Michael Dummett e la filosofia analitica, di Patrizia Manganaro
In un testo ormai celebre, pubblicato alla fine degli anni Ottanta, il filosofo inglese Michael Dummett indica l’elemento qualificante della filosofia analitica nel «convincimento che una spiegazione filosofica del pensiero possa essere conseguita attraverso una spiegazione filosofica del linguaggio».1
Stando a questa peculiare caratterizzazione, la nascita della filosofia analitica viene a coincidere con la cosiddetta linguistic turn, la «svolta linguistica» del pensiero contemporaneo. Tale svolta, incalza Dummett, pur non essendo stata portata a compimento in modo uniforme e univoco, trova la sua origine più attendibile negli studi sui fondamenti dell’aritmetica del logico e matematico tedesco Gottlob Frege,2 che risalgono al 1884. Nel tentativo di fornire risposta al quesito, inequivocabilmente kantiano, circa la «datità» dei numeri, Frege aveva infatti esposto la tesi, nota come «principio di contestualità», secondo la quale una parola ha significato solo nel contesto di un enunciato.3 Si tratta dunque di un’indagine intorno al modo in cui è possibile fissare il senso di enunciati che contengono termini numerici, cioè di una ricerca epistemologica condotta e chiarita attraverso un paziente lavoro linguistico.
In modo esplicito, Dummett attribuisce a Frege la paternità della filosofia analitica.4 Ciò non toglie il riconoscimento che tutto l’interesse del logico tedesco sia incentrato sui pensieri, e non sugli enunciati che li esprimono: ma è un fatto che l’analisi del pensiero venga da lui effettuata mediante l’analisi del linguaggio. Le nozioni di struttura dell’enunciato e di struttura del pensiero devono dunque essere sviluppate insieme. Scrive Dummett: «L’analisi della struttura logica degli enunciati può essere convertita in un’analisi parallela della struttura dei pensieri, poiché per «struttura logica» si intende una rappresentazione delle relazioni che intercorrono tra le parti dell’enunciato che sia idonea agli scopi del trattamento semantico, o meglio, del trattamento nel quadro di una teoria-del-significato: si tratta dell’analisi sintattica nei cui termini è possibile spiegare il possesso da parte dell’enunciato del significato che ne fa l’espressione di un certo pensiero. Ecco perché Frege poteva avanzare la tesi che la struttura dell’enunciato riflette quella del pensiero».5
Nei Fondamenti dell’aritmetica, tuttavia, non si dà alcuna prova o giustificazione della svolta linguistica: essa viene semplicemente compiuta, come il modo più naturale di procedere nell’indagine filosofica. Secondo Frege, i pensieri sono sempre comunicabili — l’enunciato esprime un pensiero in virtù delle sue proprietà semantiche, dell’essere cioè accertabile come vero o falso6 —; Dummett si spinge forse più oltre, quando afferma che il linguaggio è il veicolo privilegiato del pensiero, e che i pensieri sono sempre «incarnati». Che cosa significa? In accordo con il logico tedesco, per «pensiero» egli non intende il processo psicologico del pensare, bensì il significato condiviso: dunque la comprensione filosofica del pensiero si consegue attraverso — e solo attraverso — la spiegazione filosofica del linguaggio. Come Frege, Dummett ha un’idea logica, dunque non soggettivistica né psicologico-naturalistica, del pensiero: e proprio qui sta, a suo parere, la premessa necessaria e indispensabile per il corretto accostamento di pensiero e linguaggio.7 Da questo punto di vista, la logica è il solido fondamento della filosofia: come ha osservato tra gli altri Franca D’Agostini,8 con ciò Dummett assegna al termine «analitica» il senso che possedeva in origine, quando designava una parte specifica della filosofia, quella «pura» o fondazionale, più che una corrente, un orientamento o uno stile di pensiero. Con la lezione di Frege si è chiarito che l’obiettivo primario e imprescindibile della filosofia è «l’analisi della struttura del pensiero e, in secondo luogo, che lo studio del pensiero deve essere tenuto nettamente distinto dallo studio del processo psicologico del pensare; infine, si è riconosciuto che il solo metodo appropriato per l’analisi del pensiero consiste nell’analisi del linguaggio».9
Come è stato giustamente rilevato,10 Dummett ha tentato una complessa, acuta mediazione tra la tradizione logica che risale a Frege e le provocazioni provenienti dal pensiero del «secondo» Wittgenstein sulla filosofia del linguaggio. Accanto al celebre argomento contro il linguaggio privato, l’austriaco sostiene che la comprensione del linguaggio non è un processo psichico: si tratta di due tesi contro il cosiddetto «mentalismo», entrambe esposte da Wittgenstein nelle Ricerche filosofiche11 e in seguito riprese efficacemente da Dummett. Il significato ha sempre carattere pubblico: l’idea che la comprensione e/o la conoscenza del significato di un enunciato debba manifestarsi nell’uso deriva, per il filosofo inglese, «sia dall’elaborazione dello slogan wittgensteiniano «il significato è l’uso», sia dalla riflessione su una tesi di Frege nota come «principio di contestualità», secondo cui una parola sta per qualcosa solo nel contesto di un enunciato. Questa tesi è giustificata dal fatto che l’enunciato è l’unità minima con cui è possibile compiere un atto linguistico, cioè un atto concreto di comunicazione; del resto, l’acquisizione del significato di una parola avviene osservando esempi del suo uso in enunciati, e la conoscenza del senso di una parola può essere dimostrata solo usandola (cioè mostrando che si è in grado di formulare enunciati in cui essa è usata correttamente). Dunque la tesi della contestualità è strettamente connessa all’idea del carattere pubblico del significato, che a sua volta fa tutt’uno con la tesi della manifestabilità nell’uso».12
In polemica con la configurazione del linguaggio come mero strumento di comunicazione, sostenuta in tempi recenti dall’americano Donald Davidson, Dummett ne ribadisce la funzione di veicolo, e così precisa: «Il linguaggio è sia uno strumento di comunicazione che un veicolo del pensiero; per orientarsi nella filosofia del linguaggio è importante chiederci quale ruolo consideriamo primario».13 Come subito si vedrà, sono le stesse parole che aprono la conferenza Language as our means of Communication, tenuta da Michael Dummett all’Università Lateranense in occasione del Convegno di Studi su Intelligenza e linguaggio. Questioni e prospettive nel dialogo tra scienze e teologia (Roma, 25-26 ottobre 2001). Certamente lo scopo primario del linguaggio è essere strumento di comunicazione significante. Secondo Dummett, non esiste un pensiero «nudo», non accompagnato da parole o da immagini interne. E allora, il veicolo privilegiato ed efficace del pensiero umano è proprio il linguaggio: i pensieri esistono, ma la loro esistenza è «incarnata» nella loro espressione, sì che è possibile affermare che il pensiero vive nella sua espressione linguistica, è ciò che può essere espresso nel linguaggio. Quanto questa non trascurabile implicazione getti un ponte tra la frattura, apparentemente incolmabile, tra «analitici» e «continentali» rimane ancora non opportunamente esplorato.
Language as Our Means of Communication
Language serves two purposes for us: as an instrument of communication and as a vehicle of thought. Which is its primary role? Is it because we can express our thoughts in words that we can use words to convey those thoughts to others? Or is it rather that it is because we can communicate with one another by means of words that we can use words to express our thoughts to ourselves?
Informal disputes frequently arise over whether we “think in words”. Some maintain that they always think in words; others that they do so only sometimes. When two or more people are talking with one another, the process would be misrepresented by supposing that, before speaking, each first conceived a thought in his mind and then put it into the words he uttered. Occasionally, a speaker may pause to consider what to say before making his contribution to the conversation; he may be making up his mind what is the truth of the matter, or he may be pondering how best to express what he already thinks. For the most part, however, there is no such preliminary reflection: each speaker comes straight out with what he has to say. In such a case, the verbal utterance is not a linguistic formulation of a prior thought: the speaker is thinking in words, words that he utters out loud. He is framing his thoughts at the same time as he communicates them. Someone asked a question about some matter about which he has never thought may take a little time to respond. He may not be reflecting on how to express his opinion verbally: he may be reflecting on the topic itself, calling to mind the various considerations bearing on it one way and the other. Chivvied to give a response by a questioner who says, “You must know what you think”, he may reasonably, if comically, reply, “How can I know what I think until I hear what I say?”.
But how is it when we think silently to ourselves? Clearly, we often do then rehearse our thoughts in sentences: but are we, in such a case, thinking for the first time the thoughts expressed by those sentences, or merely thinking how to express our thoughts in words? To say the latter would be to say that we are thinking how to express in words the thoughts we already have. Sometimes, indeed, this is a correct description of what is taking place — for instance, when we have already expressed those thoughts in words, but in words that would be unsuitable for some forthcoming occasion. This may be because, so expressed, our hearers would not readily understand, or because they might be offended or embarrassed. It is plain, nevertheless, that it would often falsify the process of thinking to oneself to describe it as thinking how to express our thoughts in words: we are often in a position in which our inner verbal formulation of the thought is the first time we have ever framed just that thought, one in which we could say to ourselves, "I did not know that I thought that until I said it to myself’.
Thought needs a vehicle: there is no such thing as a bare thought, one not expressed by anything, either by externally observable speech or action, or by internal imagery. Often the vehicle is linguistic; but this need not mean that the thought is formulated in sentences which, if spoken aloud, would adequately convey it to others. I can employ, as a vehicle of my thought, the kind of shorthand that will suffice to record a thought in notes I make for my own use. Another person might be unable to interpret those notes, because they rely on associations peculiar to me. The notes do not fully express the thoughts they record, but adequately, serve to call them to my mind; and in the same way, when I am engaging in a rapid train of thought, the words that go through my head may encapsulate my successive thoughts for me, although they do not fully express them.
A thought can, however, be expressed otherwise than in words, but an outward action other than speech. If I sit up in my chair, turn my head and assume an alert look, I may well be giving expression to the thought, “What was that strange noise?”; if after a moment I shrug my shoulders and return to what I was doing, I may express the thought, “It was probably nothing important”. To ascribe such thoughts to me is not to suggest that those words, or any words, went through my head: my thoughts were sufficiently expressed by my actions. When someone is driving a car or steering a boat, he may have to think very rapidly how to take avoiding action when another car or boat approaches in an unexpected direction. His thought is unlikely to be conducted in words. It is more likely to be conducted in mental imagery superimposed upon the scene he sees with his eyes, even though, later, he can recount his thoughts in language.
The process of thought that goes through the mind of someone driving a car or steering a boat is the kind of thought in which an animal can engage. Animals can often solve quite difficult problems about how to deal with physical situations confronting them; but their thought can bear only on what they can presently perceive, because its vehicle can only be imagined manipulation of their perceptions, just like the thought of the person driving a car or steering a boat. A human being, by contrast, may frequently be struck by a thought having nothing to do with his current activity: he may stop in his tracks, turn round and go back, because he has just remembered something he had forgotten to do or to take with him. An animal cannot behave in this way, because his thoughts cannot float free of his present environment.
Language is thus very often, but not invariably, the vehicle of human thought. One can have any given thought only if one has grasped the concepts involved in that thought: one cannot think, “That is not a river, but a lake”, unless one has the concepts of a river and of a lake. A great many of our concepts we have acquired only through our learning of language; most of these could not have been acquired in any other way. As Wittgenstein remarked, a dog may be expecting his master to come, but he cannot be expecting him to come in an hour’s time, or next week. Our knowledge of language makes us capable of thoughts that we could not have if we had no language, whatever may be their vehicles on a particular occasion.
A language is essentially a communal possession, that is, essentially an instrument for communication between different people, people who share that language. It is a mistake of principle to think of each individual as having a language peculiar to himself — an idiolect — and to regard the primary notion as being that of the idiolect, a common dialect being a set of overlapping idiolects and a language, as we ordinarily think of it, a set of closely resembling dialects. Of course, the word “dialect” is used in different senses. Sometimes it merely means a language that lacks prestige: one not used in the legislature, one in which newspapers are not printed or sermons preached. But a dialect in the other sense, that is, properly so called, is not a language in its own right, but a way of speaking a language, as an accent is a way of pronouncing a language. A foreigner ought to try to use a language as the natives do, following them in vocabulary, syntax and idiom: it is impolite of him knowingly to insist on speaking it in some other way. But when two people are speaking together in the same language, each using his own language, it is usually impolite for either to mimic the other; it gives the impression of mockery. This often applies to the use by one of them of the dialect that the other is using. I am afraid that I cannot illustrate this in Italian, because I do not know it nearly well enough, so an English illustration will have to serve. When a Scotsman and an Englishman are speaking to one another, the Scotsman may use the expression “wee balms”, meaning ‘little children’. He will expect the Englishman to understand him — that much of Scottish dialect is widely known. But he will not expect him to copy him. If the Englishman, in reply, were to speak of ‘wee bairns’, he would give the impression of mocking him, just as he would do if he copied his accent instead of using his own.
An idiolect is not related to a language, or even to a dialect, as a dialect is related to a language. Of course, each individual speaker has his personal linguistic habits. Of the vocabulary which the dictionary lists, he will know and freely use some words, he will understand but hardly ever himself use certain other words, and he will be quite unfamiliar with yet other — probably many other — words. He is likely, moreover to misunderstand and misuse a certain range of words. And the extent of these four categories will vary from individual speaker to individual speaker.
These personal peculiarities of individual speakers do not, however, make of idiolects structures that could subsist of themselves; rather, they are ways that an individual has of speaking the language that he shares with others, and have just been explained by reference to that common language. An individual’s idiolect could not exist without there being a common language of which it was an idiolect: it could not be his private language, unrelated to any other, in which — since by hypothesis he could not use it for communication — he might frame his private thoughts and, perhaps, keep his diary.
In his Philosophical Investigations, Ludwig Wittgenstein gave a famous argument to the effect that there cannot be a private language. For the purpose of this argument, he understood the expression “a private language” in a very restricted way: in this sense, someone has a private language only if he is the sole person who could understand it. The reason for this restriction is that Wittgenstein’s immediate target of attack was a very common conception of our cognitive relation to our own sensations — our manner of thinking about them. One notorious way of expressing this conception is by saying that it is impossible to convey to others what our sensations are like: for this reason, it is maintained, I cannot know whether something that we agree in calling “yellow” does not appear to you as what we should both call “red” appears to me. Another way to express the same conception is to say that words for phenomenal features such as colour have a dual meaning in the minds of all who use them: they have the public meaning which we invoke when we converse with others; and they have the private meanings which for each subject expresses the sensation that he experiences when he perceives something of the given colour. It is because he attaches this private meaning to a colour-word that he is able to use it in accordance with its public meaning. He recognises the colour-sensation to which some object gives rise in him as that which he calls by that colour-name, and uses the colour-name to convey to others what he sees. This is effective because our colour-sensations are systematically correlated with the physical properties of the light that strikes our eyes, but, so far as we shall ever be able to tell, in a way that differs from individual to individual.
According to this idea, the private meaning that one subject attaches to the colour-word cannot in principle be communicated to any other subject; if the subject had a private word with that meaning, he could never explain its sense to anybody else.
I believe, however, that Wittgenstein’s argument suffices to prove the impossibility of a private language in a broader sense, that of a language which only one person does understand. The fact that someone could, or believes that he could, instruct another person in the use of his private language makes no essential difference to the situation, so long as he has not done so. Once he has taught the language to another, and he and the other start to converse in it, then we have a new situation: now it really is a language, even if, so far, with only two speakers. But, while it remained the property of only a single individual, it was not yet a genuine language at all.
Of course, we must take care how we frame this thesis. A code is not a private language, even if known to only one person. If someone knows a common language, but uses, for soliloquy or for his private notebook, certain otherwise non-existent words of his own invention in place of certain words of the common language, he is not employing a private language in the relevant sense, because the use of the words he has invented remains responsible to the use of the words of the common language that they encode. A private language of the kind which Wittgenstein’s argument aims to show cannot exist, is one the meanings of whose words are not given by reference to the meanings of the words of any other language, but are directly conferred on them solely by the individual whose language it is.
The argument turns on the evident principle that the meaning of a word or more complex linguistic expression depends upon there being a right and a wrong concerning its use. This right and wrong relate both to construction and to application. In a common language such as English, a gross grammatical error destroys meaning, if the number-word “three” or the noun “problem” is used as a verb in some sentence, that sentence is, as it stands, meaningless. It may be possible to work out what the speaker is intending to say; but what he has said has no meaning, as he said it. More importantly, even if what a speaker says has a determinate meaning in the language, there must be, if he was making an assertion, an agreed criterion for whether what he said was true. If he was asking a question, there must be an agreed criterion for what would be the right answer; if he was declaring what should be done, there must be an agreed criterion for whether his demand was complied with; similarly if he was giving voice to a wish, making a bet or the like. Very often, it will be far from straightforward to determine the truth or falsity of an assertion: but it must be a matter of common agreement what is to count as a ground for its truth, and what forms of argument for its support or against it are valid. The criteria for the truth of very simple forms of statement, such as “It is snowing” or “The bottle is on the shelf”, must be applicable by quite direct observation.
A private language cannot have objective standards of right and wrong. There is no one to judge whether what is said is right or wrong, whether it is correctly or incorrectly constructed, whether the predicate is or is not true of the object to which it is applied. If a sensation is being identified, there is no standard by which it can be decided whether it was identified correctly or incorrectly, whether the sensation so identified really is or is not the same as the sensation to which that name was originally attached. There is only the individual speaker, who is judge in his own case. As Wittgenstein says, whatever will seem right will be right: and that means that there is not in reality any right or wrong here at all. A private language is no genuine language, because what is of the essence of language, principles that determine whether anything has really been said or not, and other principles that determine whether what has been said is true or false, are lacking.
It follows that there cannot be a language unless it is common to a whole community, a body of speakers. This does not of course imply that a language is dead when only one person who knows it remains: it dies when that last speaker dies. It was still a genuine language even when only one person knew it, because it had been the common possession of a community of speakers. A dead language is still a language, because it was once the common possession of a community of speakers; if it has left behind a literature, we can still discover from this what were the norms that governed it.
Nor does it imply that an utterance in a language is correct provided only that a majority of the linguistic community — most or even all of its speakers — would judge or be inclined to judge that it was correct. There can of course be massive communal error in judging the truth of propositions whose truth-value is not readily determinable: the most obvious example of this is the long adherence of humanity to the geocentric picture of the universe. There can also be a widespread tendency to make erroneous judgements about more particular matters, for instance to think that the thunder follows the lightning instead of merely being heard after the lightning is seen. When we learn a language, we do not merely acquire propensities to speak in certain ways and to judge the truth-value of what is said in certain ways. Rather, we come to grasp certain principles governing our employment of language: principles that determine the formation of sentences out of their component words, and how the meanings of those sentences are derived from the meanings of the words, and principles that govern by what means we are to decide the truth or falsity of a declarative statement. These principles, if they are truly to govern our use of the language, must be the object of common agreement, but their application, though often obvious, is by no means always straightforward: hence the disagreements that so frequently arise between us. It is the principles that we learn to accept which determine whether what is said is correct or incorrect, true or false, not whether speakers at large are disposed to think it one or the other.
From this we must infer that the primary purpose of language is as an instrument of communication. The words and sentences of a language could not have the meanings that they have were the language not used as an instrument of communication, because they would not have those meanings unless it were the common possession of a community of speakers. It is because it serves as an instrument of communication that it can also serve as a vehicle of thought. Language is essentially, not accidentally, a social phenomenon; we have language, and hence are able to have the thoughts we have, only because we are members of a human society.
As we are reminded with wearisome frequency by professors of linguistics, languages change; if they did not, we here should all be speaking Proto-Indo European. The fission of Latin into the many Romance languages took place in historical times. The least interesting reason for linguistic change is the introduction of new words, or new senses for old words, for new kinds of object or newly -discovered objects or kinds of object. But languages change also because the rules or principles governing them come to be flouted. A language must be subject to rules; otherwise its expressions would have no meanings and we could not use it to say one thing rather than another. But the paradox of language lies in the fact that, although it has rules, there is no authority to enforce those rules. The only penalty for failing to abide by the rules of a language is that of failing to convey anything to one’s hearers. If it were not possible to convey one’s meaning at all unless one adhered strictly to the rules governing the language one was speaking, then there would be no linguistic change save that required for making reference to new kinds of object. But this is of course far from being so. When someone utters a grammatically incorrect sentence, or uses a word in a garbled form or in the wrong sense, it is very often possible to guess what mistake he was making, and hence what correctly phrased sentence would have expressed the meaning he was intending to convey. But some hearers will rightly guess at the intended meaning without realising that any mistake had been made. If such hearers repeat the mistake, or other speakers make the same mistake, it will gain currency: if it gains sufficient currency, it will cease to be a mistake. A grammatical rule will have been altered, or a word have acquired a new form or a new sense. The paradox of language is that those who must obey the rules if language are at the same time the rule- givers: if they come to concur in speaking in accordance with systematic violations of the existing rules, those ways of speaking cease to be violations and come to exemplify the new rules that are now in force.
The professors of linguistics who so incessantly remind us that languages change often speak and write as though an utterance bears its meaning on its face, as if its meaning were as observable a feature as the way it is pronounced. They say that a native speaker may express his meaning however he chooses, just because there is no external authority to compel him to speak as the rules that govern the language require. Obviously this is not so. An utterance that is correctly expressed bears the meaning that it does precisely because of the rules that govern the language: those rules determine what each word means and how they combine to confer its meaning on the sentence that is composed of them. It is on the basis of those same rules that it is often possible to guess the meaning that a speaker intended to convey by means of an utterance that he has expressed incorrectly. Such professors of linguistics frequently use their cry, “You can’t stop language from changing”, to insinuate that linguistic change is an irresistible natural force beyond the control of the speakers of the language. Obviously, it is not: if the language changes, it is the speakers who change it. Some changes are beneficial, some malign, particularly those that impoverish the vocabulary of the language by depriving a word of its former meaning and attaching to it a meaning that another word already had. Linguistic change is not irresistible: when a particular change destroys a means of avoiding ambiguity or diminishes the expressive power of the language, it ought to be resisted and sometimes can be successfully resisted.
Two people can communicate with one another by means of language — the only means we have to communicate any but the simplest thoughts and wants — only if there is a language that they both know, at least to some extent. Speaking and listening to the speech of others are conscious activities. They might not be. We can imagine being in a condition of not being able consciously to discriminate the sounds that others, or that we ourselves, made when we spoke: after all, it is a common experience to hear others talking, but to be unable to make out what they are saying. In such a condition it might be that the words that others addressed to us aroused in our minds the thoughts they were expressing, without our being able consciously to articulate their utterances into their component words, and that likewise, when we spoke ourselves, we were aware of the thoughts we were expressing but unaware in any detail of the sounds by means of which we were expressing them. After all, we can remember the gist of what someone said, that is, its content, without being able to recall the words he used: might it not be like that for us at the very time of hearing someone speak? It is interesting to consider whether this fantasy is a genuine possibility or not; but it is at least a fantasy. Our knowledge of a language is a conscious knowledge, and our use of it a conscious activity.
In what does someone’s knowledge of a language consist? What are the rules that determine what meanings the words and sentences of a language have? The explanatory priority must be given to the meanings of sentences: as the famous statement of Gottlob Frege has it, it is only in the context of a sentence that a word has a meaning. The meaning of a word is its contribution to the meaning of a sentence of which it is part: we can explain the meaning of a word only as contributing to the meaning of a sentence in which it occurs, and hence only if we have a conception of what, in general, a sentence may mean. Sentences of course serve many different purposes. Some may be used to state what is the case; others may be used to enquire what is the case, to express a wish that something were the case, to demand or to pray that something be the case. Of these, their use to assert what is the case may be considered primary: we cannot explain the other uses of sentences unless we understand what is meant by saying that a question has been answered, that a wish has been fulfilled, that a demand has been complied with, that a prayer has been answered. So the central problem is to account for the meanings of declarative sentences that can be used to make assertions.
The general explanation of the meanings of declarative sentences which is by far the most popular among analytical philosophers is the truth-conditional one first systematically proposed by Frege in the first volume of his Basic Laws of Arithmetic and endorsed by Ludwig Wittgenstein in his celebrated early book Tractatus Logico Philosophicus. This is the theory that the meaning of a declarative sentence is given as, or determined by, the condition for an assertion made by it on any particular occasion to be true. On this theory, we must explain a speaker’s understanding of a declarative sentence as consisting in his knowledge of the condition for an assertoric utterance of it on any particular occasion to be true. The difficulty with this theory is twofold: to explain how we learn to associate such conditions with the sentences of the language; and to explain in what our grasp of those conditions consists. We are often in a position to establish one or another statement as true, or as highly probable, by observation or by reasoning based upon observation. That is what we learn when we learn our language: we learn what is to count as establishing the statements that can be made in the language as true or as probable. We learn how to report our observations in language; and we learn what forms of reasoning are valid. Obviously, there are many statements that we understand perfectly well, but of which we have no reason to judge true or false. It is natural to take our understanding of them to reside in our ability to judge whether something would establish or render probable their truth. But this would not be sufficient from the standpoint of the truth-conditional theory of meaning: to understand the statement, we must, on this theory, know the condition that must hold for it to be true, whether or not this is a condition that we can recognise as obtaining, whenever it obtains. How are we supposed to have come by this knowledge, simply from learning to use our language? It is certainly part of it that we should have acquired the ability to recognise when the truth of a given statement has been established or shown to be probable; but it is hardly part of learning to use the language to come to grasp the condition for it to be true, independently of there being any ground for us to assert it.
An ability to recognise that the truth of a statement has been established or shown to be probable is manifested in a speaker’s employment of the language. He treats the statement as true, either by himself asserting it or else by accepting inferences drawn from it, drawing such inferences himself or in his other conduct acting upon the truth of the statement. In the case of a statement whose truth has not so far been established or shown to be probable, a speaker may well still be able to show himself capable of recognising that its truth was established or shown to be probable, if this should later occur; he would thus show that he satisfied this criterion for understanding the sentence used to make the statement. He could show it by what he said in this regard about certain hypothetical circumstances; but he could also show it by his employment in other sentences of the various words that make up the sentence in question. An account of understanding in terms of a speaker’s ability to recognise what shows a statement to be true does not rest on attributing to the speaker any inner mental processes accompanying his speech or his listening to the speech of others; it rests only on his observable use of the language. It is this — his observable use of the language — which provides the ordinary basis on which we judge that he does or does not understand this or that expression.
The truth-conditional account of understanding, by contrast, involves ascribing to the speaker, for every statement he understands, a conception of the condition required for it to be true. This is not just a conception of a state of affairs in which he could recognise the truth of the statement as having been established, for it is essential to the truth-conditional theory that the statement may be true independently of our means of coming to know that it is true: what anyone must have, on this theory, in order to understand the statement, is a grasp of the condition for it to be true, whether or not we are in a position to recognise it to be true. When we do not know whether a statement is or is not true, we may know an effective method of finding out whether it is true. But often we do not know of any such effective method: we simply have to leave it open whether or not we shall hit upon reasons to think it true or reasons to think it false, without any assurance that we shall ever do either. Now truth conditional theorists normally accept classical logic, and, with it, the principle of bivalence. This is the principle, namely, that every statement is determinately either true or false, provided, at least, that it is not ambiguous and that its meaning is definite rather than vague. Hence, even if we do not know any effective method for finding out whether a given statement is true or false, those who accept this principle must presume that, independently of our knowledge, it is either true or false. It may be true, even it we shall never nave any reason to think it so; equally, it may be raise, even though we shall never have any reason to deny it.
The truth-conditional theory of understanding therefore requires the attribution to a speaker, for him to understand a language, of an ability to frame, for every statement he understands, an inner mental conception of a state of affairs that would render that statement true. What form does his grasp of this inner mental conception take? The proponent of a truth-conditional theory of meaning and understanding cannot say that it consists in an ability to formulate in words the condition for the truth of any given statement, or else his explanation would be circular: he would be explaining the understanding of language in terms of a prior knowledge of language. His grasp of the truth-condition — rather than his knowledge that it is the condition for the truth of the given statement — must, rather, be effected without the help of language. That we have any such languageless grasp of the condition for the truth of every statement we understand is highly implausible: what we have is an understanding of our language. As Wittgenstein said, when we say something, knowing what we mean, we do not have meanings in our heads running along beside the words: the words we utter are the vehicle of our thought. The truth-conditional theory of understanding makes it uncertain whether anyone else understands what we say as we do. On this theory, we cannot tell for sure whether he understands from what he says or how he responds to what we say. We could tell for sure only if we could look into his mind to see if he had the same conception as we do of what would render our statements true; for he might without having that conception know all that was needed to use the language correctly.
The logical positivists did not subscribe to the truth-conditional theory of meaning; instead, they advocated the verificationist theory according to which the meaning of a statement consists in the manner in which it might be verified. So far, this agrees with the account of understanding as the ability to recognise when the truth of the statement has been established. So far, in my opinion, the positivists were not in error. They went wrong, however, in their narrow conception of what the verification of any statement must consist in. They thought that the verification of any empirical statement must consist in a suitable sequence of sense-impressions. The meanings of statements of logic and mathematics had then to be explained in a completely different way. This was an atomistic theory of meaning: it allowed it as in principle possible that someone should understand any one sentence of a language without understanding any other of its sentences, nor, therefore, any of the words in that one sentence. In a sense, indeed, a tourist armed with a phrase-book can associate the correct meaning with a sentence of a language he does not know: but, while he may be said to know the meaning of the sentence, he does not in the true sense understand it, because he does not grasp the meanings of the component words nor how they are put together to form a sentence with that overall meaning. Moreover, the conception of verification as receiving an appropriate sequence of sense-impressions bears little relation to our actual procedures for recognising the truth of a statement: we should be hard put to it to say what sense-impressions led us to endorse the statement “The Government is far less popular than it was just after the election”.
Where the logical positivists went astray was in failing to comprehend the interconnectedness of language. Many expressions cannot be understood unless other expressions are first understood. The positivist conception of verification does not tally with what in practice we take as establishing the truth of a statement, either conclusively or with probability. In the normal case, this will not consist in unassisted sensory observation, but will involve inferential reasoning, and will thus depend on our understanding of the other statements that form part of the argument. Sentences do not, as the positivists thought, fall into a dichotomy of empirical ones and mathematical ones. Rather, there is a scale at one end of which stand sentences capable of being used as reports of observation, and may therefore be verified without appeal to any kind of reasoning, and at the other end of which stand mathematical theorems, established by deductive reasoning unaided by observation. Most of the statements we make occupy intermediate positions on this scale: the manner in which we establish their truth involves both observation and inferential argument. A plausible theory which treats the meanings of sentences as given by what is required to establish statements made by means of them as true will explain this in terms of the way in which we actually so establish them, and not as the occurrence of sequences of sense-impressions. Because the word “venficationism” has misleading associations with the mistaken theories of the logical positivists, I should prefer to call such a theory ‘justificationist’ rather than ‘verificationist’: it explains the meaning of a statement in terms of what would justify asserting it.
In particular, the meaning of a theological statement, such as “God knows the secret impulses of men”, “God forgives the sins of the penitent” or the fundamental “There is a God”, will be given, on a justificationist theory of meaning, by whatever arguments in favour of them we correctly recognise as cogent. When acceptance of any particular such statement rests on revelation, this will be a feature of its meaning, together with the grounds for believing it to -have been revealed. Of course, these are propositions over which there is and has been great dispute: but those who argue for and against them do so on the assumption that such arguments are objectively either valid or fallacious: and their validity or invalidity turn on our considered judgements of their soundness or unsoundness. On the truth-conditional theory of meaning, by contrast, the meaning of a theological statement, like that of any other statement, is given by the condition for it to be true. In this context, this phrase comes very close to being meaningless: what can it mean to say that someone knows the condition for the statement “God exists” to be true? The justificationist theory connects the meanings of our statements, including theological ones, with our actual linguistic practice, not with a prelinguistic capacity to envisage confirmatory states of affairs.
Relazione svolta nell’ambito di un Seminario organizzato dal SEFIR (Scienza e Filosofia nell’Interpretazione del Reale), area internazionale di ricerca della Pontificia Università Lateranense, dal titolo Intelligenza e Linguaggio (25-27 ottobre 2001). Ringraziamo il prof. Piero Coda, Responsabile dell’area di ricerca, per l’autorizzazione ad anticipare in questa sede il testo. (E.B.)
M. Dummett, Alle origini della filosofia analitica, tr. it. di E. Picardi, Il Mulino, Bologna 1990, p. 11. ↩︎
Cfr G. Frege, Fondamenti dell’aritmetica, tr. it. in: Logica e aritmetica, a cura di C. Mangione, Boringhieri, Torino 1965. ↩︎
Nel § 106 dei Fondamenti dell’aritmetica Frege enuncia il “principio di contestualità” come segue: «Abbiamo formulato il principio che il significato di una parola non vada spiegato isolatamente, bensì nel contesto di un enunciato; solo osservando questa massima è possibile, io credo, evitare la concezione fisica del numero senza cadere in quella psicologica». ↩︎
Per la verità, con un’espressione scherzosa Dummett lo definisce «il nonno della filosofia analitica», e così continua: «Frege offrì la prima spiegazione plausibile nella storia della filosofia di che cosa sono i pensieri e di come concepire il significato degli enunciati e delle parole che li compongono. Coloro che si trovarono spinti ad analizzare il pensiero attraverso l’analisi del significato linguistico non ebbero altra scelta che costruire sulle fondamenta che egli aveva gettato» (M. Dummett, Alle origini della filosofia analitica, cit., p. 20). ↩︎
M. Dummett, La base logica della metafisica, tr. it. di E. Picardi, Il Mulino, Bologna 1996, p. 16. ↩︎
Come è noto, la teoria fregeana del senso si basa sulla nozione di verità: cfr G. Frege, Il pensiero, tr. it. in: Ricerche logiche, con un’Introduzione di M. Dummett, Guerini e Associati, Milano 1987, p. 68. Il nesso tra senso e verità fu successivamente ripreso e chiarito da L. Wittgenstein nel Tractatus logico-philosophicus: un’importante delineazione dell’influsso di Frege su Wittgenstein si trova in M. Dummett, Frege e Wittgenstein, tr. it. di L. Anselmi, in: AA. VV., Capire Wittgenstein, a cura di M. Andronico, D. Marconi, C. Penzo, Marietti, Genova 1988, pp. 229-240. ↩︎
Cfr M. Dummett, Filosofia del linguaggio. Saggio su Frege, tr. it. di C. Penco, Marietti, Casale Monferrato 1983; Id., Frege and Other Philosophers, Clarendon Press, Oxford 1991. ↩︎
Cfr F. D’Agostini, Breve storia della filosofia del Novecento, Einaudi, Torino 1999, p. 220. L’Autrice nota che, in questo senso, sono a rigore «analitici» anche alcuni filosofi «continentali» come Heidegger e Husserl, e conclude: «Tra filosofi analitici e filosofi continentali si crea allora una linea trasversale, in cui la differenza è data solo dall’uso della logica formale, attivo e inaggirabile nella filosofia analitica, e trascurato o considerato improprio in quella continentale» (Ibid.). Della D’Agostini si possono utilmente consultare anche Filosofia analitica. Analizzare, tradurre, interpretare, Paravia, Torino 1997, e Analitici e continentali, Mondadori, Milano 1997. ↩︎
M. Dummett, La verità e altri enigmi, tr. it., Il Saggiatore, Milano 1986, p. 66. ↩︎
Cfr D. Marconi, Dalla filosofia della logica alla filosofia del linguaggio, in: Storia della Filosofia, a cura di P. Rossi e C.A. Viano, Vol. 6, Tomo II, Laterza, Roma-Bari 1999, p. 725 ss. ↩︎
L. Wittgenstein, Ricerche filosofiche, tr. it. di R. Piovesan e M. Trinchero, Einaudi, Torino 1995. ↩︎
D. Marconi, Dalla filosofia della logica alla filosofia del linguaggio, cit., p. 726. ↩︎
M. Dummett, Una graziosa confusione di epitaffi. Alcune note su Davidson e Hacking, tr. it. di L. Perissinotto, in: D. Davidson, I. Hacking, M. Dummett, Linguaggio e interpretazione. Una disputa filosofica, Unicopoli, Milano 1993, p. 134. ↩︎