, Volume 78, Issue 3, pp 523–546

Pragmatics in Carnap and Morris and the Bipartite Metatheory Conception


    • Philosophy, School of Social SciencesUniversity of Manchester
Original Paper

DOI: 10.1007/s10670-011-9352-5

Cite this article as:
Uebel, T. Erkenn (2013) 78: 523. doi:10.1007/s10670-011-9352-5


This paper concerns the issue of whether the so-called left wing of the Vienna Circle (Carnap, Neurath, Frank) can be understood as having provided the blueprint for a bipartite metatheory with a formal-logical part (the “logic of science”) supporting and being supported by a naturalistic-empirical part (the “behavioristics of science”). A claim to this effect was recently met by a counterclaim that there was indeed an attempt made to broaden Carnap’s formalist conception of philosophy by the pragmatist Morris, but that this initiative failed and that Carnap showed no interest in it. To defend the original claim this paper provides an analysis of Carnap and Morris on the subject matter of pragmatics in order to show that and how Carnap adopted Morris’s proposals in so far as they agreed with bipartite metatheory conception.

1 Introduction

This paper addresses a discussion over the nature of the philosophy of science developed by the so-called left wing of the Vienna Circle in the 1930s.1 Against the claim that the efforts of Rudolf Carnap, Otto Neurath and Philipp Frank can be understood as providing the blueprint for philosophy of science as a bipartite metatheory with a formal-logical part (what Carnap called the “logic of science”) complementing and being complemented by a naturalistic-empirical part (what Neurath called the “behavioristics of science”), it has been countered that within the vicinity of the Vienna Circle there was indeed an attempt to broaden Carnap’s formalist conception of philosophy of science but that that initiative came neither from Neurath or Frank but from Morris and was proposed to unify the logical empiricist philosophy of the Vienna Circle with American pragmatism of the Mead-Dewey variety. Moreover, contrary to what’s asserted by the bipartite metatheory thesis, Carnap showed no interest in—nor made any opening for—anything other than his own formalist logic of science.

This interpretive disagreement has considerable bearing on how the philosophy of science developed on the left wing of the Vienna Circle may be judged. For one thing, the bipartite metatheory conception affords to the efforts of members of the left wing a theoretical cohesion that would be lacking still even if it were recognized that Carnap’s formalist logic of science and Neurath’s naturalistic behavioristics did not, after all, contradict each other.2 (On first appearances they may well do and, historically speaking, one disagreement springing from misunderstanding did run deep). For another thing, the bipartite metatheory conception thereby gives substance to the claim, made in the Vienna Circle’s unofficial manifesto authored by future members of its left wing, that the new philosophy arising from the “scientific world-conception” was a collective undertaking.3

By contrast, the criticism exploits the undeniable fact that the bipartite metatheory conception as such was not inaugurated by its practitioners with comparable fanfare but rather has to be reconstructed from occasional remarks and engagements. Such lack of programmatic affirmations would be understandable, of course, if the conception at issue were but what an earlier well-publicized program for philosophy had morphed into, gradually and in the course of not insignificant internal debate. Now that is what I think happened, but that is unlikely to assuage critics. It is fortunate therefore that Thomas Mormann has drawn attention to the rough parallel between Neurath’s and Frank’s attempt to develop what Frank in later years called a “pragmatics of science” and Morris’s proposals to unify logical empiricism and pragmatism as a test case, making Carnap’s attitude towards the latter indicative of his attitude towards the former.4

I propose to consider the matter further by offering an analysis of Carnap and Morris on the subject matter of pragmatics. Carnap’s selective use of pragmatics was also significant for his debate with Quine; my discussion will touch on this but the focus is different. It will be argued that while Morris’s program for the unification of pragmatism and logical empiricism was by no means adopted in toto, significant aspects of it were—namely, precisely those (and only those) aspects that sustain the blueprint for the bipartite conception of philosophy as metatheory. To establish this I will argue as follows. After some preliminaries on Carnap’s views on the logic of science and the landscape of theories of meaning before Morris entered the Central European scene (Sect. 1), I will show, first (Sect. 2), that Morris offered to Carnap and the Vienna Circle an organizational schema for philosophical work under the aegis of unified science that appears (with one important qualification to be noted below) broadly comparable to the bipartite metatheory conception. Second, I will show (Sects. 35) that and how this suggestion was picked up by Carnap and entertained throughout the 1940s and 1950s. Third, I will argue (Sects. 67) that Carnap’s striking failure to designate discourse about external questions as “pragmatical” does not mean that he dropped the conception of pragmatics compatible with the bipartite metatheory conception which remained alive and well amongst the surviving members of the left wing of the Vienna Circle.

2 Carnap Before Pragmatics (1934–1938)

Carnap famously wrote: “[O]nce philosophy is purified of all unscientific elements, only the logic of science remains. …The logic of science takes the place of the inextricable tangle of problems which is known as philosophy.” (1934/2002, 279) The logic of science investigated scientific theories, their internal structure and their relation to their evidential base in purely logical terms (originally only with reference to their logical syntax, but soon also with reference to semantics).5 Note that the logic of science is a second-order inquiry and what Carnap said about its relation to other second-order inquiries. Rejecting the term “theory of science” (“Wissenschaftslehre”) for the logic of science specifically, he held that designation to be “more appropriate to the wider domain of questions which, in addition to the logic of science, includes also the empirical investigation of scientific activity, such as historical, sociological, and, above all, psychological inquiries.” (Ibid., 279) Carnap here recognized as perfectly legitimate other second-order inquiries noting that these, unlike the logic of science, were of an empirical nature. Their complementary workings make for the so-called bipartite metatheory of science.

Concurrently (at the 1934 Prague Preliminary Conference for the First Unity of Science Congress in Paris) Neurath expressed the same idea under the heading “program of unified science”: “We would establish the ‘cross connections’ from science to science and thus create a structure that knows no ‘philosophy,’ no ‘epistemology’ with special propositions—whichever one of these two is applicable has found its place either in the ‘logic of science’ or in ‘behavioristics’.” (1935/1983, 115).6 This behavioristics of science included not only psychology, but also sociology and history of science and other empirical studies of science as practiced. Thus Frank had noted already in 1932: “Only the following can meaningfully be meant by ‘philosophy of natural science’ or ‘philosophy of nature’: either the precise analysis of the meaning of scientific symbols, i.e. of the relations of coordination—something which really constitutes a significant part of every natural scientific theory—or the investigation of certain observable processes concerning the relation between the theories of natural science and other expressions of human activity—something which is to be regarded as a part of scientific sociology.” (1932a, 156)

Now it will be noted that Carnap was more concerned to distinguish his logic of science from these empirical studies than to pursue them. It should be added therefore that is no part of the thesis that the left Vienna Circle shared a bipartite metatheory conception of what nowadays we call philosophy of science (at the time the term “philosophy” itself was contested), that Carnap actively pursued anything other than the logic of science in his own work. That is why the case of pragmatics is of interest: as we shall see, there too Carnap allowed for complementation of his own speciality in order to get a comprehensive account of the phenomenon under consideration without pursuing the relevant inquiries himself.

Before turning to Carnap’s views on pragmatics we must consider how Carnap conceived of conceptions of meaning before Morris’s intervention to be able to see what changed and what did not. While he abjured concern with meaning as such in his Logical Syntax, Carnap nevertheless remarked on the use of term “Semantik” (semantics) (1934/2002, 9).7 Opposing its use for anything like syntax (even if that was differently conceived of from his own like by Chwistek), he noted its earlier use by Breal (1897) and regarded it as closely linked to “Semasiologie” or “Bedeutungslehre” (theory of meaning). This “Semasiologie” considers, as part of linguistics, the meanings of historically given languages. It has to be distinguished from Bühler’s concept (1933) of “Sematologie” which denoted “the empirical (psychological, sociological) theory of the application of symbols in the widest sense.”8

These terminological remarks are notable for two reasons. First, already in 1934 Carnap recognized a division between a narrowly linguistic investigation of meaning and a broader psycho-sociological investigation of it (Both of these, moreover, were descriptive of historically given languages in the way his own inquiry into logical syntax was not). Second, Carnap indicated familiarity with Bühler’s essay and thus the distinction drawn therein between the three functions of language: the “Darstellungsfunktion,” the symbolic function of representation, the “Ausdrucksfunktion,” the symptomatic or expressive function of states of the speaker, and the “Appellfunktion,” the signal function of appeal to or direction of the hearer (Importantly, more than the descriptive function of language was recognized here). What both points suggest is that Carnap already in 1934 had to hand concepts that allowed him to draw something like the distinction that later was drawn explicitly between semantics and pragmatics: unlike the symbolic function which is studied by semantics, both the expressive function and the signal function of language are studied by pragmatics. By 1934, of course, the term “pragmatics” was still absent from Carnap’s discussion.

Four years later, in “Logical Foundations of the Unity of Science,” Carnap (re)asserted that the investigations of the statements of science by his logic of science “abstract … from the persons asserting the statements and from the psychological and sociological conditions of such assertions” (1938, 43; compare 1934/2002, §5). Comprising now both logical syntax and referential semantics, it was thus “important” for the logic of science “to distinguish between the semantical and the psychological use of the word ‘meaning’” (1938, 44). Again neither the subject matter nor the term of pragmatics was mentioned, though the expression “psychological use of the word ‘meaning’” may well be understood to refer to what one year later he designated as the domain of pragmatics.

3 Morris Introduces Pragmatics (1934–1938)

In a series of papers culminating in his contributions to the International Encyclopedia of Unified Science, Charles Morris proposed the term “scientific empiricism” to designate the coming together of the Vienna Circle’s logical positivism and American pragmatism (Morris 1935a, b, c, 1936a, b, 1938a, b).9 In doing so Morris laid out a program for scientific philosophy according to which it could “do more” than what Carnap’s logic of science so admirably achieved, namely “discussions of formal meaning, of the place of rules in determining formal necessity, of the multi-valued logics, of the foundations of mathematics, of physicalism and the unity of science” (1935c/1937, 10). But Morris was also critical of the conception of philosophy as the clarification of meaning, associated with Schlick and Waismann. To remedy the shortcomings, Morris was particularly concerned to broaden the logical positivists’ theory of meaning which he held was not “sufficiently general” (ibid., 12n).10

Morris opposed the “unexamined individualistic hangover” (1936a/1937, 24; cf. 1936b/1937, 69) in the early forms of the verificationism espoused by the Vienna Circle and also held that “neither the formal nor the empirical mode of speech is to be given any wholesale preference over the other” (1936a/1937, 29). Instead, he demanded that the syntactic or formal viewpoint on meaning taken by Carnap’s logic of science be complemented by two additional dimensions. “Unless … the pragmatic and empirical dimensions of meaning [are] considered on a par with the formal dimension, no significant alternative is presented to the conception of philosophy as formal logic.” (1935c/1937, 14) One senses that Morris felt the logical positivists to be unduly phobic about ordinary facts of the matter:

Too much mystery is thrown around the analysis or specification of meaning: the meaning of a term is completely specified when it is known what objects the term designates, what expectations it produces in the person for whom it has a meaning, and what its connections are with other terms in the language of which it is a part. The determination of the first gives the empirical dimension of meaning, the determination of the second gives the pragmatic dimension of meaning, and the determination of the third gives the formal dimension of meaning. Morris (1935c/1937, 13)

At the 1935 Paris congress on the Unity of Science, Morris unveiled the threefold sub-division of “semiotic, the science of signs” as comprising investigations of the “existential meaning” (“the relation of sign to objects”), of the “pragmatic meaning” (“the psychological, biological and sociological aspects of the significatory process”) and of the “formal meaning” (“the syntactical relations to other symbols within the language”) (1936/1937, 65).11 Now apprised of the then recent advances of logical positivism, Morris took care to integrate his proposal into their program more precisely:

To specify the triple set of relations which a sign sustains to objects, persons, and other signs is to exhaust the meaning of that sign … All meanings are potentially intersubjective; theoretically what any sign means can be exhaustively determined by another person. There is therefore nothing peculiarly personal or private about meaning, though there are of course private aspects of the experience of meaning as of the experience of anything else. …The present position therefore reveals itself as in agreement with the first part of the thesis of physicalism, namely, the view that all propositions are in principle verifiable by everyone; it is compatible with the second part of the thesis (the view that all propositions are reducible into propositions of physics) under some interpretations of ‘reduction,’ but it does not depend upon the truth or falsity of this latter position. (Ibid., 66)

This semiotics, Morris claimed, “would at the same time be the novum organum of the special sciences and of the philosophy of scientific empiricism” (ibid., 67).12
In his first contribution to the International Encyclopedia of Unified Science Morris finally introduced the threefold distinction of semiotics under the headings since familiar: syntax, semantics, pragmatics (actually, he still used “syntactics”) and stressed again the planned organum function of semiotics. Philosophy itself was now conceived of as “the widest possible generalization of scientific method” (Morris 1938a, 69). This meant that it was “willing and able to admit into the scope of its considerations everything involved in the scientific enterprise as such, together with the implications of this enterprise for other humans interests” (ibid., 68). The primary focus of its considerations, however, was the language of science which was to be studied under different aspects. Thus he wrote:

Science, as a body of signs with certain specific relations to one another, to objects and to practice, is at once a language, a knowledge of objects, and a type of activity; the interrelated study of the syntactics, semantics, and pragmatics of the language of science in turn constitutes metascience—the science of science. Discussion of the specific signs of science must be carried on in terms of some theory of signs, and so semiotic, as the science of signs, occupies an important place in the program—indeed, the study of the actual language of science is an instance of applied semiotic. (Ibid., 70)

Earlier in the same monograph Carnap had distinguished the study of the language of science, albeit in abstraction from its speakers and the psychological and sociological conditions of their assertions, from the study of science as an activity in order to distinguish the domain of the logic of science (1938, 42–43). Here Morris advocated the threefold division of semiotics as an organizing principle for the encyclopedia project itself and thereby for the pursuit of philosophy in unified science. “The theory of signs gives the general background for the consideration of the language of science. …In this way, and on a comprehensive scale, science is made an object of scientific investigation; metascience appears both as a tool for, and as an element within unified science.” (1938a, 74) So understood philosophy of science of course was part of science itself.
Morris stressed that “any adequate account of the science must take account” of the “psychological, methodological and sociological aspects of scientific practice” (ibid., 72). This meant, of course, that the pragmatical component of semiotics was no mere afterthought. What is of particular interest in this connection is how the pragmatical component was characterized.

A third concern of metascience [besides syntax and semantics, TU] arises from the fact that the signs which constitute the language of science are parts and products of the activity of scientists. The study of the relation of signs to scientists may be called, in the spirit of pragmatism, the pragmatical investigation of the language of science. Here belong the problems as to how the scientist operates, the connection of science as a social institution with other social institutions, and the relation of scientific activity to other activities. (Ibid., 70)

Note what Morris here included under pragmatics: experimentation and hypothesis testing (“‘confirmation’ is a concept which contains irreducible pragmatical features” (ibid., 72)), the social organization of science and the role of science in social and cultural life generally.13 One might well ask: was there anything that could be regarded in relation to science that was not included?14 (And one may wonder whether there was any disciplinary unity to these “pragmatical investigations”).15

In the Introduction to his monograph for the International Encyclopedia, Foundations of the Theory of Signs, Morris claimed with explicit reference to his earlier contribution: “it is possible to include without remainder the study of science under the study of the language of science” (1938b, 2). While not further enlarging on this wide understanding of pragmatics he had previously outlined, Morris now also recognized a more narrowly delimited understanding of pragmatics (even though he did not expressly distinguish it) when he introduced the term “pragmatical rule.” “The statement of the conditions under which terms are used, in so far as these cannot be formulated in terms of syntactical or semantical rules, constitutes the pragmatical rules for the terms in question.” (Ibid., 35) This meant to cover rules for expressive language use (“interjections”), what nowadays we call speech acts (“commands,” greetings) and “various rhetorical and poetical devices” (ibid.). I shall therefore speak of a “narrow” and “wide” understanding of pragmatics, the former concerning pragmatical rules for linguistic expressions, the latter concerning anything that involves speakers of scientific languages. While narrow pragmatics is part of linguistics, wide pragmatics embraces anything from methodology to the sociology of science (and beyond).

Whatever we make of the plurality of subjects collected under the heading of “wide pragmatics,” it is clear that Morris’s conception of semiotics as metascience is broadly related to the idea of philosophy of science as scientific metatheory. It is particularly notable that his wide conception of pragmatics fully encompasses the empirical sciences of science.16 To be sure, semiotics as metascience is by no means the same thing as the bipartite metatheory. Morris’s conception was ultimately focused upon the language of science, whereas the bipartite conception need not make a comparable claim of the subservience to language of the phenomena covered by the pragmatics of science (Does the economics of science merely deal with an aspect of the language of science?) Moreover, Morris’s wide pragmatics reached still beyond the empirical sciences of science into moral-political discourse (Unlike in his Paris lecture, however, Morris did not here enlarge on the axiological aspect that for him lay behind what he called the “humanistic implications of semiotic” 1938b, 57–58).17 Even so, there is a very significant overlap between semiotics as metascience and the bipartite conception of scientific metatheory.

4 Carnap Adopts the Semiotic Paradigm (1939)

In Foundations of Logic and Mathematics Carnap officially adopted the division of “pragmatics, semantics, and syntactics” from Morris “in agreement” with “its chief features” (Carnap 1939a, 5).18 (Carnap did not specify points of disagreement).19 I shall first document his use of the relevant concepts, then ask whether it was commensurate with Morris’s and then turn to Carnap’s understanding of pragmatics specifically.

For Carnap, pragmatics dealt with “the action, state, and environment of a man who speaks or hears” a certain linguistic expression (ibid., 4). As his examples make clear, pragmatics studies “the mode of use” of expressions and “both the cause and the effect” of utterances, thus also “the preferences of different social groups, age groups or geographical groups in the choice of expressions” and “the role of language in various social relations.” He summarized: “pragmatics is an empirical discipline dealing with a special kind of human behavior and making use of the results of the different branches of science (principally social science, but also physics, biology and psychology).” (Ibid., 6)

Semantics, by contrast, restricts its “attention to a special aspect of the facts concerning the language,” namely “the relations between the expressions … and their designata,” and seeks to “lay down a system of rules establishing those relations” (ibid.). Now importantly, these “semantical rules … are not unambiguously determined by the facts” (ibid.). It follows that “the rules which we lay down are not rules of the factually given language B” but that they “constitute a language system corresponding to B … the semantical system B–S” (ibid., 7). The following contrast thus obtained:

The language B belongs to the world of facts; it has many properties, some of which we have found, while others are unknown to us. The language system B-S, on the other hand, is something constructed by us; it has all and only those properties which we establish by rules. Nevertheless, we construct B-S not arbitrarily but with regard to the facts about B. …The previously mentioned pragmatical facts are the basis … of some of the rules to be given… (Ibid.)

This gives a clear sense in which the investigations of the logic of science “abstract” from the language used by scientists in their work (as he had put it in 1938, 43). It also suggests that the logic of science has no interest in pursuing psychological or sociological aspects of the use of the language of science but considers it only as far as syntactic constraints and designation relations are concerned.20 But for the addition of semantics into the logic of science and the recognition of pragmatics outside of it, Carnap’s position remained that taken already in (1934) (see Sect. 1 above).
It does not seem to be the case, however, that Carnap’s understanding of the threefold semiotics fully agreed with Morris’s. There is the question which Carnap did not address until Introduction to Semantics, of just how “interrelated” (in Morris’s words 1938a, 74) the investigations of the different branches of semiotics were. Consider Carnap’s distinction between “pure” and “descriptive” syntax (1934/2002, §2) and his parallel distinction between “pure” and “descriptive” semantics (1942, §5), and what relations pragmatics stood in to each of them (Since analogous characterizations apply to syntax, we need consider only how Carnap described the distinction as it applies to semantics).

By descriptive semantics we mean the description and analysis of the semantical features either of some historically given language, e.g., French, or of all historically given languages in general. The first would be special descriptive semantics; the second, general descriptive semantics. Thus, descriptive semantics describes facts; it is an empirical science. On the other hand, we may set up a system of semantical rules, whether in close connection with a historically given language or freely invented; we call this a semantical system. The construction and analysis of semantical systems is called pure semantics. The rules of a semantical system S constitute … nothing else than a definition of certain semantical concepts with respect to S, e.g., ‘designation in S’ or ‘true in S.’ Pure semantics consists of definitions of this kind and their consequences; therefore, in contradistinction to descriptive semantics, it is entirely analytic and without factual content. (1942, 11–12, orig. emphasis)

Descriptive semantics is the task of empirical linguistics and aims to provide true descriptions of how expressions of a historical language relate to the world; pure semantics is undertaken by logicians and establishes a purely logical system of designation relations. (In later years Carnap would call this a semantic framework.)
Now importantly, Carnap held pragmatics to stand in different relations to semantics and syntax, depending on whether we think of their descriptive or pure varieties.

Linguistics, in the widest sense, is that branch of science which contains all empirical investigations concerning languages. It … consists of pragmatics, descriptive semantics and descriptive syntax. But these three parts are not on the same level; pragmatics is the basis for all of linguistics. However, … [o]nce the semantical and syntactical features of a language have been found by way of pragmatics, we may turn our attention away from the users and restrict it to those semantical and syntactical features. … descriptive semantics and syntax are, strictly speaking, parts of pragmatics. (Ibid., 13, orig. emphasis)

This leaves open whether descriptive semantics itself was envisaged as a formalized theory or not. But importantly, Carnap went on: “With respect to pure semantics and syntax the situation is different. These fields are independent of pragmatics.” (Ibid.) What follows is that, at the time, Carnap did not envisage a “pure pragmatics”: all pragmatics was thought of by him as descriptive and empirical. For Carnap, pragmatics had no bearing on the construction and investigation of constructed languages, on semantical and syntactical systems. Instead, pragmatics was linked exclusively to historically given languages.

By contrast, Morris distinguished between “pure semiotic, with the component branches of pure syntactics, pure semantics, and pure pragmatics” on the one hand and “descriptive semiotic” with the component branches of descriptive syntactics, descriptive semantics, and descriptive pragmatics (1938b, 9). Pure semiotic was characterized as a “deductive system, with undefined terms and primitive sentences which allow the deduction of other sentences as theorems” (which was, he added, “the form of presentation to which science strives”). Morris noted that even “the preliminary systematization in the component fields has hardly begun”; even so, pure semiotic remained an important “goal” (ibid.).21 Pure semiotics “elaborated in systematic form the metalanguage in terms of which all sign situations would be discussed. The application of this language to concrete instances of signs might be called descriptive semiotic (or syntactics, semantics or pragmatics as the case might be).” (Ibid., orig. emphasis) For Morris, pure pragmatics arose “out of the attempt to develop a language in which to talk about the pragmatical dimension of semiosis” (ibid., 30).

How significant is this difference? What did the independence of pure syntax and semantics from historically given languages amount to? For Carnap, the languages that pure syntax and semantics developed did not have to conform to the constraints of humanly possible languages.22 And as later became clear in his debate with Quine over the tenability of the analytic/synthetic distinction, the categories used by Carnap in the development of semantical systems did not necessarily have counterparts in the purely descriptive analysis of natural languages.23 Whereas it looks as if pure semiotics for Morris was still empirical, albeit at a high level of abstraction from concrete examples, Carnap’s pure syntax and semantics were a priori investigations. As for pure pragmatics, at the time Carnap recognized no such thing.24

Whether Carnap or Morris were aware of this difference right away is not entirely clear (Morris commented on it only a few years later and urged the introduction of pure pragmatics).25 It did not make much of a practical difference for their work on unified science and in the 1950s a broad convergence of their views obtained—with the important difference that showed up in Carnap’s and Neurath’s neglect of Morris’s interest in axiology.26 In retrospect, it would appear that both Morris and Carnap got hold of important considerations: Morris stressed that all human thinking about languages cannot ultimately transcend its grounding in the experience of human language, while Carnap stressed that how we conceptualize language-related phenomena in syntax and semantics is not ultimately determined by our experience as “sign-using animals” (Morris 1938b, 1).

5 Carnap’s Employment of the Wide Conception of Pragmatics (1939–1950)

Turning to Carnap’s understanding of descriptive pragmatics, it is notable that it dealt with a great variety of language-related behaviors.

If in an investigation explicit reference is made to the speaker, or, to put it in more general terms, to the user of the language, then we assign it to the field of pragmatics. … Examples of pragmatical investigations are: a physiological analysis of the processes in the speaking organs and in the nervous system connected with speaking activities; a psychological analysis of the relations between speaking behavior and other behavior; a psychological study of the different connotations of one and the same word for different individuals; ethnological and sociological studies of the speaking habits and their differences in different tribes, different age groups, social strata; a study of the procedures applied by scientists in recording the results of experiments, etc. (1942, 10)

Of these examples, only some fall happily under general linguistics and psycho- or socio-linguistics. The first set of examples rather pertains to the physiology and neurology of speech behavior and the last set points in a different direction altogether.27
Note therefore especially the very wide remit that Carnap gave to pragmatics.

It has turned out to be very fruitful to look at the problems of theoretical philosophy from the point of view of semiotic, i.e. to try to understand them as problems which have to do with signs and language in one way or another. Among problems of this kind we may first distinguish between those problems—or components in complex problems—which are of a factual, empirical, rather than logical nature. They occur especially in the theory of knowledge and the philosophy of science. If construed as problems of semiotic, they belong to pragmatics. They have to do, for instance, with the activities of perception, observation, comparison, registration, confirmation, etc., as far as these activities lead to or refer to knowledge formulated in language. (Ibid., 245)

Here pragmatics comprises the study of all activities that “lead to or refer to knowledge as formulated in language.” Clearly much that does not have to do with language use in any direct way falls under the heading of pragmatics in this very wide sense. It fits, of course, that Carnap had previously assigned methodology of science to the empirical part of the metatheory of science: methodology seeks to regulate activities intended to lead to knowledge.28 Thus when he corrected “Testability and Meaning” for a 1950 reprinting, Carnap also added a note updating its terminology: “descriptive concepts” discussed in the chapter “Empirical analysis of confirmation and testing”—namely: observable, realizable, confirmable, testable— “belong to pragmatics” (1936–1937/1953, 50).29

It is not hard to see, however, that a certain ambiguity is now introduced into Carnap’s concept of pragmatics. When he claimed that “pragmatics is the basis for all of linguistics,” it was not this wide sense of pragmatics that Carnap could have had in mind, but only the narrow sense centered on the investigation of historically given linguistic behavior. The sense in which pragmatics encompasses also “empirical” problems of epistemology and philosophy of science having to do with “the activities of perception, observation, comparison, registration, confirmation, etc., as far as these activities lead to or refer to knowledge formulated in language” is not the basis for all linguistics as an empirical science at all—at best only for the empirical study of the language of science. We may distinguish these two senses in Carnap by calling the wide one “methodological pragmatics” and the narrow one “linguistic pragmatics.”

The question arises how far Carnap was following Morris’s lead. As noted, he clearly did not follow the latter’s muted reference to axiology and it might even seem that his taste for the empirical sciences of science was decidedly limited, namely to more or less methodological issues. What must be remembered here, however, is that Carnap was addressing concerns arising in his own work as a logician of science: those included, as we just saw, “defining concepts of an empirical analysis of confirmation” (ibid.). So by adopting pragmatics in the wide methodological sense, Carnap by no means limited his recognition of the empirical sciences of science as complementary to his logic of science so as to exclude, say sociology or history of science.

6 Carnap’s Employment of the Narrow Conception of Pragmatics (1955–1963)

Carnap’s use of “pragmatics” did not remain constant, however. When he used it in the mid-1950s he used it typically in a narrow linguistic sense. Thus he wrote:

The analysis of meaning of expressions occurs in two fundamentally different forms. The first belongs to pragmatics, that is, the empirical investigation of historically given natural languages. This kind of analysis has long been carried out by linguists and philosophers, especially analytic philosophers. The second form was developed only recently in the field of symbolic logic; this form belongs to semantics (here understood as in the sense of pure semantics, while descriptive semantics may be regarded as pragmatics), that is, the study of constructed language systems given by their rules. (1955a/1956, 233, orig. emphasis)

Previously, Carnap’s main efforts had been addressed to pure semantics, for his concern did not lie with “understanding the behavior of individuals and … the character of and development of whole cultures” (ibid., 234). But now he added that

also for the logician a study of pragmatics may be useful. If he wishes to find out an efficient form for a language system to be used, say, in a branch of empirical science, he might find fruitful suggestions by the study of the natural development of the language of scientists and even of the everyday language. Many of the concepts used today in pure semantics were indeed suggested by corresponding pragmatical concepts which had been used for natural languages by philosophers or linguists, though usually without exact definitions. Those semantical concepts were, in a sense, intended as explicata for the corresponding pragmatical concepts. (ibid., 234)

Carnap then set out to “clarify the nature of the pragmatical concept of intension in natural languages” in order to give a “practical vindication for the semantical intension concepts” (ibid., 235). To this end, he outlined what he called the “conceptual framework of theoretical pragmatics” further in a follow-up paper by giving formal explications of an expanded notion of the pragmatical concept of intension, making use now in addition of the also formally explicated concepts of belief and holding true, and by adding formal explications of the concepts of assertion and utterance (all but the latter he regarded as theoretical terms) (1955b/1956, 248–249).

Three things are notable about this “pragmatical turn” by Carnap in 1955. First that Carnap here continued a debate with Tarski and Quine that had started when Carnap was a visiting professor at Harvard in 1940–1941 and that turned on the viability of drawing a sharp distinction between logical and factual truth and the use of intensional concepts in logic and semantics.30 In his 1955 papers Carnap in these advised the study of pragmatics in order to justify the admissibility of intensional languages: some of the intensional concepts of his constructed languages were meant, after all, to serve as explicans, as clarifying reconstructions, of contested concepts of ordinary language use. This may appear to signal a shift in Carnap’s conception of the relation between what had been called “pure” and “descriptive” semiotics. That it was “clear” to him that “it should be possible” to provide “an empirical, behavioristic criterion in pragmatics with respect to natural languages” of “the concept of intension or meaning” (1963, 67) does suggest that Carnap, perhaps under continued pressure from Quine, had come to regard the relation between pure and descriptive semiotic inquiries as closer than before. It does not mean, however, that Carnap conceded that each and every concept used in pure semantics, say, needed to be shown to be capable of a clear and unambiguous application to natural language: the use of some notions—“analytic statement,” for example—were justifiable on grounds of their practical value to the project of explication itself.31

The second point to note concerns what Carnap dealt with in the present context under the heading “pragmatics”: the meaning of natural language expressions that was intended by their users. When he responded to Morris’s contribution to the Schilpp volume, Carnap again used “pragmatics” in this narrow sense: while describing it as “the theory of the use of language by human beings” he discussed under its heading his efforts to indicate “some concepts (belief, intension, assertion) which, together with related concepts, might serve as a basis for theoretical pragmatics” (1963, 861). As we shall see, this raises the question of what became of Carnap’s understanding of the wide methodological notion of pragmatics.

The third point to note is that Carnap’s talk of “theoretical pragmatics” signals the earlier mentioned belated rapprochement with Morris’s conception of pure semiotics. When Morris, in his contribution to the Schilpp volume, repeated his demand for pure pragmatics as “coordinate with pure semantics and pure syntax” and “supplying a framework for the intensive studies in descriptive pragmatics now under way in psychology, the social sciences and the philosophy of science” (1963, 88, 90), Carnap replied that his theoretical pragmatics amounted to the same thing: “I agree with Morris that there is an urgent need to develop pure pragmatics, which would supply a framework for descriptive pragmatics.” Carnap reasoned that what he now called (interchangeably) “theoretical” or “pure” pragmatics studied “the logical relations” between the concepts forming its basis by means of meaning postulates (1963, 861). In consequence, for both Carnap and Morris the different branches of descriptive semiotics were but applications of the conceptual frameworks developed in the branches of pure semiotics. Note, however, that the theoretical pragmatics now admitted by Carnap only provided a pure counterpart for a very small part of the originally very wide domain of descriptive pragmatics and concentrated on a narrow linguistic version that was not even yet concerned with speech acts and the like but focused only on descriptive semantics, the meaning of declarative expressions of historically given natural languages.32

7 Carnap’s Non-Adversion to Wide Pragmatics in Reference to Framework Questions

Carnap noted that the defense of the semantical intension concept against his critics may also take other forms, namely that of demonstrating their “usefulness through [their] application in the further development of language systems” (1955a/1956, 235). In “Empiricism, Semantics and Ontology” he had argued that the choice of linguistic forms was to be decided by whether their use was “expedient and fruitful for the purposes for which semantical analyses were made, viz. the analysis, interpretation, clarification, or construction of languages of communication, especially languages of science” (1950b/1956, 221). Now what Carnap invoked there—assessments of linguistic forms for their “efficiency as instruments” (ibid.)—are results of investigations that fall under the heading of pragmatics in the wider methodological sense. Yet Carnap did not call these considerations “pragmatic” in this paper but instead spoke of “practical questions” and “practical decisions” (ibid., 207, 214). In Meaning and Necessity Carnap had likewise spoken only of “practical differences concerning the choice of a method for semantical analysis” (1947/1956, 204). And when Carnap returned to state that “many problems concerning conceptual frameworks seem to me to belong to the most important problems of philosophy” and added that he was thinking “both of theoretical investigations and practical deliberations and decisions with respect to an acceptance or a change of frameworks” (1963, 862), he again refrained from using the term “pragmatic” or “pragmatical” (cf. ibid., 66, 872–873).33

Thus a rather curious situation has arisen. When discussing questions of framework change generally, Carnap did not refer to the wide methodological sense of pragmatics as being able to help in answering them but, by contrast, when he did invoke pragmatics to help decide framework questions of a particular kind, it was the narrow linguistic version of pragmatics that he invoked. The question arises whether Carnap had begun to drop the wider methodological sense around 1950.34 As the passages from 1955 indicate, by then the term “pragmatical” was once more tied closely to the investigation of aspects of natural languages. One may even wonder whether it was not now tied to the investigation of the meaning of natural languages, but it probably would be wrong to say that now pragmatics became what previously he had called “descriptive semantics.” If Carnap’s advice to regard the latter “as pragmatics” is read as advice to regard it as part of pragmatics, then the continuity of his usage is preserved. Yet even with this clarification Carnap’s mid-1950s use of “pragmatics” stands in apparent contrast to his use of the term in 1942. Now he appealed only to the narrow linguistic sense of the term in pursuit of his aim “to outline a behavioristic, operational procedure” for determining natural language meanings (1955a/1956, 235).

What are we to make of this? Several possibilities need to be considered. (1) Carnap did not drop the wide methodological sense of “pragmatics” but simply had no occasion to mention it. This is not very plausible—unless more were to be told: as we saw, the issue of framework change concerning the admissibility of terms for abstract objects provides a fitting opportunity to bring the wide concept of pragmatics (the empirical sciences of science) to bear on it. (2) Carnap did drop the wide methodological sense of “pragmatics” because he began to find it objectionable for some reason or other. Again we would need to hear more. (3) Carnap had a yet different reason that, as a matter of fact, leaves it open whether he did or did not reject the wide notion of pragmatics. I favor a version of (3). To see how it works, let’s allow Carnap to recall the problem situation with his customary clarity:

If a philosopher asks a question like ‘are there natural numbers?’, he means it as a question so-to-speak outside the given language, raised for the purpose of examining the admissibility of such a language. Therefore I called philosophical questions of existence of this kind external questions. But then I pointed out that for these questions no interpretation as theoretical questions has been given by the philosophers. I proposed to the philosophers who discuss such questions that they interpret them as practical questions, i.e. as questions about the decision whether or not to accept a language containing expressions for the particular kind of entities. Various reasons may influence the decision about the acceptance or non-acceptance of the framework for such expressions. My main point is the rejection of the customary view that the introduction of a linguistic framework is legitimate only if the affirmative answer to the external question of existence (e.g., ‘there are natural numbers’) can be shown to be true. In my view, the introduction of the framework is legitimate in any case. Whether or not this introduction is advisable for certain purposes is a practical question of language engineering, to be decided on the basis of convenience, fruitfulness, simplicity and the like. (1963, 66)

Crucially, “external” questions are “not theoretical.” Yet pragmatics is a theory. It follows that discourse about external questions is not pragmatical discourse in sense of pragmatics. If this is true then Carnap’s choice of terminology has nothing to do with whatever he may have thought at the time about the wide sense of pragmatics.

But what makes “practical questions” of language choice “not theoretical”? The answer is that, for Carnap, they are normative questions asking for a volitional decision. They are not factual questions to be answered by reference to empirical or logical-mathematical evidence.35 Practical questions involve (i) determining the instrumental purpose of language to be chosen and (ii) determining a relevance ranking for various measures of efficiency (Carnap mentioned some, as we saw). Even if (i) were fixed one way or another, (ii) would still be open. Now pragmatics, be it the narrow or the wide version, has nothing to say to (i). But also with regard to (ii) it can at best inform the decision maker about specific efficiencies but cannot prescribe their ranking (which depends on various specifics of the case at hand).

So Carnap was only particularly conscientious when he refrained from calling external discourse pragmatical. Of course, the empirical sciences of science were consulted on those occasions, but this still did not mean that the issues were straightforwardly pragmatical in the sense in which he typically understood the term pragmatics, namely as a descriptive discipline.36 (Pure pragmatics also was unable to help here). A theorist like Morris who did not draw a sharp distinction between factual and normative discourse would not share Carnap’s qualms, but Carnap himself observed the further distinction that this one commited him to.

8 Frank’s Late Reminder to Carnap of the Wide Conception of Pragmatics

The most important question in the context of the present essay, however, is what all this means for Carnap’s views, first, of pragmatics in the wide sense and, second, of the conception of philosophy as bipartite metatheory of science. Carnap’s relative silence about the wide conception of pragmatics after 1950 may indicate that he began to find it implausible to press the empirical sciences of science into the semiotic schema (echoing an earlier complaint of premature oversystematization by Neurath)—but it need not.37 And even if he did, that need not have a bearing on whether Carnap continued to be open to the bipartite metatheory conception. This would have to be decided on independent grounds.

Again we must note that Carnap himself never worked on pragmatics in the wide sense of wider than methodology—this did not change with his turn to discussions of pragmatics in the narrow sense. Yet Carnap did give a very significant hint about how he continued to think about the former which can be found in his response to Frank’s contribution to the Schilpp volume, “The Pragmatic Components in Carnap’s ‘Elimination of Metaphysics’.” There Frank discussed (and reprinted in translation) a Marxist-Leninist critique of Carnap’s famous piece by V. Brushlinsky from the official philosophical journal of Soviet Russia, Pod Znamemem Marksisma (Under the Banner of Marxism) of 1932. Commenting upon it, Frank pointed out that “not only Soviet philosophers but also Western philosophers who follow the Pragmatism of John Dewey” would hold that “one cannot overcome metaphysics fully by proving that it is meaningless but by understanding its meaning fully and exposing it. The word ‘fully’ means: ‘Including the pragmatic component’.” Carnap’s critique remained ineffective “because he restricted his universe of discourse to the logical component of language.” (Frank 1963, 164)38

What’s interesting here is what Frank understood by pragmatics and how Carnap reacted (As noted, already in 1932 Frank had suggested adding a sociological dimension to Carnap’s then emerging logic of science: Frank was one of the instigators, together with Neurath, of the bipartite metatheory conception of philosophy of science).39 Frank noted that and how Carnap admitted pragmatics in Foundations and stated:

If we consider the pragmatic component of scientific language, we have to speak about the influence of his environment upon the builder of scientific theories, in other words, we have to investigate the reasons which have induced the scientist to introduce such formulations of principles in which [contested] words like ‘god’ or ‘reality’ enter. (Carnap 1963, 164)

Note that Frank appealed to “pragmatic component of scientific language of science” to make his point about what in his Philosophy of Science: The Link Between Science and Philosophy he called the “pragmatics of science” (1957/2004, 360). Frank was quite happy to let his pragmatics of science sail under the semiotic flag. When we turn to this book we find that Frank endorsed “the triadic scheme defined and elaborated in the International Encyclopedia of Unified Science by Rudolf Carnap and Charles Morris” (ibid., 349).

Carnap reacted somewhat testily, though not without chagrined sympathy.

As I have pointed out elsewhere [in the Schilpp volume], the Vienna Circle, essentially because of Otto Neurath, did recognize the importance of a sociological analysis of the roots of philosophical movements. But unfortunately, a division of labor is necessary, and therefore I am compelled to leave the detailed work in this direction to philosophically interested sociologists and sociologically trained philosophers. (Carnap 1963, 868)

Agreeing with the “importance of a sociological analysis of metaphysics,” Carnap pointed to “detailed work” done by pragmatists “in connection with investigations concerning the pragmatic component in language.” But he conceded that “the pragmatic component has so far not been sufficiently investigated” by logical empiricists, even though “its importance has been acknowledged theoretically” by him and his colleagues (ibid.).

This exchange is highly significant for our present concerns. It shows not only that Frank was happy to subscribe to the semiotic schema and comprehended the empirical sciences of science under the heading of pragmatics of the language of science (in its wide sense), but also that Carnap joined him in this—both in how such investigations were to be categorized and in the assessment of their importance. Carnap’s reaction thus strongly suggests that when he turned to work on theoretical pragmatics in the narrow sense he had not abandoned its wide sense either, for we can now recognize its continued endorsement in the very paper which introduced his pragmatic turn: “Nobody doubts that the pragmatical investigation of natural languages is of greatest importance for an understanding both of the behavior of individuals and of the character and development of whole cultures.” (1955a/1956, 234, emphasis added)

9 Conclusion

Does this late agreement between Frank and Carnap prove the thesis that their and Neurath’s efforts can be understood as providing the blueprint for a bipartite metatheory with a formal-logical part supporting and being supported by a naturalistic-empirical part? I believe that together with the history of Carnap’s views on pragmatics it does provide significant support. Far from failing across the board, Morris in fact succeeded in furnishing a conceptual vehicle for developing the bipartite metatheory conception of philosophy of science within unified science that was adopted by Carnap and Frank. That within the framework of the encyclopedia project this blueprint for the bipartite metatheory was not developed as far as Neurath and Frank had hoped and that most contributions only concerned the logic of science in one way or another, does not contradict this basic fact—nor that there was a serious falling-out between Neurath and Carnap or that Morris’s own conception of the unification of logical empiricism and pragmatism failed.40

To be sure, a clear affirmation of the bipartite metatheory conception of philosophy of science under this moniker was never issued either by Carnap or Neurath or Frank. The explicatory nature of the bipartite metatheory thesis is clearest in the case of Carnap—both Neurath and Frank affirmed the substance of the thesis in related terms at various times. But Neurath too had terminological difficulties, in his case with Morris. Therein lies the key, I believe, to why the bipartite metatheory conception did not receive the kind of promotional fanfare accompanying other programmatic schemes of his. For Neurath did not only object to the “Kantian tendency” for tripartite conceptual architecture, but also had “the most far-reaching doubts” about the expression “science of science.” The reason was the same in both cases: their use holds “the danger,” he once wrote to Morris, “that everything that one meets with must be brought under this schema and that certain possibly fruitful matters get neglected precisely because they do not fit into it.”41 Neurath feared that under the guise of Morris’s terminology a new “system”-thinking, a new metaphysics could develop. Thus he preferred to promote what here has been called the bipartite metatheory conception under the old heading of the “program of the unified science.”

The very term “science of science” figures prominently in the present explication of his model of unified science, however. Is this a distortion of the historical material? A related question arises in Carnap’s case. Did he not reserve the name “philosophy” exclusively for the logic of science and did he not treat pragmatics (especially the wide variety of it) as a placeholder subsuming under it all the metatheoretic investigations he was not interested in pursuing himself? With regard to the latter the answer is “yes,” but this does not detract from his de facto subscription to the bipartite metatheory conception.42 Likewise with Neurath: his terminological qualms need not prevent a redescription of the substance of his ideas. His rejection of the term “philosophy” was as much a function of the long-term dialectical context in which he perceived himself as Carnap’s retention of the same term for the logic of science alone. Why then my redescription of their views under the current heading? While it underscores what Richardson has termed the “methodological naturalism” that characterizes “scientific philosophy”—“the insistence that philosophy can employ no methods other than those of science” (2003, 12)—this redescription also brings out, for instance, that the logic of science was not for one moment intended by its inventor to operate in isolation, divorced from first- and second-order factual research. This conception of philosophy was not shared across the whole Vienna Circle—nor across orthodox post-World War Two logical empiricism. Frank’s contribution to Schilpp’s Carnap-Festschrift may be understood as a reminder, issued by one whose work on the pragmatics of science in the 1950s was increasingly ignored in mainstream philosophy of science, that the bipartite metatheory conception was still very much “on the table,” as Carnap readily acknowledged. That it was so unreservedly forgotten after both of their deaths by the philosophical profession is yet another matter.43


See Mormann (2010) and Uebel (2010).


This was how far Uebel (2001) pressed the matter in light of some well-known differences between Carnap and Neurath; in Uebel (2007, 435) the analysis was extended to the attribution of a “broadly shared program for a bipartite metatheory serving as a joint replacement for [traditional] philosophy”.


See, e.g., Carnap et al. (1929/1973, 306). Hahn belonged to the left wing but is not mentioned further here since due to his early death there are no published contributions of his to the debates under investigation.


“The failure of constructing a comprehensive ‘scientific empiricism’ in Morris’s sense should be taken into account when we seek to assess the chances of Uebel’s ‘bipartite metatheory’, even if the parallelism between Uebel’s and Morris’s proposals is limited” (Mormann 2010, 38). Mormann now speaks of Morris’s “Paris programme for scientific philosophy”; for a sympathetic portrayal of it, see his (forthcoming).


See Carnap (1942, §39) for a discussion of the changes this wrought for the theses and terminology offered in his (1934/2002).


By “behavioristics” Neurath meant to designate theories that accounted in physicalistically acceptable ways for human behavior and to indicate a critical distance from Watson’s behaviorism (see 1933, §5). His own examples over the years indicate that a non-reductive physicalism was intended (see, e.g., 1936/1983, 162).


This is not to say that Carnap did not think that some aspects of meaning could be treated syntactically. For a brief while in 1932, Carnap used “logical syntax” and “semantics” as cognates (see 1932/1959, 177) and even fancied “Semantik” as the title for his (1934/2002). And since Coffa (1977) it is known that in Logical Syntax Carnap used supposedly syntactic means to accomplish tasks we now regard as semantic.


From Bühler’s essay, but not from Carnap, one can learn that “Semasiologie” stems from Heinrich Gomperz. For a description of Gomperz’s “Semasiologie” in his (1908) see Ogden and Richards (1923/1936, 274–277). See Carnap (1942, 238–239) for an updating of his own 1934 terminological remarks on “semantics”.


Thus Morris declared that “formalism, empiricism and pragmatism are complementary phases of the scientific temper. I propose to recognize this fact by designating the implied philosophical attitude as scientific empiricism” (1935c/1937, 21, emphases dropped).


Morris also recommended adding “empirical axiology” and “empirical cosmology” (1935c/1937, 14 and 16). I do not discuss these here as they represent that aspect of Morris’s proposal that was not taken up by either Carnap or Neurath or Frank.


While “semiotic” was used from Locke through Peirce to Morris and his collaborators, the current usage—which I shall follow outside of quotations—favors “semiotics”; see Sebeok (1971).


Elsewhere Morris had written: “Philosophy as organon becomes general semiotic.” (1937, 5).


Morris’s expression “the study of the relation of signs to scientists” resonates both with Carnap’s talk of investigating science as an activity (see above) and Neurath’s “behavioristics of scholars” which comprises (according to examples given at 1936/1983, 160) the history and the psychology of science but contrasts with logic of science.


Note that Morris’s later characterization of pragmatics as “that portion of semiotic which deals with the origin, uses, and effects of signs within the behavior in which they occur” (1946/1971, 302) would still allow for similar expansion, depending on what is specified as the relevant behavior in which the signs occur.


Neurath can be seen to be motivated by a related worry when he replied to Ake Petzäll’s reply to his delineation of the bipartite metatheory conception (in response to Petzäll’s earlier criticism of Vienna Circle philosophy): “The two disciplines, logic of science and behavioristics of scholars which Petzäll mentions in connection with my article are therefore to be seen only as examples; here could be other examples from other disciplines.” (1936/1983, 168) What Neurath meant to stress was that a great variety of disciplines make up the empirical science of science. On related concerns expressed by Neurath in the correspondence with Morris, see Dahms (1997, 54–63) and Reisch (2003a, 206–207) and Sect. 8 below.


Previous discussions of Morris’s semiotics tend to neglect this “wide” aspect of his pragmatics.


It is the axiological aspect of Morris’s program that Mormann stresses as distinctive in his (forthcoming).


Martin wrote that in these “splendid paragraphs we encounter a characterization of the modern trivium of syntax, semantics, and pragmatics arising from the sea almost fully blown like Venus herself” (1987, 303). This clearly overlooks the influence of Morris even though, as we saw, Carnap was indeed already favorably disposed towards it.


Carnap’s reference was Morris (1938a) not (1938b). Three years later he noted that Morris defined pragmatics as “the field dealing with the relations between speakers (or certain processes in them) and expressions” but added that “in practice there does not seem to be a sharp line between investigations of this kind and those which refer also to designata” (1942, 9–10).


Carnap gave a sustained defense of the method of abstraction underlying the logic of science in (1950a, 215–219).


Later Morris added, with a silent reference to Carnap, that “semantics has not yet attained a clarity and systematization comparable to that obtained by certain portions of syntactics” (1938b, 22); Carnap (1942) set out to remedy this, of course.


Though he did not engage in doing so after 1932, Carnap never renounced the possibility that the logic of science could construct private protocol languages, for instance.


“As I now understand Quine, I would agree with his basic idea, namely, that a pragmatical concept, based upon an empirical criterion, might serve as an explicandum for a purely semantical reconstruction, and that this procedure may sometimes, and perhaps also in the present case, be a useful way of specifying the explicandum. On the other hand, I would not think that it is necessary in general to provide a pragmatical concept in order to justify the introduction of a concept of pure semantics.” (Carnap 1963, 919).


In response to Morris, Carnap admitted in later years: “In the case of pragmatics, I was mainly thinking of empirical investigations” (Carnap 1963, 861). In fact, it is difficult to see how there could be a pure pragmatics of the kind envisaged by Morris for the wide understanding of pragmatics that Carnap appears to have taken over from Morris: Morris’s own remarks about pure pragmatics apply to the narrow understanding of it as concerned with linguistic rules. When Carnap in later years conceded a role for pure pragmatics, it was likewise limited to systematize the employment of meaning-related and psycho-linguistic notions.


The reason Morris gave was that “in pragmatics too we can distinguish between formative and lexicative ascriptors, and so generalize the distinction between pure and descriptive semiotic” (1946/1971, 329). The relevant distinction was one between the “formative discourse of semiotic and its designative discourse, that is, between semiotic as logic and semiotic as scientific discourse” (ibid., 303), such that the former discourse spelled out what the semiotic relations were that were taken for granted in the latter.


What Carnap and Morris must have been aware of were their very different views on metaethics, but these were not discussed by them in publications until Morris’s contribution to the Schilpp volume (Dewey had raised the matter without naming names in his (1939, 6–19).) This point has been highlighted recently by Mormann (2007) and Richardson (2007). That in respect of their metaethics quite generally not much convergence between logical empiricists and pragmatists can be discerned is true but irrelevant to the issue at hand, namely, whether a shared adherence to the blueprint for the conception of a bipartite metatheory can be discerned in Carnap (and Neurath and Frank) and Morris.


Since Richardson (2003) discussed the far-reaching agreements between Carnap and Morris (albeit from the perspective of the reception of Carnap’s views by Morris, not of Morris’s by Carnap), it should be noted that the implicit distinction or unnoticed tension between the narrow and the wide conception of pragmatics here focused upon does not coincide with the difference that Richardson rightly diagnosed between, on the one hand, Quine’s understanding of what is pragmatic and Carnap’s and Morris’s understanding on the other (ibid., 17). It also does not coincide with the difference Richardson rightly sees Carnap drawing between the theoretical and the practical (2007, 299): both the narrow and the wide sense of pragmatics are theoretical.


See Carnap (1938, 42). Once he also provided the following illustration of pragmatics: “the study of the methods of testing hypotheses or theories by first deriving predictions from observation sentences with their help, and then comparing these predictions with new observation sentences which report the results of experiments.” (Carnap 1939b, 222).


The corrections and additions made for the 1950 reprint of “Testability and Meaning” published under the aegis of the Graduate Philosophy Club of Yale University (see the entry in the Arthur Benson’s bibliography of Carnap in Schilpp 1963, 1039) were taken account of (where applicable) in the partial reprint of the paper in Feigl and Brodbeck (1953). (None of the changes made were substantive). Carnap had engaged in similar updating of his terminology when already in February 1938 he mentioned to Quine that some of the examples in Logical Syntax, §75, most of which he now regarded as belonging to semantics—not syntax—actually belong to “pragmatics” (in Creath 1990, 244).


Carnap commented on this obliquely in (1942, vii) and explicitly in (Carnap 1963, 64–65).


See Carnap (1963, 1003) and compare fn. 16 above.


Kasher correctly pointed out that Carnap’s theoretical pragmatics “does not supply us with a general framework for presenting constitutive rules of use, but rather with a segment of a metalanguage” (1975, 272). Carnap’s theoretical pragmatics did not completely answer to Morris’s demand for pure pragmatics.


This contradicts the letter of a claim by André Carus who attributes to the mature Carnap the “dialectical conception” of an “applied discipline of ‘conceptual engineering’” serving as “the successor to philosophy” (2007, 20). For Carus, Carnap’s logic of science was paired with what he calls “pragmatics”: “Carnap never got very far in characterizing the kind of external discourse in which ‘theoretical investigations and practical deliberations and decisions with respect to an acceptance or a change of frameworks’ would be carried on. He gave it a name, ‘pragmatics’, but remained vague about details.” (Ibid., 266) Note that Morris (1963, 88–90) claimed to recognize Carnap’s appeal to “pragmatics” in (i) his designation of the concepts “observable, realizable, confirmable, and testable” as “descriptive concepts”—not logical ones—in “Testability and Meaning”; in (ii) the not further specified decisions that go into setting up the meaning postulates for linguistic frameworks; and in (iii) the settlement of framework questions by expedience, simplicity and fruitfulness. (See also his (1964, 46).) This supports Carus’s claim, but like him Morris fails to engage with Carnap’s own terminology on this point. A discussion of Carus’s Carnap interpretation from the perspective taken here is offered by Uebel (forthcoming).


To be precise, when he wrote “Empiricism, Semantics and Ontology”; as we saw, he still adhered to it when he added correction to a reprinting of “Testability and Meaning”.


Note the convergence with Richardson’s claim that for Carnap the distinction between the theoretical and the practical, of “the realm of representation and belief” and “the realm of decision and the will” is “fundamental” (2007, 299 and 297).


See Carnap’s insistent response (1963, 923) to Sellars’s dark admonitions to the contrary (1963).


In his correspondence with Morris Neurath objected more generally to the threefold division of metatheoretical inquiries as prematurely systematic and harboring the danger that phenomena that do not easily fit into the schema get neglected (letter of 10 January 1938, quoted in Dahms 1997, 58). This complaint was remembered—and conceded to have been justified in retrospect—by Morris as Neurath’s fear that it “would engender pseudo-problems and distract attention from genuine problems” (1946/1971, 301).


Earlier Frank had put matters thus: “Carnap does not investigate elaborately the pragmatic component but does not, on the other hand, deny its existence and relevance” (1963, 164).


“The events around Galileo make it clear that the passionate conflicts connected with a physical theory have nothing to do with its suitability to represent natural processes but much more with their relationships to the political and social events of the time. Therefore there is no need to amplify the positivist conception of science by a metaphysical concept of truth but only by a more comprehensive study of the connections that exist between the activity of the invention of theories and the other normal human activities.” (1932b/1998, 14).


On the falling out see Reisch (2003b), on Morris’s declining influence see Reisch (2005, Ch. 16).


Neurath to Morris, 10 January 1938, partially quoted in Dahms (1997, 58) and Reisch (2003a, 207).


In Uebel (2011) it is argued that recognition of this subscription is crucial to understanding Carnap’s appreciation of Kuhn’s Structure of Scientific Revolutions.


I wish to thank Thomas Mormann for many discussions and two anonymous reviewers for their constructive criticisms which much improved this paper.


Copyright information

© Springer Science+Business Media B.V. 2011