husser! review that identity is a logically primitive
notion is "not very convincing" since "it is only possible for Frege to say this
because he takes the sign of idenlity to do duty also for the biconditional,
which is in turn possible only because he assimilates sentences to names, viz.,
of truth-values; and in any case it seems more natural to take a defmition as a
stipulation of the interchangeability of two expressions, rather than of the
truth of a sentence connecling them.... [rJhe thesis of the indefinability of
identity does not seem to play any important role in Frege's philosophy"
(Dummett 1981: 543). Frege in one way does and in another way does not
countenance a sharp distinction between stipulating interchangeability
among (linguistic) expressions and capturing, analyzing, proving, and
asserting by means of a real defmition. Compare Benacerraf, "Frege: the Last
Logicist" (1995).
What is "logically simple" cannot, according to Frege, be defmed (c( 1903:
§146, n. I; 1980a: 42).
Rosalind Carey pointed out to me that in "Insolubilia and Solution by
Symbolic Logic" Russell writes that "the notion of defmition ... oddly
enough, is not defmable, and is indeed not a defmite notion at all" (1906
essay, reprinted in Russell 1973: 209). The remark occurs in the context of
Russell's treatment of the Richard Paradox and the Vicious Circle principle.
Tappenden discusses Frege's paper on Boole's logical calculus, and suggests
that between at least 1880 and 1884 Frege developed an interesting and novel
account of the fruitfulness of definitions and concepts-albeit a view which
came into tension with the later Sinn/BedeutuTlf. distinction. Tappenden argues
that qualllificational cOlllplexity in a
corresponds to fruitfulness of a
defmition at this stage of Frege's development. This may be related to Frege's
claim to be able to define identity, although I cannot pursue this conjecture
here. See Tappenden 1995. Compare Picardi 1988.
For an informative account of the difficulties facing Frege's early view, see
Mendelsohn 1982.
Basic Law III in the
formalizes Leibniz's law, but does not purport
in any way to be a defmition of identity. Section 65 of the GnmJItJgen could be
•.•
30
31
32
33
34
35
36
37
38
163
read as explicidy advocating Leibniz's law as a defmition of identity,
although even in the Husser! review Frege mentions the law as expressing
something important about identity (see Frege 1984: 20). Jan Harald Alnes
has persuaded me that it is doubtful that in the Grundlngen Frege held that
identity could be defined.
More precisely, the Zerlegungen contain names of judgments which will be
made in the corresponding Al!fbau, and some of these judgments will contain
defmed terms, as well as primitive terms of the formula language.
This denial is quite explicit in Weiner 1990: ch. 6.
The word "analytic" does not occur in Grundgesetze. However, in 1914 Frege
says that the sentence expressing the judgment of identity corresponding to a
defmition is "a tautology" ("Logic in Mathematics," in 1979: 208).
(Presumably he writes this qfter he and young Wittgenstein had spoken.)Jan
Harald Alnes has suggested in convenation that sameness of sense on two
sides of a lrue identily statement
only whrn Ihe identity springs from
an explicit defmition or is an instance of a basic law. This would explain why,
even alier introducing the Sinn/lkdeutung
Frege holds in
and
that Ihe Iwo sid('s of an inslance nf Ra.qjc lA'\w V
"express the same sense, but in a dillerenl way," while simultaneously holding
that derived truths (such as "2+2=4") involve expressions with differing
scnses on eilher side of the identity sign (Frege 1984: 143; 1967: 130).
This formal point points toward several unclarities in I'rege's attitude toward
defmitions in the context of his "reduction" of arithmetic to logic. See, in
particular, lknacerraf 1995, and the response to Benacerraf by Weiner
1984. Sec also Picardi 1988, Dummell 1991b (ch. 2), and Tappenden 1995.
And, of course, in the section on "special definitions" in Grundgeset.c:.e Chapter
I and in the table of definitions in
Appendix.
I am aware that there is a long established usage in which the metalanguage
is dubbed the "syntax language" of the object language. In this sense what I
am calling "metasyntactic" would be called "syntactic."
In conversation, correspondence and in Chapter I section 3 of his unpublished manuscript on Frege.
Further evidence for Ricketts' view, as Ricketts himself has suggested, is
§48 # 12, labeled "Citation of Definitions." Here, according to
Ricketts, Frege sets out the usc of the double stroke as a rule of inference of
the formula language. A critic of Ricketts' view would have to claim that this
section, rather than setting forth a rule of inference, is giving what Frege
called in §47 a "supplementary point." Note, however, that in Grundgesetze
§53, Frege does speak of "the rules to which I appeal in the Zerlegungen" and
refers the reader back to 48.
I was stimulated to think about Frege's treatment of definitions as a result of
responding to papers by Hans Sluga and Joan Weiner at a December 1995
American Philosophical Association symposium on "Frege and the Origins
of Analytic Philosophy." Both Sluga and Weiner argued against 5CVtral
widespread interpretations of Frege's distinction between Sinn and
and their arguments got me to think more carefully than I had about Frege
on identity and defmitions. Thanks are due to them for their stimulation,
both at the symposium and in other published works. I also thank Anat
Biletzki and Anat Matar for their enthusiasm and generosity in organizing
the January 1996 Tel Aviv conference at which an earlier version of this
in
1
•
Semantics and the double tkfinition stroke
164 Juliet Floyd
paper was read. The audiences at Tel Aviv and at Oslo University in June
1996, contributrd
for which I am Kl'ateful.
My understanding of Frege and controversies about how to read him is
enormously indebted to Burton Dreben, and I thank him for our many
hours of discussing Frege since 1983, and for his sage editorial and philosophical advice at every stage of this paper's composition. Jan Harald Alnes,
Rohit Parikh, and Sanford Shieh have also offered helpful comments on
dralls of my paper, as has Thomas Ricketts,
manuscript 011 Frege's
philosophy has been especially instructive, especially its Chapter I, §3 on
Frege on defmitions.
BmUOGRAPHY
Baker, G. P. and Hacker, P. M. S. (1984) 1JJgieal Excavations, Oxford: Oxford
University Press.
Benacerraf, P. (1995) "Frege: The Last Logicist," in W Demopoulos (ed.) Frege's
of MathemoJics, Cambridge, Mass.: Harvard University Press.
Conantj. (1992) "The Search for Logically Alien Thought: Descartes, Kant, Frege
and the Tractatus," in Hill, C. (ed.) Philosophical Topics 20/1.
de Rouilhan, P. (1988) Frtge: us Paradoxes de La Rtprbmlation, Paris: I,rs Editions de
Minuit.
of Mathematics, Cambridge, Mass.:
Demopoulos, W (ed.) (1995) Frege's
Harvard University Press.
Diamond, Cora (199Ia) The Realistic Spirit, Cambridge, Mass.: MIT Press.
---{I 991 b) "Ethics, Imagination and the Method of Wittgenstein's Tractatus,"
R. Heinrich and H. Vetter (eds) Bilder tier Philnsophie, Wzener R£ihL 5, 55-90.
Dreben, B. (1962-96) Unpublished lecture transcriptions.
--"Frege on Foundations," 4/6/92 lecture to the Boston University Colloquium
for the Philosophy of Science, unpublished transcription.
---{1992) "Putnam, Quine and the Facts," in Hill, C. (ed.) Philosophical Topics
20/1.
Dreben, B. and van Heijenoort,j. (1986) "Introductory note to 1929, 1930 and
1930a," in S. Feferman, J. W Dawson, Jr, S. C. Kleene, G. H. Moore, R. M.
Solovay,J. van Heijenoort (eds) Kurt GOrkl, Collected Papers, L New York, Oxford:
Clarendon Press.
of Language, Cambridge, Mass.:
Dummett, M. (1973, 1981) Frege:
Harvard University Press.
--- (1978) Truth and Other EniRma-r, Cambridge, Mass.: Harvard University Press.
--(1981) The Interpretation of Frege's Philosophy, Cambridgc, Ma.,s.: Harvard
University Press.
--{1984)
Unsuccessful Dig: Critical Notice of G. Baker and P. Hacker,
lAgieal Excavations" in C. Wright (ed.) Hege: Tradition and Iriflumce, Oxford:
Blackwells.
(1987) "Review of Kurt GOdel Collected J#J'*s," Mind96: 570-5.
---{1991 a) The LJgieal Basis qf
Cambridge, Mass.: Harvard University
Press.
--{1991 b) Frege and Other Philosophers, Oxford: Clarendon Press.
---{1993a) The Seas of lAnguIlge, Oxford: Oxford University Press.
---{1993b) Origins of AnalYtic
Cambridge, Mass.: Harvard University
Press.
---(1994) Frege: Philosophy
165
of Mathematics, Cambridge, Mass.: Harvard University
Floyd, j. (1995) "On SayinK What You Really Want to Say," in j. Hintikka (ed.)
F1rJm 1Jdlkind to Giitkl, Dordrecht: Kluwer.
Floyd,J. and Shieh, S. (eds) (forthcoming) Futurt Pasts, Cambridge, Mass.: Harvard
University Press.
Frege, G. (1879) &grilfsschrift, Halle, Verlag von Louis Nebert, second edition,
(ed.) Hildesheim: Georg Olms (1964); trans.J. van Heijenoort
Ignacio
(ed.) From Frege to GOdel: A Sourcebook in Mathematical Logic 18791931.
---(1884) Die Grundlagen rkr Arithmetik, Breslau; trans. J. L. Austin as The
Foundations of Arithmetic (2nd edition, Oxford, 1953; reprinted by Northwestern
University Press, 1980).
---{1893, 1903) Grundgeset<.e rkr Arithmetik, Vol. I, Vol. II; reprinted by Hildesheim:
Georg Olms Verlagsbuchhandlung (1966); partially translated into English
(through s.52) as The Basic Laws of Arithmetic by M. Furth, Berkeley: University
of California Press (1964).
---{I 967) Kleint Schriftm, I. Angelelli (ed.) Hildesheim: Georg Olms
Verlagsbuchhandlung.
---{I 969) Nachgelassene Schriftm, H. Hermes, et aL (eds) Hamburg: Felix Meiner.
- - ( 1971) On the RJundations of Geometry and Formal Theories of Arithmetic, trans.
Eo Kluge, New Haven: Yale University Press.
--{I 976) WlSsenschtifllicher Bril.fivechsel, G. Gabriel et at. (eds), Hamburg: Felix
Meiner.
---{1979) Posthumous Writings, Hans Hermes, Friedrich Kambartel, Friedrich
Kaulbach (eds), trans. P. Longand R. White, Chicago: Universityof Chicago Press.
---{1980a) Translationsfrom tire Philosophical
of Gottlob Frege, trans. M. Black
and P. Geach (eds), third edition, Totowa, 1'{J: Rowman and Littlefield.
- - (1980b) PhilosophicalandMathematical Correspondence, G. Gabriel etal. (eds), abridged
by B. McGuinness, trans. H. Kaal, Chicago: Universityof Chicago Press.
--(1984) Collected Papers on Mathematics, 1JJgie and
B. McGuinness (ed.),
trans. M. Black et al., Oxford: Blackwell.
Gt'idel, K. (1986) Kurt GOrkl, Collected Papers, 1, S. Feferman,J. W Dawson,Jr., S. C.
Kleene, G. H. Moore, R. M. Solovay, J. van Heijenoort (eds) New York:
Oxford, Clarendon Press.
Goldfarb, W. (1979) "Logic in the Twenties: the Nature of the Quantifier," Journal
of Symbolic Logie 44,3: 351-68.
Heck, R. (forthcoming) "Frege and Semantics," in Ricketts (ed.) The Cambridge
Companion to Frege, Cambridge: Cambridge University Press.
Hill, C. (ed.) (1992) Phiiosophical10pics 20, I: 114-80.
Hintikka,j. (1979) "I<'rcge's Hidden Semantics," Revue Inttrnationale de Philosophie 33:
716-22.
-and Hintikka, M. (1986) Investigating Wlttgenstein, Oxford: Blackwell.
Kemp, G. (1995) "Truth in Frege's 'Law of Truth'," SyntJrese 105: 31-51.
Marion, M. and Voizard, A. (eds) (forthcoming) Frege, 1JJgifjue MaJJrJmatique et
Philosophie, Paris: I'Harmattan.
Mendelsohn, R. (1982) "Frege's Begrilfsschrift Theory of Identity," Journal of tire
History of Philosophy XX, 3: 279-99.
Picardi, E. (1988) "Frege on Defmition and Logical Proof," in Atti rkl Congresso Temi
e prospettive rklla logiea e rklla filosofia della scien;:,a contemporanu, Cesena 7-10
gennaio 1987, Vol. I CLUEB, Bologna Italy: 227-30.
166
Juliet Floyd
Putnam, Hilary (1992) "Reply to Conant," "Reply to Dreben," in PhilJJsophi£al
Topics 20, I.
Quine, W V. 0. (1977) "Facts of the Matter," in R. W Shahan and K. R. Merrill
(eds) American
from Edwards to Qyine, Norman, Ok.: University of
Oklahoma Press.
Ricketts, 1: (1985) "Frege the Tractatus, and the Logocentric Predicament," NoUs
9,1: 3-15.
--(1986a) "Generalit); Meaning and Sense in Frege," Pacift PkiUisophical QyarterlY
67,3: 172-95.
----{1986b) "Objectivity and Objecthood: Frege's Metaphysics of Judgment," in
L. Haaparanta and J. Hintikka (eds) Frege SyntksiQd, Dordrecht: D. Reidel:
65-95.
---(1996a} "Logic and Truth in Frege," Proceedings of the Aristotelian Society.
----{1997) "Truth-Values and Courses-of-Value in Frege's GrwuigesetQ," in W W
Tait (ed.) EarlY Analytic PhiJDsopf!y: Essays in honor if Leonard Linsky, Chicago: Open
Court Press.
---(forthcoming) unpublished manuscript on Frege's philosophy.
Ricketts, 1: (ed.) (forthcoming) Th CambriJige Companion to Frege, Cambridge:
Cambridge University Press.
Russell, Bertrand (1903) The Principles of Mathematics, Cambridge: Cambridge
University Press; second edition (with a new Introduction), New York: W W
Norton & Company, 1938.
--and Whitehead, Alfred North (1910, 1912) Priruipia Mathematica (2 vols.),
Cambridge: Cambridge University Press; second edition, 1925.
---{ 1919) Introduction to MaJhemaJ.i.cal PhilosopJpi, London: George Allen & Unwin,
Ltd; (second edition 1920).
---{1973) Essays in AnalysU, D. Lackey (ed.) New York: George Braziller.
Shieh, S. (forthcoming) "On Frege's Logic and Semantics," in M. Marion and A
Voizard. (eds.) Frege, /..ogiIfue MathimoJique et Philosophie, Paris: I'Harmattan.
Sluga, H. (1980) Gottlob Frtge, London: Routledge and Kegan Paul.
--(1987) "Frege Against the Booleans," Notre Dame Journal if Formal LJgic 28 I:
8D-98.
Stanley,
J.
(forthcoming) "'Truth and Metatheory in Frege," Pacific PhilosophUal
Q!wterry.
Tappenden, j. (1995) "Extending Knowledge and 'Fruitful Concepts': Fregean
Themes in the Foundations of Mathematics," NoUs-. 427-67.
van Heijenoort,j. (1985) Sel«ted Essays, Naples: Bibliopolis.
van Heijenoort, J. (ed.) (1967) From Frege to Giid8/: A Sourcebook in Math.em.atiud Logic
1879-1931, Cambridge, Mass.: Harvard University Press.
Weiner,j. (1982) "Putting Frege in Perspective," Ph.D. thesis, Harvard University.
--{1984) "The Philosopher Behind the Last Logicist," in C. Wright (ed.) Fregt
Tradition and Irifluence, Oxford: Blackwell.
--{1990) Frege in Perspeaice, Ithaca: Cornell University Press.
---{forthcoming)"Theory and Elucidation," in J. Floyd and S. Shieh (eds) Fu/.IIrt
Pasts, Cambridge, Mass.: Harvard University Press.
Wittgenstein, Ludwig (1922) Tractatus LJgico PhiUisophicu.r, with an introduction by
Bertrand Russell, trans. C. K. Ogden, London: Routledge and Kegan Paul
(corrected edition 1933).
Wright, C. (ed.) (1984) Frtge: Tradition and Irifluence, Oxford: Blackwell.
9 Ar
o
lli
Gil,
Since its
founders, :
its conceIT
been done
analytic p:
, believed tl:
logically t
• and presu
· quences.
.
Contrlll
· concerned
, linguistic"
, lite
.
reason
. profound!)
'. The idea'
tool of c(
• language<wrong ane
tions. And
, held to be
,1Ionds of 1;
language, <
again
; important
. everywhen
.", But
byr·
lions it
SUi