Curt Ducasse, Philosophy as a Science, 1941


Philosophy as Logical Syntax of the Language of Science

CONSPICUOUS among recent attempts to differentiate philosophy from other intellectual enterprises is the one made by the logical positivists. Carnap, who is perhaps at present the most prominent member of the "Vienna circle," in particular has contended that philosophy is the logical syntax of the language of science. It is this contention that we must now examine.

I. What Carnap Says Philosophy Is Not. -- Reduced to their most summary form, Carnap's contentions concerning the nature of philosophy are as follows:

The traditional problems of philosophy may be classified under three heads: logical, psychological, and "metaphysical " Metaphysical propositions Carnap defines as "all those propositions which claim to represent knowledge about something which is over or beyond all experience."{1} But no proposition is even theoretically verifiable unless from it are deducible propositions of perception, different if it were true from what they would be if it were false. Propositions metaphysical in the sense of over or beyond all experience are therefore essentially unverifiable. But a proposition which is not even theoretically verifiable is, speaking logically as distinguished from psychologically, without sense - -- meaningless.{2} Propositions that are metaphysical in the latter sense, and therefore nonsense, include, Carnap asserts, nor only the propositions of what has traditionally been called metaphysics, but also those of traditional epistemology and of normative ethics. They are really pseudo-propositions; that is, they seem to express a judgment or assertion, but in fact express only a command or an emotional attitude. Psychological propositions, on the other hand, are not meaningless, since they are empirically verifiable; but to say that they are so verifiable is to say that they belong to empirical science, not to philosophy. This leaves for philosophy only problems of logical analysis.{3}

I am here concerned only with the thesis that philosophy is logical syntax and I shall therefore nor pause to examine Carnap's dismissal of metaphysics. I may say in passing, however, that although his use of the term metaphysics seems to me unfortunate, I hold that a statement is ambiguous for us if we can not specify any respect in which our experience (whether perceptual or other) would, under specified circumstances, be different if the statement were false and if it were true. It seems to me, moreover, that many of the statements found in books on metaphysics, epistemology, and other parts of philosophy, are of precisely this kind; that is, we do not know of any test of them which, if applied, would either prove them, or disprove them, or show them either less or more probable than their alternatives. Therefore, for the purposes of anyone seeking knowledge, such statements are wholly negligible. And this, I believe, is substantially what Carnap means when he describes them as "nonsense" or "meaningless."

2. Philosophy as Logical Analysis, i.e., Syntax of the Language of Science. -- But what is logical analysis? It is the sort of investigation of which the results are expressed in syntactical sentences -- in sentences of logical syntax.{4} A syntactical sentence is a sentence expressing either a formation rule or a transformation rule of a language system (or a consequence of such rules), that is, either a rule for determining how kinds of expressions to be called sentences of that system can be constructed out of the different kinds of symbols or words (e.g., nouns, articles, adjectives, verbs, etc.) or a rule for determining how given sentences may be transformed into others.{5} Whether a given word is, e.g., a noun, or a verb, etc., is then a matter not of what it stands for but either of the rules for combining it with others or of the form of the word itself.{6} Syntactical sentences thus concern only the form of a language; that is, they do not concern the sense of any sentence of that language or the meaning of any of the words entering into its sentences, but only the arrangements of its words, and the derivation of sentences of it one from another. Using the word "formal" in this sense, the logical syntax of a language may therefore be defined as "the formal theory of that language."{7}

By the language of science, Carnap means "the use of language for making assertions," viz., the part of language consisting of declarative sentences as distinguished from questions, commands, exclamations, etc.{8} Therefore when he defines philosophy as the logical syntax of the language of science, he means that philosophy is concerned exclusively with the rules of formation and transformation of declarative sentences (and the consequences of those rules), wholly without regard to what the words which make up the sentences stand for.

3. Ambiguity of the Term Syntax as Used by Carnap. -- However, the reader who is attempting to discover what exactly he is to understand by the word syntax is much handicapped by the fact that Carnap uses it in three or perhaps even four more or less different senses:

1) In some places, he refers to syntax as being investigation -- specifically, analytical investigation, viz., analysis: "The logical syntax of a language system S consists of . . . . the investigation or analysis of the formation rules of S, and that of the transformation rules of S.{9} Similarly, by implication, when, having defined language as "the system of the rules of speaking," he refers to languages as "objects of logical syntax,"{10} that is, presumably, objects of the kind of investigation called logical syntax. Again, in such an expression as "the task of syntax is . . . ."{11} syntax is apparently being conceived as an enterprise, viz., an investigative one.

2) Elsewhere, Carnap uses "logical syntax" as the name of a theory, "a theory which we will call Logical Syntax";{12} that is, apparently, as the name of a particular sort of theory (in the way in which we might refer, e.g, to the corpuscular as distinguished from the wave theory of light). But he also and more deliberately and often refers to logical syntax as the theory of a particular sort of subject, viz., of the formal aspects of language: "the formal theory of language (known as 'syntax' in our terminology)";{13} "the formal theory of the linguistic forms of . . . . language."{14} Such theory, i.e., logical syntax, is then more particularly described as consisting of the system of formation and transformation rules of a language: "we apply the name 'syntax' . . . . to the system containing [not formation rules only but] both kinds of rules together."{15}

But as pointed out above, a system of such rules is elsewhere called by him a language and languages are called objects of logical syntax. Elsewhere, a language is stated to be a calculus, and a calculus itself stated to be a system of rules of formation and transformation; and logical syntax is then described as "the construction and manipulation of a calculus."{16}

3) On still other occasions, Carnap speaks of "the syntactical method,"{17} syntax then apparently being conceived neither as language nor as theory nor as investigation, but as a method of investigation.

4. Syntactical Sentences and the Formal Mode of Speech. -- This erratic employment of the word syntax is puzzling indeed to the reader, who finds the word or its adjective used on practically every page. However, for the purpose of grasping and assaying the contention that philosophy is logical syntax, it need perhaps be remembered only that a "syntactical sentence," in any case, is a sentence formulating either a formation or transformation rule of a language or a consequence of such rules, and that, according to Carnap, if a sentence is neither metaphysical (and therefore devoid of sense), nor psychological, and yet is philosophical, it is then always really a syntactical sentence. Sentences obviously syntactical are said to be in the "formal mode of speech," the mode, namely, which we use in so far as we speak exclusively of words or other symbols and of the rules for constructing sentences out of them and transforming these sentences into others."{18}

The contention that all genuine philosophical sentences are really syntactical is based by Carnap on a distinction between what he calls "real-object" sentences, "pseudo-object" sentences, and syntactical sentences. He holds that all philosophical sentences which have sense but do not present themselves as syntactical are of the pseudo-object kind; that every sentence of this kind can be translated into a syntactical sentence; and that every pseudo-object sentence is thus shown to concern not a real object but only a linguistic form. Let us first be clear as to the kinds of sentences Carnap designates by those names.

5. Real-Object Sentences. -- Real-object sentences are "sentences which concern not linguistic expressions but extra-linguistic objects,"{19} stating, for example, properties or relations of such objects.{20} Or, more technically, real-object sentences are "synthetic" sentences: synthetic sentences are the genuine statements about reality."{21} A synthetic sentence is itself defined as a sentence which is both not analytic (i.e., is not a consequence of every sentence) and not contradictory (i.e., is not such that every sentence is a consequence of it). A synthetic sentence, in other words, is one which is a consequence of only certain sentences, and of which only certain sentences are consequences.{22}

6. Pseudo-Object Sentences and the Material Mode of Speech. -- A pseudo-object sentence, on the other hand, is one which seems to concern objects because there are used in it words designating objects or matter, but which really concerns linguistic forms.{23} More explicitly, pseudo-object sentences "are formulated as though they refer (either partially or exclusively) to objects, while in reality they refer to syntactical forms, and, specifically, to the forms of the designations of those objects with which they appear to deal."{24} From this it follows that in reality "it is only possible, in any domain of science, to speak either in or about the sentences of this domain, and thus only object-sentences and syntactical sentences can be stated."{25} Pseudo- object sentences, which are said by Carnap to be the "material mode of speech," are accordingly described by him as "quasi-syntactical sentences of the material mode of speech."{26}

7. "Translation" of Sentences of the Material Mode of Speech into the Formal Mode. -- The majority of philosophical sentences, Carnap points out, present themselves in the material mode of speech; and, as already mentioned, his ground for asserting that in spite of their appearing thus to concern objects, they nevertheless really concern linguistic forms and therefore really are disguised syntactical sentences, is that (he believes) all sentences of the material mode of speech can be translated into the formal mode. This translatability, indeed, is declared by him to be "the touchstone for all philosophical sentences, or, more generally, for all sentences which [have sense but] do not belong to the language of any one of the empirical sciences."{27}

The process he designates as "translation" from the one into the other mode of speech turns on the existence of syntactical qualities that are "parallel" to qualities of real objects in the following sense: "a syntactical quality Q2 is called parallel to the quality Q1 [of an object] if it is the case that when, and only when, an object possesses the quality Q1 does a designation of that object possess the quality Q2."{28} On the basis of this definition, the criterion of a pseudo-object sentence (at least when those of the simplest form are concerned) is this: "such a sentence attributes to an object (say a) a quality Ql to which a parallel syntactical quality Q2 can be found. Such a sentence 'Q1 (a)' can then be translated into the syntactical sentence 'Q2 ('a')' which attributes the quality Q2 to a designation of that object."{29}

8. "Translatability" from the Material into tb Formal Mode No Proof that Philosophy is Syntax. -- Before turning to the question whether the process called by Carnap translation from the material into the formal mode of speech is really translation, it is worth noting that even if it really were translation, his conclusion that sentences in the material mode really are disguised syntactical sentences, and therefore that philosophy is really syntax, would not follow. What would follow would be that either they are disguised syntactical sentences, or the syntactical sentences into which they are translatable are disguised sentences of the material mode, i.e., are really sentences about objects (and if so, their truth-value can be ascertained by observation of objects). For if a sentence S is truly and exactly a translation of a sentence Z, then Z is equally truly and exactly a translation of S. And therefore one could assert that such translations into the formal mode are in truth pseudo-syntactical sentences -- i.e., are quasi-objective sentences of the formal mode. One could assert it on the very same ground and the only ground -- viz., translatability -- offered by Carnap for asserting that philosophical sentences are in truth pseudo-object sentences and are to be described as quasi-syntactical sentences of the material mode.

It would thus be perfectly arbitrary which one of such a pair of sentences we chose to describe as a "disguise" of the other, and which one therefore to describe as what the other "really" is; and therefore it would be arbitrary also whether we chose to say that philosophy is really syntax or that syntax, or at least a certain part of syntax, is really philosophy. This point, which in itself has perhaps no great importance, is good at least to show that, even on Carnap's own premise of translatability, his numerous statements that syntactical sentences are the "correct" way of stating what sentences of the material mode "really" are, and that the latter are but "disguises" of the former, represent nothing that he has in the least demonstrated, and are thus wholly dogmatic. But I believe and shall now try to show that even the alleged translatability is not a fact.

9. What Carnap Calls Translation into the Formal Mode of Speech Not Truly Translation. -- An example offered by Carnap of a sentence in the material mode of speech -- or a pseudo-object sentence -- is, "This book treats of Africa"; and the syntactical sentence which he offers as "translation" of it into the formal mode is, "This book contains the word Africa" (or an expression synonymous with this word).{30}

Common sense, however, distinguishes between a sentence and the fact, opinion, or hypothesis which the sentence formulates; and when in ordinary English one sentence is said to be a translation of another, what is meant is that both sentences formulate exactly the same fact, opinion, or hypothesis. An example would be the two sentences "This book treats of Africa" and "Ce livre traite de l'Afrique." But I submit that the two sentences "This book treats of Africa" and "This book contains the word Africa" do not formulate the same fact but each a different fact, and therefore that the second is not a translation of the first at all. For the continent called Africa is one thing and the word Africa is another, wholly distinct; and no statement of the relations the word Africa has to certain other words can possibly be a translation, properly so called, of -- that is, be strictly synonymous with -- any statement about something, such as the piece of land called Africa, which is other than a word.

"It may be possible, it is true, to give a rule for matching certain statements about the piece of land called Africa each with a statement about the word Africa having the same truth-value. There would then be a parallelism of truth-values between the statements of the two sets, and perhaps even between statements derivable from those of each set. But parallelism of truth-values, even if systematic, is one thing, and synonymy is another; for in spite of the parallelism of truth-values, the statements of the two sets would remain formulations of facts or opinions in the one case about a word, and in the other about a piece of land, and thus cannot be translations of each other.

That the sentence "This book contains the word Africa" is not a translation of the sentence "This book treats of Africa" would in any case be proved by the fact that the truth-value of each is independent of that of the other. For there can be a book that contains the word Africa and yet is not about Africa -- for instance, a rhyming dictionary, or a book describing Australia but miscalling it Africa. And there can be a book about Africa which does not contain either the word Africa or an expression synonymous with it. An example would be a book consisting of the following sentence: "The continent about which I am writing contains forests in which gorillas live." For the expression, "the continent about which I am writing," is not in English synonymous with the word Africa. It obviously does not have the same connotation; and as regards denotation, whereas the word Africa does orient the attention of English-speaking readers or hearers to a specific region of the earth's surface, the expression given does not do this even if the author's name appears on the book and much less if the book happens to be anonymous. Indeed, even if the expression were instead "this continent," I should have to insist that it would not be synonymous with the word Africa; for what identifies for another person the continent concerned is not the words "this continent" by themselves (as on the contrary the word Africa by itself does identify it), but perception by the other person both of the pointing gesture performed at the time by the utterer of those words, and of the presence of a continent at the place pointed to.

Before passing to other examples of "translation" offered by Carnap, let us examine what he claims to be the difference between the so- called pseudo-object sentence "This book treats of Africa" and the real-object sentence "Mr. A. visited Africa." He states that "it is really a quality of Africa to be visited by Mr. A.," but that "it is not a quality of Africa to be treated of in that book, because one might know every-thing about Africa and nevertheless nothing about that book."{31}

Obviously, however, the assertion that one might know everything about Africa and yet not know, about Africa, that it is treated of in that book, is a contradiction unless being treated of in that book is not a quality of Africa. Carnap's "evidence" that it is not a quality of Africa thus begs the very question at issue. What he needs to show, he merely asserts.

One can only guess at the criterion he employs to distinguish what is and what is not a quality of Africa. Being visited, i.e., presumably at least walked on, by Mr. A. is counted by Carnap as a quality of Africa. Would then being circumnavigated or being flown over, but never landed on, by Mr. A. be also counted as a quality of Africa? And on the other hand, if being treated of in a certain book is not a quality of Africa, then being treated of in the particular manner called being misrepresented would not be a quality of it either. But would Carnap be willing to say similarly that one could know everything about, for instance, the late Captain Alfred Dreyfus, without knowing that he was misrepresented as a traitor in certain documents? I submit that unless one knows this, one does not know one of the most outstanding facts about him. Likewise, unless one knows that certain Spaniards believed that a golden city or a fountain of youth existed in America, one does not know everything about America, for even certain physical facts about America, viz., that certain Spaniards explored it, can be correctly explained only by reference to the fact that America was thought of by them as containing such things.

Carnap's contention that "This book treats of Africa" is "really" a statement of the relation of the word Africa to certain other words, has already been shown to rest only on his misdescribing as "translation" a process which is other than translation. The remarks just made show that his corresponding assertion -- that being treated of in a certain book is not a quality of Africa -- is supported only through a petitio principii; and that, beside being thus wholly dogmatic, it is on its own merits hardly plausible.

Let us now examine another of the examples Carnap gives of what he calls translation from the material into the formal mode of speech. He takes the sentence "The rose is a thing," and offers as translation of it into the formal mode the sentence "The word 'rose' is a thing- word."{32} The reason he gives for classifying the sentence "The rose is a thing" as a pseudo-object sentence, and for asserting that the other is a translation of it into the formal mode, is that the truth of the sentence "The rose is a thing" can be ascertained "without observing any rose, by only considering to what syntactical kind the word 'rose' belongs, namely that it is a thing-word."{33}

But, we must now ask, how can one tell whether or not a given word of the English language is a thing-word? The answer that would naturally suggest itself would be that it is a thing-word if that to which we choose to apply it is a thing -- as distinguished from, for example, a property, or a place, or a relation, etc. But this does not seem to be what Carnap would answer. I believe he would say, rather, that a word is a thing-word if it is one which, according to the rules of formation of the language, can be combined only in certain manners Ml, M2, etc., with certain other words W1, W2, etc. -- perhaps, for instance, the word milk, in the following manners with the given other words: fresh milk, good milk, cold milk, etc.; drink milk, fetch milk, use milk, etc., etc.{34} For only an answer of this type would constitute a syntactical definition of thing-word.

The question would then arise, however, whether a word which is a thing-word as so defined is also a thing-word in the sense of being a word which is the name of a thing (as distinguished from a property, a relation, etc.). It seems that Carnap would decide this question in the affirmative by stipulation, that is, by ruling that whatever one chooses to designate by a given word is a thing if the word is a thing-word in the syntactical sense of this term; for he gives the rule, for example, that any two objects are to be "assigned to the same genus if their designations belong to the same syntactical genus"!{35}

The proposal embodied in this rule, however, seems to make the extraordinary assumption that conventions (either made by us or already existing) as to word combinations create real likenesses between, or other real characters in, any real entities to which one may choose to apply the words entering into the conventions. But the plain fact, I submit, is on the contrary that the real characters of real entities are what determine the syntactical relations of the words by which we have chosen to designate the entities; and therefore that when we assert about such a real entity as a given rose (or inductively about any entity which is a rose), that it is a thing, our assertion of this is based on the fact that observation of the given entity (or of any other entity which is a rose) has revealed to us that, on the one hand, it resembles such other real entities as trees,. houses, mountains, stones, etc.; and that, on the other hand, it differs from such other real entities as solubility, combustibility, fusibility, malleability, etc. (which we label not things but properties), and differs also from such other real entities as "later than," "between," "two inches from," "helper of," "together with," etc. (which we label not things but relations).{36}

If the "rules" for the combining of given particular words such as rose and thing are not thus dictated by, and because of this, parallel to, the real relations empirically discovered as holding between the real entities we have chosen to label by those words, then the laying down of formation rules is just a verbal game with which we please to amuse ourselves; and any parallelism there may turn out to be between such rules and the relations of any entities that are real in the sense of being other than words is then a pure matter of luck. But obviously, the language to which the everyday words thing and rose belong was not constructed in this arbitrary way; nor were the words thing and rose retained in it because the syntactical relations that had been assigned to them luckily turned out to parallel the real relations of certain real entities nor, still more strangely, because these words magically conferred on any entities to which we chose to apply them real relations parallel to those syntactical ones.

But if, as obviously is the fact in the case of the word thing, the syntactical rules for the combining of it with others are dictated by the empirically observed relations between the real entities to which we have tacked this and the other words, then these "syntactical rules," as well as the more particular sentences from which they are then generalizations, are in truth sentences formulating relations between real entities; and these sentences, if presented in the formal mode of speech, should then be described as quasi-objective sentences of the formal mode of speech.

The upshot of these remarks is then that when we assert of a given real entity that it is a thing, we do so because we observe it to possess a certain set of real (but determinable) : characters; and that when we further assert of the same entity that it is a rose, we do so likewise because we observe it to possess a certain additional set of real (and likewise determinable) characters. The set of characters labelled thing, however, happens to be a part of the set labelled rose; and therefore if we have already observed that a given entity possesses the set of characters labelled rose, we can show, merely by analysis of what we have already observed, i.e., without additional observation of the entity itself, that it possesses also the set of characters labelled thing. But that in such a case this can be shown analytically is no evidence whatever for the contention that the analysis by which it is shown is syntactical analysis, viz., analysis of the relation of the word rose to a class of words called thing-words. For the reasons already stated, I contend on the contrary that it is real analysis, viz., analysis of the relation which a certain set of real characters has to a certain other set of real characters.

I contend, further, that actually we know that the word rose is a thing-word not by discovery that this is a consequence of the syntactical rules of English, but by inference from jointly (a) the empirically discovered fact that the set of real characters we chose to label thing is a part of the set we chose to label rose -- i.e., the fact that the rose is a thing -- and (b) the fact that if it is true that the rose is a thing then it is necessarily true also that the word rose is a thing-word. Theoretically, it might have been through investigation of the syntax of English that we learned that the word rose is a thing-word. We should have learned it in this way if, for instance, we had come upon books written in English and, without being able to know what real entities any of the words in them stood for, we had found that the word rose was often coupled in them by "is" with the word "thing" but never with such words as "property," "relation," etc. But actually this is not the way we learned it.

However, had the sentence used as example been instead, "The word 'chiliagon' is a plane-figure word," the contention that its truth is known by syntactical analysis would have been sound, for the word chiliagon came into English not as a label for the set of characters common to certain perceived real entities, but as the result of a prescriptive verbal definition; and therefore if we should ever identify a perceived real figure as a chiliagon, we should be doing so on the basis of that verbal definition.

On the other hand, it is worth noting that in the case of the word "circle," which -- unlike chiliagon -- came into the language as the name for certain perceived shapes, we do identify perceived real figures as circles (e.g., the figure of a coin, of the full moon, etc.) on the basis of their perceived shape and not on the basis of the geometer's prescriptive verbal definition of the word circle. Indeed, no figure we perceive can strictly be identified as a circle in the geometer's sense of the word.

To sum up now the argument of this section: Carnap contends: (a) that the truth of "the word rose is a thing-word" can be ascertained by only considering to what syntactical kind the word rose belongs; (b) that "the word rose is a thing-word" is a translation of "the rose is a thing"; and (c) that the truth of "the rose is a thing" can therefore be established without observing any rose, and merely by syntactical analysis.

As against this I have argued (a) that actually we know that the rose is a thing by empirical observation of the relation of the two sets of real characters we chose to label respectively rose and thing; (b) that actually we know that the word rose is a thing-word not through investigation of English syntax, but as an inference from our prior knowledge that the rose is a thing. And I now further urge (c) that although the truth of "the word rose is a thing-word" follows from the truth of "the rose is a thing," the first sentence concerns the relation of a word (viz., rose) to certain other words, whereas the second sentence concerns the relation of entities other than words (viz., real roses) to certain entities that are also other than words (viz., real things); and therefore that the two sentences do not formulate the same fact at all and cannot correctly be described one as translation of the other. Carnap would, I believe, grant that the discovery that a given entity is a flower is empirical and real and likewise the discovery that a given entity is a rose; and there seems no good reason why he should not grant also that discovery of the less specific fact that a given entity is a thing (and not perhaps a relation or a property) is likewise empirical and real. It is true that if the entity concerned is not given concretely, but only its name (e.g., "Caro") is given, then he can maintain, as he does, that if that name is not a thing-name, then (by the formation rules of the language) "Caro is a thing" is not to be called a sentence.{37} But this only brings up all over again essentially the same question we have discussed, viz., whether Caro is a thing-name as a matter of syntactical convention (whether already existing, or made by ourselves), or on the contrary in the sense that the real entity to which we choose to apply it is empirically observed by us to have a certain set of real characters, the set, namely, which we have previously chosen to label by the word thing.

Another allegedly pseudo-object sentence used as example by Carnap is, "The evening-star and the morning-star are identical." He gives as translation of it into the formal mode the sentence, "The words 'evening-star' and 'morning-star' are synonymous."{38} Carnap regards the first sentence as illustrating "the deceptive character of the material mode as to the subject matter of its sentences,"{39} since that sentence seems to assert a relation between two objects, although only one planet is concerned.

That sentence, however, actually illustrates only the fact that, in the material mode of speech as in the formal, it is possible to say only clumsily what one means; for the sentence can easily be freed of the defect mentioned and yet remain in the material mode. What it expresses only clumsily can be expressed accurately by "the star seen in the evening is identical with the star seen in the morning." This sentence formulates a belief really about a star observed in the morning and a star observed in the evening; and it resolves a possible doubt as to whether the star observed at the one time is or is not the same as the star observed at the other time. This sentence is thus not about the syntactical relation of one word to another at all, and the sentence, "The words evening star and morning star are synonymous," which formulates a syntactical relation, is therefore not a translation, properly so called, of it. What is true is only that there is a systematic parallelism of truth-value between two sentences so related; but although if a sentence is a translation of another such parallelism is present, the converse does not hold.

There is another example, viz., "This letter is about the son of Mr. Miller," which, although analogous to the one concerning Africa, is worth mentioning because Carnap there considers the possibility that Mr. Miller has no son. He states that even then one can deduce from this sentence, by the ordinary rules of logic, the sentence, "A son of Mr. Miller exists," which is nevertheless then false. And Carnap claims that this shows "that the use of the material mode of speech leads to contradictions if the methods of inference which are correct for other sentences are thoughtlessly used also in connection with it."{40}

But I submit that the example shows no such thing. It shows only that if one starts with a sentence which is false, then, whether it be of the material or of the formal mode, other false sentences can be logically deduced from it. Carnap, it is true, asserts that even if Mr. Miller has no son, the sentence, "This letter is about the son of Mr. Miller" may still be true (the letter "will then merely be telling a lie"!). But obviously this is not so. If Mr. Miller has no son, then (although the letter may still contain the expression "the son of Mr. Miller") what will be true of the letter will be, not that it is about the son of Mr. Miller, but only that it is about an imaginary son of Mr. Miller; and from this one cannot deduce that a son of Mr. Miller exists.

It is worth noting, moreover, that from the formal-mode sentence, "This letter contains a sentence in which occurs the description 'the son of Mr. Miller,' " one can logically deduce that "a letter exists containing a sentence in which . . . . etc." Yet this existential assertion may be false, for it may actually be false that this letter contains any sentence in which that description occurs, or indeed that any letter is there at all.

10. "Psychologism" vs. "Chemistry of Symbols." -- Examination of additional examples would only confirm that what Carnap calls translation from the material into the formal mode is not really translation at all; and since the allegation that it is translation is the basis on which he rests the contention that philosophy is logical syntax, this contention collapses.

This, however, does not preclude the possibility that logical syntax is a part of philosophy. I do not now propose to go into its merits, but a word seems in order concerning what Carnap calls "psychologism," viz., the mistake, as he considers it, "that logic is a science concerning thinking, that is, either concerning the actual operation of thinking or the rules according to which thinking should proceed."{41} In this connection, he contrasts the "meaning" of a sentence in the sense of the thoughts, images, etc., connected with it, with its "meaning" in the sense of such other sentences as are consequences of the given sentence according to the transformation rules of the language to which it belongs. The latter alone is logical meaning or sense; the former, on the contrary, psychological.{42}

But to contrast logical with psychological meaning in this manner seems to me indefensible for the following reason. The laws or rules of nature are properties which the entities of nature have quite independently of man's participation in the transformations that take place among these entities according to those laws. A stone dislodged from a cliff by the frost obeys the law of gravitation wholly without need of man's co- operation. But the laws or rules of formation and of transformation of a language are not similarly properties of its words and sentences themselves. Combinations of words into sentences, and transformations of the latter into other sentences in accordance with such rules, do not occur independently of performance of the combinations or transformations by a human mind. The fact plainly is that the formation and transformation rules of a language are properties not of its words and sentences but of some human mind or minds. That is, they are habits that some mind already has or that it proposes to adopt at least for a time. Indeed, a verbal entity is truly a word or a sentence at all only in so far as the habits of some human mind connect it regularly either with some non-verbal experience or with certain other verbal entities in specific ways.{43}

The rules of a language, I submit, thus are what Professor Carnap denies that they are, viz., ways in which certain minds think or propose to think on certain occasions, and ways in which the thinking of certain minds should proceed on certain occasions if it is to succeed in what it aims at. To deny this would be to imply that words can be endowed by man with quasi-chemical properties of their own in accordance with which they then become capable of combining independently of his presence.

That a language is a system of habits of human beings is indeed stated by Carnap himself in a recent publication.{44}It is quite true, as he also states,{45} that one can abstract from the speaker and deal only with the expressions of the language and their relation to their designata, or even deal only with the expressions. But to abstract from the speaker or hearer can properly mean only to take him for granted throughout without mention; not, of course, to suppose that logical rules, which are habits in him, are properties of symbols independently of him. The possibility of abstraction from speakers or hearers thus does not render compatible the two assertions made by Carnap, viz., that the rules of logic are not rules of the thinking of certain human beings, and that language and therefore its rules are a system of habits of human beings.

11. Syntactical Treatment of Philosophical Problems. -- The criticisms set forth in the preceding pages seem to me to dispose of Carnap's contention that the only genuine philosophical problems are problems of the syntax of the language of science; but these criticisms should not be construed either as denying that logic is essentially syntax -- for I think this is true -- or as denying that syntactical treatment of some philosophical problems can be very fruitful. To account for this fruitfulness, however, it is no more necessary to suppose that these philosophical problems are themselves problems of syntax than it is necessary, in order to account for the power of arithmetic to anticipate in certain respects the outcome of physical operations, to suppose that these physical operations are themselves operations of arithmetic.{46}
[Table of Contents] [Chapter 8]


{1} Philosophy and Logical Syntax (1935). [Back]

{2} The Unity of Science (1934), p. 26. [Back]

{3} P. L. S., pp. 9-34. also, The Logical Syntax of Language (1937), pp. 277-81. [Back]

{4} P. L. S., p. 68. [Back]

{5} P. L. S., pp. 41, 43. [Back]

{6} L. S. L., p. 2. [Back]

{7} P. L. S., p. 39. [Back]

{8} Foundations of Logic and Mathematics, International Encyclopedia of Unified Science, I, No. 3, p. 3. [Back]

{9} P. L. S., p. 45. See also L. S. L., p. 315, where the logical analysis of physics is stated to be "the syntax of the physical language," See also F. L. M., p.1, where the logical analysis of language is presented as consisting of semantics and syntax. [Back]

{10} P. L. S., p. 41. [Back]

{11} P. L. S., p. 58. [Back]

{12} P. L. S., p. 39. [Back]

{13} L. S. L., p. xvi. [Back]

{14} L. S. L., p. 1. [Back]

{15} P. L. S., p. 47; L. S. L., p. 2. [Back]

{16} P. L. S., p. 4, 5. [Back]

{17} P. L. S., pp. 39, 68; L. S. L., p. 8, "the method of syntax," and p. 11, "the syntactical method." [Back]

{18} L. S. L., p. 64. [Back]

{19} L. S. L., p. 60. [Back]

{20} L. S. L., p. 277. [Back]

{21} L. S. L., p. 41. [Back]

{22} L. S. L., p. 39, 40. [Back]

{23} P. L. S., pp. 60, 64. [Back]

{24} L. S. L., p. 285. [Back]

{25} L. S. L., p. 331. [Back]

{26} L. S. L., p. 285. [Back]

{27} L. S. L., p. 313. [Back]

{28} P. L. S., p. 63; also L. S. L., p. 287. [Back]

{29} P. L. S., p. 63. [Back]

{30} P. L. S., p. 61, 65. [Back]

{31} P. L. S., p. 65; see also L. S. L., pp. 285-86, where Carnap uses "property" instead of "quality." [Back]

{32} P. L. S., pp. 61-62; L. S. L., pp. 293, 297. [Back]

{33} P. L. S., p. 62; L. S. L., p. 293, [Back]

{34} Cf. Bloomfield, Linguistic Aspects of Science, Int. Enc. Unif. Sci., I, No. 4, p. 26, from which this example is borrowed, and to which Carnap refers his readers elesewhere (F. L. M., p. 5). [Back]

{35} L. S. L., p. 293. [Back]

{36} In referring to such properties and relations as real, I do not mean anything more or less metaphysical than does Carnap when he refers to, e.g., a rose, as a real object -- viz, that it is other than verbal. Since Carnap regards the questions debated in the controversy over universals as pseudo-questions (ibid., p. 311), I assume that he would not embrace the nominalism which would, as he does, class, e. g., a rose, as real, but would refuse to call likewise real the properties or relations it has. [Back]

{37} L. S. L., p. 293, ex. 1. [Back]

{38} P. L. S., pp. 61, 66. In L. S. L., p. 290, Carnap makes clear that by "synonymous" he means in this case "P-synonymous" and not "L-synonymous" -- the difference between the two, stated in more familiar language, being roughly that between sameness of denotation and sameness of connotation. [Back]

{39} P. L. S., p. 67. [Back]

{40} L. S. L., p. 291. [Back]

{41} P. L. S., p. 34. [Back]

{42} L. S. L., p. 42. [Back]

{43} Cf. the witer's "Symbols, Signs, and Signals," Journal of Symbolic Logic 4 (1939), p. 41. [Back]

{44} F. L. M., p. 3. [Back]

{45} F. L. M., p. 4. [Back]

{46} Cf. "Symbols, Signs, and Signals," pp. 50-52. [Back]

[Table of Contents] [Chapter 8]