Why Semantic Ascent Fails
From: METAPHILOSOPHY, Vol. 14, Nos. 3 & 4, July/October 1983, pp. 276-290. Page numbers in "< >" refer to this journal.
1. Introduction. Linguistic philosophy is rumored to be dead and gone, its passing unnoticed until long after the fact, and largely unmourned. Gilbert Harman, for one, places its demise well into the past, referring to “. . . the arguments that undermined [note the past tense] the old linguistic philosophy and dating its end two decades ago, in 1960. According to him, “The period of linguistic philosophy . . . [ran] from 1930-1960”, after which philosophers came “. . . to believe again that philosophy need not be restricted to, the analysis of language”. This change of conviction at least partly resulted, he suggests, from arguments by Willard Quine and others which showed “. . . that there can be no real separation between questions of substance and questions of meaning”.1
One might suspect, however, that the matter is not so clear as such statements would make it seem. Some even respond by saying that there never was such a thing as “linguistic philosophy” anyway. But, while this denial doubtlessly makes all important point, surely there did exist for some decades a discernible philosophical tendency which it is not entirely unuseful or misleading so to designate. And, just as surely, something of the practice of that movement still persists, even though its major dogmas are long abandoned. At least one can confidently say that no clear substitute for language, as the primary subject matter of philosophical analysis and speculation, has emerged. Quine himself, whose arguments are alleged to have helped undermine linguistic philosophy, continued to assign “semantic ascent” an important, if not essential, role in philosophical discussion. And no less than a few years ago, human knowledge—which remains the focus, more than anything else, of philosophical analysis—was usually treated as something essentially linguistic: possibly a “fabric of sentences”.
However, what most obviously seems in need of resistance is any suggestion to the effect that linguistic philosophy was done away with by means of arguments that showed it to be in some way fundamentally mistaken. The arguments to which Harmon refers above in fact appeared as arguments within the genre favored by linguistic philosophy. They certainly did not have the force of negating the Linguistic Turn in general, but at most that of showing that it must be taken only in a certain manner—which we may loosely call tile “pragmatic”. G. J. Warnock’s words about the fall of Idealism, earlier in this Century, seem a fair characterization of what happened to linguistic philosophy:
<277> Such systems are more vulnerable to ennui than to disproof. They are citadels, much shot at perhaps, but never taken by storm, which are quietly discovered one day to be no longer inhabited. The way in which an influential philosopher may undermine the empire of his predecessors consists, one may say, chiefly in his providing his contemporaries with other interests.2
But this sort of explanation leaves us with the question of whether or not there really was something fundamentally mistaken in “the old linguistic philosophy”. If not, then the turn away from it is something which ought to be resisted. But, if so, then to make clear its error will fortify our bored declension with some good reasons. In order to make the issue manageable, I shall focus upon one of the less radical versions of the Linguistic Turn: Quine’s method of semantic ascent. Semantic ascent, as Quine understands it, does indeed fail as a general methodological device for philosophy; and—although I shall not argue this point here--the reason why it fails seems also to apply to the other interesting versions of the Linguistic Turn which were exemplified in the career of linguistic philosophy.
2. The main assumption of semantic ascent. Semantic ascent is a methodological strategy in philosophy in which one turns (or “ascends”) from speaking—or attempting to speak--of certain apparently non-linguistic matters to speaking of correlated entities, events, or structures that are constituents of language, or are in some sense linguistic. Typically, one turns to talking of correlated words or sentences. Quine remarks:
It [semantic ascent] is the shift from talk of miles to talk of ‘mile’, it is what leads from the material (inhaltlich) mode into the formal mode, to invoke an old terminology of Carnap’s. It is the shift from talking in certain terms to talking about them . . . .3
The point of semantic ascent, for Quine, is to allow those who use philosophically interesting terms or locutions differently—and who, therefore, cannot use them in discussions with each other without confusion and begging of questions at issue--to withdraw from the use to the mention of the terms and locutions in question, there to find a common conceptual ground. Thus Quine says:
The strategy of semantic ascent is that it carries the discussion into a domain where both parties are better agreed on the objects (viz. words) and on the main terms concerning them. Words, or their inscriptions, unlike points, miles, classes, and the rest, are tangible objects of the size so popular in the market place, where men of unlike conceptual schemes <278> communicate at their best. The strategy is one of ascending to a common part of two fundamentally disparate conceptual schemes, the better to discuss the disparate foundations. No wonder it helps in philosophy.4
And discussing, in this same passage, the transition from an allegedly hopeless discussion about whether miles exist to a discussion about the uses of the word “mile”, Quine concludes that “...then we can get on; we are no longer caught in the toils of our opposed uses”.
Clearly the assumption of semantic ascent in Quine’s hands is, then, that when philosophical disputants use the appropriate names of words or sentences to mention the words or sentences in question—and when they use the related terms and conceptual apparatus required to formulate statements or sentences about those words or sentences-differing philosophical viewpoints or theories of the disputants will not be presupposed, or will be presupposed to a significantly smaller degree. To repeat the crucial phraseology: “Both parties are better agreed on the objects (viz. words) and on the main terms concerning them.” But is this in general true? Can’t it be shown false in significant cases? If so, we well may have found a reason why semantic ascent must fail as a general philosophical strategy. Although it may sometimes be innocent and useful, there will be at least some, and possibly many, important philosophical discussions in which semantic ascent will not help, and may indeed be harmful, because of opposed philosophical views precisely concerning words, sentences, or other linguistic items or structures mentioned in making the ascent.
3. Apologia for McX. Consider Quine’s encounter with the fabulous McX. Quine purports to have used the method successfully with McX. But McX, being fabulous, has of course been in no position to talk back.
The initial difficulty between Quine and McX arises when they wish to disagree about the existence of something such as Pegasus. Since McX believes that in some sense Pegasus exists, he, at least, can consistently state that there is an entity which Quine rejects. But Quine, it might seem, cannot say this, precisely because he rejects Pegasus as an entity.5 As Quine states the point:
If Pegasus were not, McX argues, we should not be talking about anything when we use the word; therefore it would be nonsense to say even that Pegasus is not. Thinking to show thus that the denial of Pegasus cannot be coherently maintained, he [McX] concludes that Pegasus is. (p. 2)
The problem, then, is to get around the initial divergence in theory which determines how the crucial terms are used, and to find a common ground from which to start. Semantic ascent is supposed to accomplish this, and <279> there is no question but that Quine Supposes that he has used semantic ascent with McX to achieve this end. Turning from Pegasus, he deals with “Pegasus”. On page sixteen of “On What There Is”, he states that, as a result of operating on a semantical plane”, he is able consistently to
...describe our disagreement by characterizing the statements which McX affirms Provided merely that my ontology countenances linguistic forms, or at least concrete inscriptions and utterances, I call talk about McX’s sentences.
He holds, then, that “withdrawing to a semantical plane” allows him “to find a common ground on which to argue” with McX.
Disagreement in ontology involves basic disagreement in conceptual schemes; yet McX and I, despite these basic disagreements, find that our conceptual schemes converge sufficiently in their intermediate and upper ramifications to enable us to communicate successfully on such topics as politics, weather, and, in particular, language. In so far as our basic controversy over ontology can be translated upward into a semantical controversy about words and what to do with them, the collapse of the controversy into question-begging may be delayed. (loc. cit.)
But when we look at what Quine actually did with McX, we find that he simply attacked McX’s position (or analysis or theory) concerning meaningful names. Specifically, he argues against McX’s (alleged) view that “...Pegasus...must be because otherwise it would be nonsense to say even that he is not”. He retreats, not from philosophically contested points to philosophically neutral ground, but from philosophically contested points about what exists to philosophically contested points about the nature and function of names. One will surely search in vain for passages in which Quine and McX “... communicate successfully on such topics as ... language”.
And what, if there were such a passage, might we safely imagine it to contain? Certainly only such philosophically neutral and uninteresting things—truly on a par with talk of politics and the weather—as that “Pegasus” has seven letters in it, is to be found in black on this page, occurs in some (English) books on mythology, functions grammatically as a noun etc. etc. But even these modest statements may quickly raise questions that are philosophically debatable and debated. For example, how can “Pegasus” both be on this page and also in books on mythology? The type/token distinction is often invoked at this point. But it in fact cannot be invoked if we are to remain at the level of unquestionably “successful communication”. As we shall discuss below in more detail, there is much disagreement, and possibly confusion also, among philosophers on what, precisely, the type/token distinction is as applied to “words”; and the drawing of it by a given philosopher normally pulls a considerable load of ontological freight. Very little <280> indeed can be said of “words” that is philosophically clear and uncommitted from the outset.
In fact, when Quine says (above) that he can consistently describe his disagreement with McX by “characterizing the statements which McX affirms”, one can only wish him well on getting a characterization of McX’s statements which will not, beg other ontological or philosophical points against McX, concerning the nature of statements themselves. Quine even seems to use, “statement” and “sentence” interchangeably in this passage (p. 16). After speaking of characterizing statements, the next sentence reads: “Provided merely that my ontology countenances linguistic forms, or at least concrete inscriptions and utterances, I can talk about McX’s sentences.” Statements, sentences, and concrete inscriptions and utterances! Thus quickly has semantic ascent led us into an ontological briar patch! And, given what we already know of McX’s philosophical proclivities, it is at least a fair bet that he will not take his sentences to be concrete inscriptions or utterances, nor his statements to be sentences, and will insist that the discussion was, in any case, supposed to be about the ontological assumptions of using “Pegasus” in a linguistic act of saying that Pegasus does not exist.6
When we then come to Quine’s explicit attempt to “take steps” against McX’s position, we find that he merely begs the question against him. Quine proposes that we can “...meaningfully use seeming names without supposing that there be the entities allegedly named” by following Russell’s theory of descriptions, and treating the sentences in which the names occur as equivalent to certain compound sentences in which the names do not occur. But McX might—and if he is a hard-nosed philosopher, probably he would—simply reply that if the meaningful use of the compound sentence does not imply the existence of what was named in the use of the original sentence, then the paraphrase is not equivalent to that original sentence after all. He well might respond: “To say that, upon substitution of the paraphrase for the original sentence, the meaningful use of the original is shown not to imply the existence of what was named in that use, is like saying that strychnine is not poisonous when sugar is substituted for it.” Quine has his way here with McX only by presupposing McX’s consent to a broad swatch of a particular philosophy of language that by no means consists of, or follows from, simple observations about those “...tangible objects [words] of the size so popular in the market place, where men of unlike conceptual schemes communicate at their best”.
<281>It must be emphasized that the point here has nothing to do with who is right or wrong in this dispute about the analysis of names. Specifically, it is not claimed that McX’s views are correct. Rather, the point is that, in this case of the quarrel with McX, semantic ascent does not provide a domain in which philosophical disputants can agree, or at least disagree in a manner which does not force the begging of philosophically significant questions or outright inconsistency.
It should also be emphasized that semantic ascent does have the advantage of locating premisses from which ontological conclusions are frequently drawn. In the case at hand, it allows us to see that McX rests his case upon what others well may regard as a wholly gratuitous theory about the use of names. He can no longer, given semantic ascent, hope simply to use his favorite argument upon Quine.7 But the difficulty of finding a common conceptual ground upon which the ensuing disagreement, now within the philosophy of language, can be conducted falls heavily upon the parties involved. The function of securing a common conceptual ground was supposed to have been served by semantic ascent into metalanguage. This ascent has failed in this respect in this case. Further ascents, into meta-meta-etc-languages do not obviously offer promise of help where the help is needed, It may even, with some justification, be thought that several decades of work by those who have taken the linguistic turn seriously shows that it is rare to find shared, wide-ranging conceptual commitments in the remarks of those who have semantically ascended. (Are not Wittgenstein and Carnap, John Wisdom and Gustav Bergmann only by courtesy or confusion said to have been talking about the same thing?) And insofar as in our metalanguage we are still “speaking of objects”, Quine would be the very first to deny that semantic ascent will free us from ontological presumption. At most we can hope that the one we are trying to talk to is equivalently presumptuous—is ontologically relativized to the same reference points—as we now discuss words and language.
However, it is seldom clear, in Quine’s writings, just when he supposes he is standing within the domain of that successful, non-partisan communication allegedly provided by discourse about words, and when he regards himself as having moved beyond that, into the domain where philosophical and other theories about words (and, more generally, about language) govern the discussion. One clearly should not assume that all of his remarks about language are intended by him to be of the former sort. Yet it does seem that, when Quine is speaking to some philosophical problem, almost nothing of importance that he says about words or language is philosophically unprejudiced, or free from some thoroughly philosophized conceptual scheme. Thus, poor McX comes in for a beating because of “...the occult entities [attributes] which he posits under such names as ‘redness’ ”. (p. 10) Quine admits, of course, that there are red houses, roses, and sunsets, but denies <282> that things such as these have anything in common which would constitute the attribute of redness. Using, we may suppose, the same tone in which he discusses politics and the weather, he remarks:
The words ‘houses’, ‘roses’, and ‘sunsets’ are true of sundry individual entities which are houses and roses and sunsets; but there is not, in addition, any entity whatever, individual or otherwise, which is named by the word ‘redness’, nor, for that matter, by the word ‘househood’, ‘rosehood’, ‘sunsethood’. (loc. cit.)
Certainly if Quine intended this to be one of those statements “...where both parties are better agreed on the objects (words) and on the main terms concerning them”, he has not succeeded with his intention. For just what is involved in the word “red” being true of each of the sundry individuals that are red is both puzzling and highly contestable. Consider only those problems involved in the choice between substitutional and objectual interpretations of quantification. And again, “the word ‘red’ ” certainly cannot just refer to the particular inscription which Quine produced when he wrote these lines decades ago, and which probably no longer exists. If it has ceased to exist, the word“red” doubtlessly would still be regarded by Quine as “true of” all of the various red things. But the word “red” then is certainly some sort of abstract entity;8 and its status as something different from any of “its” concrete inscriptions or utterances, and actually extending to (being “true of”) all and only red things (members of its extension), is perhaps at least no less occult than McX’s attributes.
Questions about, and theories of, the unity and character of the term “red” itself aside, one should at least expect McX to reply that to postulate the extending of “red” to (or its being “true of”) certain objects alone, without an attribute—present in just those objects and no others—to guide its “reach”, is merely to assume that his own view of the extension of a term is false. For him, the extension of a term clearly is not something intrinsically intelligible and philosophically innocent. Semantic ascent, once again, only brings Quine and McX from one philosophical standoff (over the existence of attributes) to another (over what it is for “a word” to be “true of” a certain range of objects).
But enough of McX! Can we make any general statements on what it is about language (or “words”) that makes semantic ascent an attractive strategy in philosophy, and yet does not allow it to succeed?
4. Token, tone, and type. The trouble seems to lies in the fact that “words” lead at least a double life. This is something which seems to be generally understood, although its implications for the practice of semantic ascent have not been fully appreciated. Insofar as philosophically and theoretically <283> innocent discourse about words is a significant possibility, it will mainly apply to what Peirce called “tokens”. The word as token is an individual physical entity or event, publicly observable, which can be written, spoken, erased, heard, seen, misspelled, or eaten (on birthday cakes or dangerous notes). It can be loud, soft, black, white, located in the corner of the blackboard, and so on. Because tokens really are “...tangible objects of the size so popular in the market place”, statements mainly about them--or straightforwardly verifiable by reference to them--do indeed enter into communication on a par, for philosophical neutrality, with talk about politics and the weather. And hence one can readily see how a clarity-hungry philosopher might be drawn to talk of “words”. Carnap, at one time, was a case in point. He once held that the sentences of philosophy
...are in part sentences of arithmetic, and in part sentences of physics, and they are only called syntactical because they are concerned with linguistic constructions, or more specifically, with their formal structure. Syntax, pure and descriptive, is nothing more than the mathematics and physics of language.9
The reference to physics suggests that he is referring mainly to particular physical entities or events, which would of course be tokens. At least one of his defenders10 explicitly stated that the “formal mode” was entirely about tokens. If so, that may serve to explain why a few turns in the formal mode was supposed to clear the mind of philosophical fogs arising from the bogs of the material mode.
However, if we are going to speak of tokens, we must take care to heed Augustine’s warning “...that we should not attribute more to words than is proper”.11 Because they are tokens, only certain characteristics can be assigned to them. Yet, to heed the warning will limit us to remarks about “words” which, at best, are philosophically banal. Possibly this explains why Carnap quite justifiably turned from syntax to semantics, from the early forties on. But then, unlike syntactical predicates, semantic predicates offer no hope at all for treatment as simple properties of tokens12—nor do pragmatic predicates fare any better in this regard. It turns out that only beyond the point where talk of “words” remains philosophically unproblematic <284> does it become philosophically interesting. For, once we depart from tokens, we find no comparable clarity with regard to other senses of the word “words”.
And this is true, specifically, of that important sense of “word”—whatever it may precisely be—in which the same “word” occurs many times, and “...can be encoded in spoken and written symbols...”,13 or in those yet more recondite “symbols” which brain-states or brain-processes may turn out to be. Such a “word” as this is often called a “type” word, with the intent of following Peirce’s terminology here also. But Peirce in fact drew a three-fold distinction with regard to words or signs.14 In addition to tokens, he spoke of signs which are tones, and those which are types. The tone (“qualisign,” or a “first”)15 is a possible sign: a shape, or a color, or a tonal quality—or some combination of all of these—which could be a predicate of many written or spoken tokens. A tone seems very close to what many subsequent writers have meant by“type” —although “type” also is frequently used to refer to a class of tokens.16 By contrast, a Peircian type (“legisign”, or a “third”)17 is a law: a specific dispositional property resident in tokens of a certain tone, in virtue of which the token constrains appropriate interpreters to think of a determinate object. Thus, Peirce’s type well might be considered the token’s meaning.
After Peirce, “type” is used in a variety of senses—disregarding, of course, its use in connection with Russell’s Theory of Types. According to R. B. Braithwaite, it was F. P. Ramsey “...who introduced Peirce’s words, ‘type’ and ‘token’, into modern logical discussion”.18 Ramsey stated that “...tokens [of sentences] are grouped into types by physical similarity (and by conventions associating certain noises with certain shapes), just as are the instances of a word”.19 Braithwaite read Ramsey as saying, for example, that “...the legisign [type] ‘the’ is the class of written words having a certain physical similarity”.20 A class, of course, is neither a tone nor a type in Peirce’s sense; and so Peirce’s type is totally omitted here, since even the class suggested is specified by “physical similarity”, which could only be that qualitative structure which is Peirce’s tone.
<285> Such omission seems characteristic of subsequent uses of “type” or equivalent expressions, as opposed to “token”. Thus, R. M. Eaton held that
Recurrence is the essence of symbols... The symbol is all the cases of its own recurrence. A symbol then is not a single object, but a class of objects. Symbols are the same symbol only if they exhibit a generic similarity as objects, and they are distinct symbols if they exhibit a generic difference. Thus a and b are distinct symbols because their contours as marks on a page are different, while a and a (if they are not ambiguous) are the same symbols because their contours are alike.21
In his Inquiry intoMeaning and Truth, Russell said that “The spoken word ‘dog’ is not a single entity; it is a class of similar movements of the tongue, throat, and larynx.... The word ‘dog’ is a universal, just as dog is a universal.“22 Hans Reichenbach stated that “... linguistic signs must be reproducible.... The individual sign is called a token.... The class of similar tokens is called a symbol.“23“Symbol” and “word” as used in these statements seem to be used in close to the same manner as “type” in the statements by Ramsey and Braithwaite. Kneale and Kneale, on the other hand, omit any reference to classes, and describe a “type-sentence” as “...a complete pattern of sounds or marks having meaning”, and, again, as “...the common features of utterances that resemble each other almost completely in sound or some other perceptible character, but may differ considerably as vehicles of communication.24
Another important but obscure issue concerning the nature and categorial status of types, and words as types, should be mentioned. That is the relationship between type and meaning. As suggested above, one well might understand Peirce as holding type to be meaning. But this is not true of “type” as subsequently used. Nonetheless, most of those who have discussed types have placed them in some significant relationship to meaning. Ramsey himself defined a proposition (as distinct from a sentence) as “...a type whose instances consist of all propositional sign tokens which have in common, not a certain appearance, but a certain sense”.25 The sentence type, for him, does not seem to be specified by a particular meaning. However, it is doubtful that he would allow that a set of marks or sounds was a sentence or word unless it <286> had some meaning or other. While a mark or sound must have some meaning in order to be a word type, we can have one type or “word” which bears different meanings in different tokens, as with ambiguous terms and homonyms, and we also can have one meaning with different type words, as with “table” and “Tisch”. This was the view of L. S. Stebbing, in her “Sounds, Shapes, and Words”.26 Five years earlier she had advanced the view that “...symbols which refer to different referends are different symbols”.27 This earlier view of hers is certainly the view suggested by Eaton’s remarks above, to the effect that if a is ambiguous, then a and a might be different symbols (type words). By saying this, he makes the particular meaning determinative of the particular word type. The same position was taken by Max Black,28 who supposed that he was following Wittgenstein in this respect. Such a view seems counter-intuitive. It seems that we often know which (type of) word we have before us, while in doubt as to its meaning. And it is not clear how homonyms and ambiguous terms are to be described, if this view of the relation of meaning to type is accepted.
Possibly a still different position on this matter makes a meaning a feature or property of the type—not a part of it. It may be that the view of Wilfred Sellars should be understood in this way. In a paper already referred to, Richard Rorty explains that on Sellars’ ”mythical” account,
...thoughts were originally theoretical entities, postulated as ‘inner’ states that explained certain sorts of behavior. But they were not merely Rylean dispositions nor Armstrongian ‘states apt...’; for they had certain intrinsic features. For example, they were true or false, and were about things, in the way in which sentences are. They shared, in other words, the ‘semantical’ features of sentences—the features sentences possessed not qua physical objects (inscriptions) but qua types (as opposed to tokens)—but had no other features.29
Everett Hall also held to this interpretation of Sellars, although, for reasons which he gave, he was puzzled as to how Sellars’ type sentence, referred to by his special dot-quotes, could be said to designate anything at all.30 Sellars’ view, that semantic features are predicates of types, should perhaps be contrasted with views such as Russell’s, expressed in the passage in Inquiry into Meaning and Truth already referred to, where “psychological characteristics” such as “intention” or “meaning” are apparently ascribed to tokens. Possibly yet another view of the relation of meaning to symbol or type—utilizing some unspecified part/whole relation—is suggested by the early Wittgenstein, who <287> incorporated meaning into the symbol by holding that “The sign [token] is that in the symbol which is perceptible by the senses”.31
Of course, given what seems to be the intrinsic obscurity of talk about meaning, any attempt to relate type (word) to meaning should be expected to obscure or, at least, theoretically complicate talk of “type words” yet further. Moreover, although a categorial gulf may lie between them, both tones and types—however specified—are abstract entities. Obviously these “Words” will not in general provide a neutral meeting ground for philosophers of divergent conceptual schemes. Nominalists and Empiricists, for example, are unlikely to be able to meet their adversaries by ascending to such “words”. And where positions on the status and knowledge of abstract entities have not already explicitly hardened, what one is ascending to when one ascends to “words” other than tokens is so philosophically questionable that no philosophical technique could in the long run be founded upon it. Clearly, even if Peirce’s way of describing “type” words had been quite unproblematic, the course of subsequent philosophizing would have made it impossible to use the term “type” in order to associate any clear and philosophically innocent concept with the word “word”. I have gone to quite tedious lengths to demonstrate this point, for it seems to me that obscurities about the word as type are, more than anything else, responsible for the failure of semantic ascent to work as a general method in philosophy.
It should be added that the problems with “word” talk are only compounded by speaking, as I have done above, of “type word” and “token word”. I have followed this usage thus far because it is fairly common, and nothing was to be gained for points made above by splitting the relevant hair. However, the suggestion that both types and tokens are words is mistaken or, at best misleading. To be sure, we do say, as Peirce pointed out, that there is but one word “the” in the English language. But this is no more to be taken au pied de la lettre than is the statement that there is only one poisonous lizard in the continental United States, viz. the gila monster. There is not one lizard which is the “type-lizard”, and many other lizards which are the token lizards. Likewise, there is not one word which is the type, and many other words which are the tokens.
5. Philosophical impasses created by semantic ascent? Given that there is so much room for divergence in the understanding of the nature of “words”, one might expect that that peculiar sort of incomprehension (and exasperation) which marks a philosophical impasse would emerge in some of the philosophical encounters where recourse is had to semantic ascent. This seems to be precisely what occurred, for example, with attempts by the older Bertrand Russell to come to grips with linguistic philosophy. His well-known comment was that the “new philosophy,” as he then called it,
<288> ...seems to concern itself, not with the world and our relation to it, but only with the different ways in which silly people can say silly things. If this is all that philosophy has to offer, I cannot think that it is a worthy subject of study. The only reason that I can imagine for the restriction of philosophy to such triviality is the desire to separate it sharply from empiricalscience. I do not think such a separation can be usefully made.32
Perhaps a similar exasperation is evinced in C. D. Broad’s reference to how his younger friends “...dance to the highly syncopated pipings of Herr Wittgenstein’s flute”.33 Surely these should not be passed off as simply the grumblings of crotchety old men, but as remarks due at least in part to a fundamental obscurity in the practice of those philosophers complained of. What one blurb called the “fiery controversy in Britain’s intellectual establishment” touched off by Ernest Gellner’s Words and Things, an attack on linguistic philosophy, is another case in point. The quality of the attack and the counter-attack, with suggestions of low motives or abilities (or both), cannot be satisfactorily explained without some reference to the essential obscurity of talk about words or language in philosophical contexts where “the ascent” is made. Attempts at discussion between “Analytic” and “Continental” philosophers often exhibit the same type of mutual—but usually kindly—incomprehension: with any exasperation courteously suppressed for the duration of the encounter. Indeed, even “Ideal” and “Ordinary” language philosophers have surely come to philosophical impasse or by-pass through semantic ascent upon occasion, as was suggested above. Exchanges between Ryle and Carnap, or between Strawson and Quine, may be profitably read with the possibility of this in mind.34 The common ground allegedly provided by semantic ascent turns out, in practice, to be much more narrow than it was hoped, and to be marked out, after all, only by shared philosophical assumptions—this time about “words”.
Even G. E. Moore, thought by some to be the father of it, found semantic ascent puzzling. In the course of his paper, “Moore’s Notion of Analysis”, C. H. Langford spoke in such a way as to suggest that Moore might be understood as analyzing verbal expressions.35 He did not, in fact, press this point of view upon Moore; but Moore, in his response, seems to go out of his way to reject it, and to stress how little sense he can make of it. After emphasizing that he never intended to use “analysis” in such a way that what he analyzed might be a linguistic expression, he remarked:
<289> There is, of course, a sense in which verbal expressions can be ‘analyzed’. To take an example from Mr. Langford: Consider the verbal expression ‘x is a small y’. I should say that you could quite properly be said to be analysing this expression if you said of it: ‘It contains the letter “x”, the word “is”, the word “a”, the word “small”, and the letter “y”; and it begins with “x”, “is” comes next in it, then “a”, then “small”, and then “y”.’ It seems to me that nothing but making some such statement as this could properly be called ‘giving an analysis of a verbal expression’. And I, when I talk of ‘giving an analysis’, have never meant anything at all like this.36
There is no doubt that Moore was serious here, and—because of the actual obscurity of philosophical talk about words?—really could not understand Langford’s suggestion.
6. Quality, idea, word. After all of this, however, one must nonetheless concede that, within certain limits, talk of words causes no philosophical perplexity and assumes no philosophical viewpoints. Further, clear references to words may make it possible for philosophers to pin down and jointly focus upon the always somewhat recondite subjects of philosophical inquiry. (If mind is itself just too slippery, let us by all means speak of, or write books on, the concept of mind—at least until “concepts” become as philosophically slippery as minds!) But it must then be added that this is no more true of words than it is of many other physical objects or events—of which words, in the clearest of senses, after all form only one sub-class.
Within limits, even our talk of qualities and experiences (“ideas”) is also quite unproblematic. In a certain sense, everyone very well knows what blue and square are; and that is at least a part of the common sense foundation of Platonism. But after centuries of philosophical controversy over how qualities are to be understood, they were dismissed in favor of “ideas”. Leibniz, for example, held sensible qualities to be “occult”.37 About the same time he did so, it was written in the Port Royal Logic that “Some words are so clear that they cannot be explained by others, for none are more clear or more simple. ‘Idea’ is such a word.“38 And this also is quite true—in a sense. We have roughly the same level of clarity about our experiences (or “ideas”) as we do about qualities, and about words and other physical items. We can identify and re-identify them, both in ourselves and in others; and we can classify them and relate them in many ways to each other and to many other things. Without this being so, the creation and appreciation of sciences and various types of art objects would—along with much else in human life, such as moral training and judgement—simply be impossible. This was the <290> strength of the appeal to experience, in its many forms, in past and present philosophy, and is the common sense foundation of both Rationalism and Empiricism.
But, while all of this is true, at some distance into Twentieth-Century Anglo-American philosophy it simply was no longer helpful; and it even became bad form to speak of ideas or experiences, especially if done with that abandon manifested in immediately previous writers. It then was language which seemed to stand in the clear. However, it took only a few decades of intensive work by the large and highly trained professional group which philosophy had become to show how very limited the clarity about language was. Once the limitation is realized, one is in a position to see why semantic ascent is philosophically futile. It fails as a general strategy for philosophy simply because language in general, and its elements (“words”) and structures in particular, are no more clear and no more subject to consensus among philosophers—nor are they more clearly positioned in reality in such a way as to provide philosophical access to all else—than are other material objects, along with minds, experiences, and the various ranges of properties and relations, of which we speak before we make the semantic turn. Once again, this should have been expected, since the only philosophically unmuddled sense of “word” is that in which words are simply one subset of physical objects and events.
7. Some disclaimers. It is sometimes useful, in doing philosophy, to talk about linguistic items of one sort or another. In particular, not only is the explicit clarification of terminology often required, but one sometimes can most advantageously specify a philosophical subject or issue, or call attention to a distinction or fact, by mentioning a word or sentence with which it is associated. There is here no intention to deny this. Also, I do not wish to deny that language may “matter to philosophy” in some yet more fundamental sense. Language and “words” are at least one important topic for philosophical investigation, and may be inherently prior in the order of knowledge to other important philosophical topics. However, the manner and the extent in which language may be fundamental to philosophy as a whole is surely at present still a highly speculative matter, to be settled only by the progress of investigations in linguistics and the philosophy of language—and possibly the progress of human knowledge in general. What has thus far happened in these investigations suggests that progress in the philosophy of language will cause the very idea of a linguistic philosophy, whether “ordinary” or “ideal”, to appear as something quaint and incredible. Perhaps it has already done so.