Home > Philosophy > 1997 After Postmodernism Conference > Schneider (backgrounds)
I would like to begin with a very rough outline of what I plan to do in this lecture, and I will draw this outline by way of commenting on the words I have used to formulate my title(1). When I speak of the calculus-side of language, I have two things in mind: First, the rules of grammar books that tell us the correct forms of particular languages like 'I go', 'he goes', 'we go', etc. And, second, the logical rules implicit in language. The first kind of rules can be called 'syntactical' in a certain sense, because they regulate the formulation of sentences but do not guarantee that they will be meaningful. Their relation to a developed, written language can be taken as clear enough in the present context, although, of course, it has problems of its own. But how logical rules (and the possibility of stating them in a calculus) relate to language is a question of direct philosophical concern.
Since we try to speak logically in our argumentations, some such rules seem to exist 'behind' or 'beneath' the rules of particular grammars. Although they are not formulated in the (traditional) grammar books, they seem to guide our understanding; they seem to be located on a semantic level, and to follow them seems to guarantee meaning. In an explicit form we know them as rules of an especially invented logical 'language', like Gottlob Frege's 'concept-script'. It is in the realm of logic that the idea of a calculus-side of language (in the stricter sense) originated, and important theorists of modern linguistics like Noam Chomsky have been influenced by ideas and procedures originally developed in this domain. It is one of the questions that I am interested in, just how much of language can be captured by spelling out a fixed structure of the kind known from the logical calculi. They, by the way, are called 'syntactical' in a second sense: Although to follow them should guarantee meaning, they are formulated in such a way that following them can be accomplished by purely formal, schematic steps.
My second term, 'imagination', stands for the creative side of language, e.g. the understanding of metaphor and the grasp of the contextual circumstances of an utterance that determine its meaning. This side of language is for me of particular interest insofar as it is impossible or pointless to capture it in terms of rules. Also on this side belongs the meaningful violation of rules as an act of free choice, and the understanding of unusual ('unruled') utterances about which "we have not settled anything", to use a formulation of Wittgenstein's(2). So we have a fixed rule-structure on the one hand, and free activities, not captured by (formulated) rules and sometimes even violatig existing rule-formulations, on the other.
My basic claim now is that an adequate understanding of language has to consider both sides. It should neither despise the structural side as the 'dead bones' of language, as the playground of mindless technocrats, as some practioners of hermeneutic philosophy tended to do thirty years ago. Nor should it neglect the side of the imagination as an epistemologically irrelevant luxury, as something that would be taken care of after the hard work of establishing a theory of meaning had been completed, or as something that sooner or later will also be treated by rules. Such a view had sometimes been taken by analytic philosophers. As the term 'interplay' is meant to express, I want to argue that the two sides of language interact on all levels. The imaginative side of free activity is necessary to establish rules in the first place, and it is constantly needed to act meaningfully according to rules, once they are formulated. It is not possible to capture the whole of linguistic competence by a network of rules. On the other hand, fixed structures are necessary to develop the typically linguistic form of creative activity. Correctly understood, structure enhances freedom and does not repress it.
In terms of some prominent names, this means that I disagree with Richard Rorty in that I think there can be an epistemologically interesting philosophy of language(3); and I disagree with the more recent writings of Donald Davidson, in that I think that the Fregean account of language did point to something essential. It is an overstatement, I think, when Davidson says that language, as analytical philosophy saw it, simply does not exist(4). On the other hand I disagree with Michael Dummett when he says that a theory of meaning has to take the form of an axiomatic-deductive theory(5). And I also disagree with him when he says that Wittgenstein's refusal to acknowledge the general concept of a proposition forces us to draw the absurd conclusion that there is no systematic connection between our understanding of the indefinitely many sentences of our native language(6). Dummett thinks that if Frege's ideas about a systematic theory of meaning are incorrect, we must have learnt our sentences one by one. In opposition to this indeed absurd conclusion, I would like to suggest that we need a broader, less formal understanding of the types of systematic connections that can exist between sentences. - So much as a very rough outline. I now turn to the first aspect, the calculus-side of language.
It is an old philosophical idea that the logical or conceptual structure of what we try to communicate by uttering a sentence of ordinary language is sometimes different from the grammatical, the linguistic structure of the sentence used. The 'structure of thought', as Gottlob Frege might have said, the structure of meaning, differs from the grammatical structure of any particular natural language. This seems to be obvious in cases in which one and the same grammatical structure can express two logically distinct relations, as e.g. in the two sentences:
(1) Your unmarried millionaire's daughter is waiting in the lobby.
(2) Your alleged millionaire's daughter is waiting in the lobby.
In sentence (1) we have a person who has the two potentially attractive properties of being a millionaire's daughter and of being unmarried, but in sentence (2) the person reported to be waiting in the lobby is not a millionaire's daughter who happens to have the additional (possibly irrelevant) property of 'being alleged'.
Since a speaker of English readily recognizes that the two sentences are in structural agreement only on the grammatical but not on the logical or conceptual level, and since she understands the logical difference perfectly well, it seems natural to say that she recognizes a second structure, in addition to the grammatical one, namely, the logical structure. Of this second structure we say metaphorically that it is located 'beneath the surface' or 'behind' the grammatical structure. To say of the speaker that she perceives the logical structure means in the case of the example that she recognizes that the word 'alleged' modifies the phrase 'millionaire's daughter' in a different way than does the word 'unmarried'. So to understand the conceptual structures of sentences seems to mean: To perceive correctly the ways in which their conceptual elements belong together to make a whole, namely, the meaning of the sentence.
Given this description of a quite familiar experience with ordinary language, it seems to be a convincing program to bring this hidden structure to the surface, to spell it out, to make it explicit. This is the basic idea of analytic philosophy: What is called the 'analysis' of a sentence should exhibit its logical structure, the structure of its content, whatever misleading peculiarities the actually used formulation may show, owing to the non-logical properties of the particular natural language used. And such an analysis should be possible of all meaningful sentences. For that reason it should be possible to give a complete inventory of types of meaningful elements and types of possible combinations of them.
The most perfect explication of the totality of possible conceptual structures would be a calculus of the type that Frege had developed for his 'concept script': The realm of the structurally meaningful, i.e. the structural characteristics of all meaningful thoughts, would be brought into the form of explicit rules that give recursive definitions of indefinitely many expressions. These rules would serve as a 'logical grammar' with the help of which the form of all meaningful utterances (in this artificial language which is taken to mirror the meaningful thoughts) would be specified. Whatever endless empirical research will be necessary for the specification of the content-side of such a 'conceptual notation', the structure of possible meanings would have been made explicit by philosophers such as Frege once and for all. It would have been laid down in the form of fixed rules for a medium independent of all the historical accidents that have shaped the many different natural languages.
According to this picture, the epistemologically interesting side of language structure (in the singular) would have been captured completely by such a calculus, once this goal would have been accomplished. We would keep in mind, of course, that as far as the many different natural languages are concerned, there would be another side of our linguistic competence: The imaginative side, the side of free activities of the speakers and hearers, which is prominent in metaphors and poetry. But when we imagine that one day the logical properties of language would have been exhaustively spelled out as a fixed structure of rules, this imaginative side will be of no epistemological interest. Admittedly, it will play a role on the content-side even in science, e.g. when new words are invented for newly discovered states of affairs. And since to some degree the unanalysed natural languages will probably remain in use for practical reasons, there will even in this area remain what Frege had described as the unwelcome necessity of guessing, owing to the implicitness and vagueness of natural languages. But as far as language structure is concerned, complete expliciteness could be attained whenever it would be necessary. In this spirit Rudolf Carnap wrote in his Intellectual Autobiography about the way he worked at the time around 1919, after having studied Principia Mathematica by Russell and Whitehead:
"When I considered a concept or a proposition occurring in a scientific or philosophical discussion, I thought that I understood it clearly only if I felt that I could express it, if I wanted to, in symbolic language."(7)
What a neat and inspiring picture of rationality! Where then are the points at which authors like the later Wittgenstein of the 'Philosophical Investigations' disagree? There are two decisive questions to ask at this point, in order to decide on the validity of the picture:
1. What is it that the philosopher, in his endeavour to make explicit the logical structure of language, looks at as the point of comparison? To what does the symbolic, logical language answer; what does it depict; what does it mean to say that it exhibits the conceptual categories of expressions and the conceptual relationships between them, in contradistinction to what philosophers like to call the merely grammatical ones?
2. Does it make sense (and if so, what does it mean) to say that a complete characterization of logical categories and relations has been formulated?
Turning to the first question, we can note that Frege believed for a long time that his concept-script would mirror the structures of a pure realm of thought. E.g. the distinction between saturated and unsaturated expressions was taken by him to mirror a distinctness of saturated and unsaturated 'thought-elements', and he took it as a discovery that in the realm of logic always two different kinds of elements are needed to form a new whole. But for a number of reasons he changed his mind, or at least he took a serious step to do so. One of these reasons had always been inherent in his conception; it is his conviction that one and the same thought can be expressed in very different formulas of his concept script. This means that the thought itself cannot directly demand one and only one way of adequate representation, and this in turn invites the idea that not all properties of concept-script expressions can be explained by reference to corresponding properties in the realm of pure thought. So Frege later spoke of 'unsaturatedness' as a property of signs. He even mentions the use of signs as the place where the unsaturatedness really belongs(8). And once this is acknowledged, we can see retrospectively that in all his earlier writings he had appealed to our knowledge of the use of natural-language expressions when he was trying to hint at what he thought of as an idependantly existing realm of thought.
He had acknowledged all along that in order to grasp something in the realm of immaterial thought, we as human beings would need language, we cannot do without the concrete, material signs. But only late in his career did he contemplate the idea that language might not be like a dirty piece of glass, disturbing the clear view at something behind it, but that it might be a 'medium' in the sense of being an instrument. With help of a musical instrument e.g. one can produce music without the idea going with it that the music had existed before its production. So possibly this analogy gives a better understanding of the relationship between language and thought than the analogy to looking through an unclean piece of glass: With help of language one can develop thoughts, but these may not have existed independently before they were developed in the medium of language.
A good part of Wittgenstein's efforts in the 'Philosophical Investigations' can now be perceived as leading just one step further in the direction indicated: Wittgenstein wants to show that in developing something like the concept-script, the philosopher does not answer to anything outside language. Logic does not depict the structure of pure thought, the world of concepts and conceptual relations, or the structure of reality itself. Wittgenstein encourages us to give up the idea of a world of thought behind language. Instead, we have many languages (logical languages among them) on the one hand, and the uses of expressions of these languages in our practical world on the other.
It follows that what appears to be the most basic form of sentencehood, Frege's concatenation of concept-expression and object-expression, is itself nothing but a particular sentence, something very ordinary. Provocatively Wittgenstein says that Frege's distinction between object and concept is nothing but the distinction between a table and its colour(9). And this particular form has been taken by us (i.e. by logicians such as Frege or the early Wittgenstein) as the general form of representation. So the structural properties of the concept script are not mirroring the relationships in a sublime realm of thought. Their generality is not mirroring something general that is given, but it is a result of our activity of projection; it is produced; it is a decision of the logicians to use this form for all non-truthfunctional propositional contents.
But granting that we no longer believe in a pure realm of thought, or in a 'mirror of nature', to use Richard Rorty's expression, could we not still have excellent reasons for this kind of projecting one particular form of expression on the widest variety of contents? I think that this is a question worth pursuing; certainly a logical language has its specific merits. But once its function as a mirror of a pure realm of thought is gone, it is clear that these merits must be the specific advantages of one language as compared to another one, for clearly given purposes. If, e.g., our purpose is to distinguish valid from invalid conclusions of a particular type, and to develop a typology of the valid ones, it can be helpful to demand of all expressions considered that they be of particular forms, like 'All As are Bs'; 'Some As are Bs'; 'No A is a B'; etc. Frege's distinction between saturated and unsaturated expressions serves related purposes. But once their special relation to the world of thought is no longer maintained, these purposes cannot been taken for granted by a simple appeal to our interest in truth and meaning. They have to be spelled out and critically evaluated.
When above I discussed the difference between two sentences containing the phrase 'millionaire's daughter', I made use of the familiar metaphor that speaks of the conceptual structures as 'hidden' in the expressions of ordinary language. According to this way of expression, it seemed as if the hearer would look at a 'second structure' and would be able to grasp it completely, in all conceptually relevant aspects, since she can comment on all conceptually relevant differences between two grammatically similar sentences. I now assume that this structure does not exist as something fixed and ready behind the expressions. This means that now a different account is needed of what the hearer grasps when she correctly understands the conceptual relations. What is she doing, if it is not a grasping of the structure of a thought? She does have access to the new content, but how is this possible?
Postponing this problem for a moment, I first want to ask another question: Given that a logical language is just a language among other languages, can we expect to construct it in such a way that the result shows the totality of all the conceptual relations between its (appropriately categorized) expressions? This was the second of the two decisive questions formulated above: Does it make sense (and if so, what does it mean) to say that a complete characterization of conceptual categories and relations will one day have been formulated with help of a 'logical language'? Could a language be developed and characterized in terms of rules that would give an exhaustive structural characterization of the realm of the meaningful? If the answer would be positive, such a result could very well be called a realization of Frege's project, even if it would no longer be interpreted as exhibiting an independently existing realm of thought.
To this second question Wittgenstein's answer is also a negative one. He argues that it is impossible to characterize all the conceptual properties, all the epistemologically relevant distinctions of kinds of expressions, by means of a logical grammar in the sense of formation rules for categorically specified expressions. The project of constructing such a language in which all conceptual differences are correctly mirrored by syntactical differences is hopeless, according to Wittgenstein. His turn to the 'use' of expressions, his neglect of the formal and structural aspects of language, regrettable and misleading as it sometimes is, has deep systematic reasons which I would now like to explore.
As is well known, Wittgenstein begins his 'Philosophical Investigations' with a quote from St. Augustine and he develops his own thoughts as criticism of St.Augustine's picture of language. One of the first points of disagreement is expressed in his sentence "Augustine does not speak of there being any difference between kinds of word."(10) Wittgenstein goes on by first giving a sketch of a language for which St.Augustine's description would be true, and by then engaging in a process that in the 'Blue Book' he describes as "gradually adding new forms"(11). You will remember that in addition to simple predicates like 'slab' Wittgenstein introduces numerals, proper names and the words 'here' and 'there'. So he himself does speak of 'differences between kinds of word' and says that these are differences in function (avoiding in this way an appeal to a realm of thought). With Frege's project in mind we expect him to aim at a complete list of categories of expressions and at an explicit formulation of the rules that distinguish admissable from non-admissable concatenations of expressions belonging to the specified kinds.
But this expectation is frustrated. Wittgenstein relativizes the concept 'kind of word' when he says "But how we group words into kinds will depend on the aim of the classification, - and on our own inclination." (PI § 17) And a few paragraphs later we read "There are countless ... different kinds of use of what we call 'symbols', 'words', 'sentences'". (§ 23) If this were unconditionally true, the notion of a 'kind of word' would be useless; there could not even be a grammar in the traditional sense of the term, because rules of inflection or sentence formation could not be formulated; they could not say words of what kind would have what sort of ending or can be put together in what order.
On closer investigation it turns out that Wittgenstein uses the expression 'kind of word' in a quite unusual way, closely parallel to his unusual use of the word 'grammar'. In accord with the old philosophical project that all conceptual properties should be made visible by a 'logical grammar', he demands that a complete specification of the 'logical form' of a given expression must be a specification of all the rules that govern its meaningful use. And it turns out that some of the 'rules' that have to be taken into account here are very different from traditional grammatical rules for natural languages as well as from logical rules for a language like Frege's concept script. And therefore they cannot simply be added to a grammar book. Later Wittgenstein does no longer use the expression 'logical form' for what he has in mind, but the expression 'grammar'. This is an unfortunate terminological choice, because it does not clearly indicate that he does not have in mind 'forms' in any sense, but something we do with the help of forms, something that cannot be captured by additional formal properties without again opening a domain of non-formal activity, and so on, ad infinitum.
Unfortunately, Wittgenstein most of the time is very unclear about the fact that form-related rules in the traditional grammatical sense do form a part of what he calls 'grammar'. He so much stresses the different uses that all living languages make of different forms, that the important role of there being forms at all sometimes seems to be forgotten.
I cannot reproduce here the many examples and subtle arguments with the help of which Wittgenstein develops and makes plausible his view; instead I will try to give a systematic sketch of his result. As in the case of object- and concept-expressions Wittgenstein's step by step development of ever more complicated language games aims at showing that all forms of language arise out of particular language-games, tied at first to quite particular purposes. Otherwise the concept of a person 'meaning something' with an utterance would be unintelligible. Some of these purposes might be found in all human cultures, like perhaps sorting out the rotten fruit from the good ones, and this may lead to our familiar object-concept form which is so useful in handeling 'middle size dry goods', as Charles Taylor ironically formulates. Still, even the distinction between object and concept starts out as a quite specific distinction, having its place in a quite specific language game. Its generality is the result of our projection; we are the ones who make it general.
The next important step that the development of Wittgenstein's model language games is meant to show is that once a structure has been developed, it can be projected (and typically is projected) into fields of use that are different from the field it originated in. We not only say 'the table is brown' but also 'Peter is sad' or 'the meeting is closed'. Such processes of projection are well known in the lexical case. That we are able to go on meaningfully in the use of a term, i.e. to apply it with communicative success to cases that are not repetitions of the original situations and often cannot have been anticipated by the community of language users, is a basic fact about our linguistic competence. What I want to stress here is that not only words, but also structures are projected, and that this also is something very familiar to all language users. To characterize this projectability of structures, I would like to revive a term that Eric Stenius had coined thirty-five years ago in his book about Wittgenstein's 'Tractatus': He speaks of 'syntactical metaphors' and describes a sentence containing such a metaphor with the words: "Its sense does not have the form suggested by its logical syntax, but nevertheless this syntax seems to be the best syntax we can give it."(12) As an example Stenius cites Frege's famous sentence 'The concept >horse< is not a concept'. When we remember Frege's troubles in explaining why he could not avoid this extremely inintuitive, seemingly contradictory expression for what he wanted to say, and when we take seriously Stenius' remark that a better syntax cannot be given in cases like this one, we can read these two comments as hints on why the conceptual relations involved here can be explained, can be commented on, but cannot easily be captured by means of forms and structures.
To give a clearer idea of what is involved here, I would like to add another, simpler example which is inspired by discussions in Frege's and in Wittgenstein's writings, but is (in the form I use it) not actually found there. Imagine a simple language game of the kind we find at the very beginning of the 'Philosophical Investigations': Building material is ordered with the help of predicate expressions; other predicate expressions are available to classify pieces according to their colour or size; the numerals have been learned in such a way that it is possible to give the order 'five bricks', e.g.; there are expressions like 'this slab', serving as expressions for objects (so the language game has a particular version of the object-concept-form, showing in a sentence like e.g. 'these slabs are broken'), and the step from orders to statements has been made. It is a part of this little language game that the numerals have so far only been used for orders or statements of the form 'five slabs', or 'four columns', i.e. a numeral has never appeared at the predicate position (the position of concept-expressions). And now imagine that a speaker 'misuses' an expression of the category of the numerals by putting it at the predicate-position and uttering 'these slabs are five'. He metaphorically projects the form 'these A are B' (where 'A' and 'B' mark the positions of predicate expressions) to a realm where it has not yet been used. From the point of view of the rules that had defined the game so far, this step is a violation of rules: The speaker uses the form of classifying objects for a speech act that is not a case of classifying, and he uses a count-word to form an expression that is not expressing an act of counting (but, as we may say, the 'result' of such an act). So undoubtedly there is a violation of rules. But on the other hand we can be pretty sure that the linguistic community will understand the utterance. A hearer will find this projective step, unusual as it is, no more difficult to follow than the projective step of using the word 'slab' not only for a piece of building material, but, say, for a slice of bread.
We know that Frege (in his "Foundations of Arithmetic") took a more conservative view. He argued that since a sentence of the form 'these slabs are five' expresses a truth claim, we have to find out which is the object that the expression 'five' says something about, and he works on a solution that shows that it is a concept that such a sentence says something about. But from a Wittgensteinian perspective it is unnecessary to search for such an object. Wittgenstein has given up the idea that all cases of truth must necessarily be cases of an object falling under a concept (or, as Frege proposes, a concept of first order 'falling in' a second order concept). For Wittgenstein, it is enough to see and understand the projective step that puts numerals in predicate-positions (or a concept-word in subject-position, like in Frege's '>horse< is not a concept'). And once we see that it is the projection that has to be perceived and understood, not some hidden conceptual structure behind the utterance, we see that language is full of these 'syntactic metaphors'. To cite just one more example, I would like to mention Donald Davidson's analysis of sentences like 'he buttered his toast slowly'. Davidson asks: What object is classified by the adverb 'slowly'? And he introduces 'events' as extra entities, that logically have to be posited in addition to the persons and pieces of toast. According to this way of thinking, we must have an 'object' suitable to be classified by the word 'slowly'; otherwise we would not really understand the additional truth claim in the step from 'he buttered his toast' to 'he buttered his toast slowly'(13).
I hope that it is clear now that not the search for the general structures of thought or the general logic of all possible speech acts leads to this conclusion, but the decision for one particular form of representation. Only someone who demands that all truth claims must appear in the form 'object x falls under the concept F' must search for an object to explain how an adverb like 'slowly' works. No doubt there are contexts (like the need for a formal or 'mechanical' drawing of logical consequences from a set of premisses) in which it is useful and even necessary to decide for that particular 'form of representation' that has the advantage of expressing every single truth claim by an extra, additional sentence. But these are particular contexts; it is not the case that Frege's concept script and its modern extensions have shown us the general structures of the domain of all meaningful thoughts or speech acts.
I would now like to turn more explicitely to the interplay of fixed structure and free activity. In particular I want to look at the reasons why it is philosophically pointless (if not altogether impossible) to 'make explicit' the free-activity side of language by spelling out more structure.
Could not someone respond to the picture drawn so far in the following way: All right, she would say, in natural languages we see the phenomenon of syntactic metaphor: All structures start out as specific ones, and they are projected to more and more conceptually different cases, so that rather few forms can serve many different purposes, and the conceptual relations expressed by these few forms are quite varied. But all the same: Should it not be possible to construct a logical language that avoids these structural ambiguities in the following way: One would start out with purely semantical structures in the sense of complex expressions in which every semantical difference is strictly mirrored by a difference in form. And as one proceeds, one avoids all kinds of metaphors, especially the syntactic ones. Where a new semantical relationship has to be expressed (as in 'alleged millionaire's daughter' as compared to 'unmarried millionaire's daughter), new formal devices are introduced, instead of a 'misuse' of already given ones. All projections are strictly avoided. Applied to the examples above, this means that it has to be formally clear that the word 'alleged' does not belong to the same category as the word 'unmarried'. It should be made formally clear that the syntactical form of 'these slabs are broken' is different from the syntactical form of 'these slabs are five'. In this sense semantics would take the lead over syntax; every difference in semantics enforces a difference in syntax. The result may be a very complicated, very clumsy logical language; but should it not be possible to spell it out? And why should it be pointless to do so?
I hesitate to maintain categorically that it would be impossible. But a few considerations about what would be involved in realizing such a program for me raise grave doubts as to its sense, especially from the point of view of a philosophy of language. First of all it is not clear to me, in what sense one can speak of semantic differences in such a way that semantics can really take the lead over syntax, i.e. that syntax in all cases follows semantics. When we look e.g. at the phrases 'my neighbour's house', 'my neighbour's wife', 'my neigbour's death', -should one distinguish three kinds of the genitive here? For surely the kind of 'belonging to my neigbor' is a different one in each of the three cases. It is possible to say 'my neigbor sold his house' but conceptually impossible to say 'my neigbor sold his death'. This is a grammatical fact in Wittgenstein's sense, but could it be made visible by more elaborate grammatical rules, in the sense of rules that distinguish categories of expressions and regulate their concatenation? How would the resulting expressions look; how much of current scientific or philosophical opinion would have to be included? E.g. how would I have to speak of 'my neigbor's soul'? Can a point be reached at which we could be sure to have made explicit all semantical differences? And in what sense could the totality of the resulting expressions be called a language?
I would like to leave this project to the linguists, although I see that they might ask the philosophers for help, as the problem with my neigbor's soul indicates. Philosophically, I think, the realization of such a program, even if it is possible, would not make sense because it would misrepresent our linguistic competence. What it would describe could at best be an artificially arrested state of the history of our constantly developing language games, and it would fail to make visible the sources of the development. As Wittgenstein's discussion about rule-following and about certainty have made clear, it is a very basic trait of our practical competence that we succeed in continuing a given number of examples in such a way that a common activity can be upheld that we share with other members of our species. Such ways of 'meaningfully going on' are the basis of all rules; they cannot themselves be explained or secured by rules.
Metaphor, in the lexical and in the syntactic sense, is a case of such a 'going on', of such an active production of analogies. For this reason the development of a logical language without metaphors would not explain our capacity for metaphor; it would not make explicit any implicit rules we use in understanding projections. It just would be a different medium, useful for purposes like the processing of language on a computer. But it would not be an explication of a pure realm of thought or of the hidden conceptual structures of all possible speech acts.
Following Wittgenstein I think that we do not look at a second structure when we are able to distinguish the grammatical from the logical form of sentences. Instead, we find ourselves able to give reformulations, translations, commentaries, on occasions where they are needed and in such a way that they are helpful on the occasion. Among our possibilities of explaining kinds of differences between expressions is also the possibility of giving reformulations in an especially invented schematic form, like Aristotle's syllogism or Frege's way of expressing generality with the logical quantifiers he invented. But to think that the ability to distinguish the meaningful from the meaningless is concerned with a totality of expressions and can be represented as a formal ability, is a mistake. To understand what Wittgenstein calls the 'grammar' of an expression is something different from the ability to categorize expressions and to judge their well-formedness according to rules of concatenation. And on the level of understanding our own abilities this remains true regardless of the success of attempts to formalize the whole of language, as spoken by particular people at a particular moment.
Looking back once more to the millionaire's daughter, we must now say that the impression, a speaker of English would, on hearing the sentence, percieve two structures, was an illusion. Her understanding of the semantic difference can more aptly be described as an understanding of one structure and, additionally, an understanding of the projection. It is a competence in the 'grammar' of her language in Wittgenstein's special sense, i.e., a competence in use. The new content that is carried by the old structure is grasped by the hearer by way of understanding the old structure and the projective step. And her ability to understand the projection is not a formal competence. The special character of this ability would be lost, if we would try to describe it in terms of formal operations in the medium of an internalized logical grammar she unconsciously follows. When she understands the new sentence, the hearer does not handle a second structure, but only one structure, and the projective steps that lead to a variety of different uses.
Thus I think that with the inclusion of metaphor and other types of projection we have indeed a broader, less formal understanding than Michael Dummett allows of the types of systematic connections that can exist between sentences. We can see that the concept of a 'propositional content' only appears to stand for something very general on the semantic level; semantically, the most diverse kinds of 'facts' are captured with help of this form, as simple sentences like 'these slabs are five' or 'the concept >horse< is not a concept' show. Still this form is a systematic tie between many of our sentences, and this is one reason why we indeed do not learn the sentences of our languages one by one. But we can understand the fact that forms can function as such ties, only if our projective capacity is taken into account. We need the forms, but we are free to 'misuse' them, to apply them in areas where no speaker before us has "settled anything" yet.
One consequence of this freedom is the relative independence of syntax (in the linguistic sense) from semantics. This independence was seen by linguists like Chomsky; he confessed to be unable to perceive of a 'purely' semantic structure. But I think that he draw the wrong conclusion from this observation, namely, to build his syntax in a completely formal way, to take the logical calculus and its separability from the realm of meaning as his ideal, thereby changing the old linguistic sense of the term 'syntax'. Instead, I would like to learn from a linguist more about the interplay between syntax (in the traditional sense) and semantics. And I would hope that some light is provided by my discussion of a related interplay, that between the fixed rules on the one hand, and the free uses of them in the activity of speaking and hearing on the other.
In his book 'The Authority of Language' James Edwards(14) has recently argued that the roots of philosophical nihilism lie in the idea that the whole of a human form of life can be captured by a set of rules. If it could, it would seem to be our duty to judge our own rule-set in comparison with our ideals and in comparison with the rule-sets defining other forms of life. We would have to rationally choose who we want to be, in a very radical way. Edwards thinks that only nihilism can be the result, and he tries to avoid it by showing that a form of life cannot be captured in its totality by a set of rules. If what I have presented in this lecture is correct, there are good reasons for Edwards' view. In this way the philosophy of language could help to refute philosophical nihilism, which is a worthwile task reaching beyond the academic ivory tower.
This lecture was prepared following an invitation from the universities of Trondheim, Bergen, and Oslo and was read there in spring 1995.
(1) This lecture tries to summarize some main points of my book "Phantasie und Kalkül. Über die Polarität von Handlung und Struktur in der Sprache", Frankfurt (Suhrkamp) 1992. It was prepared following an invitation from the universities of Trondheim, Bergen, and Oslo and was read there in spring 1995. I would like to thank my hosts in Norway for their stimulating discussion and also John Granrose for correcting my English.
(2) Ludwig Wittgenstein, Philosophische Untersuchungen/Philosophical Investigations, New York 1953, I, § 41
(3) Richard Rorty, Philosophy and the Mirror of Nature, Oxford (Blackwell) 1980
(4) Donald Davidson, A Nice Derangement of Epitaphs; in: E. Lepore (ed.), Truth and Interpretation; Perspectives on the Philosophy of Donald Davidson,433-446, Oxford (Blackwell) 1986
(5) Michael Dummett, What is a Theory of Meaning? (II) In: G. Evans, J. McDowell (eds.), Truth and Meaning, Essays in Semantics, 67-137, Oxford (Clarendon) 1976
(6) M. Dummett, Frege and Wittgenstein; in: J. Block (ed.), Perspectives on the Philosophy of Wittgenstein, 31-42, Oxford (Blackwell) 1981
(7) Rudolf Carnap, Intellectual Autobiography. In: P.A. Schilpp (ed.), The Philosophy of rudulf Carnap. La Salle, Ill., 1963 (The Library of Living Philosophers, Vol. XI), pp. 3-84; quote: p. 11.
(8) G. Frege, 'Gedankengefüge', in: Frege, Kleine Schriften, hrsg. von I. Angelelli, Darmstadt (Wiss. Buchgesellschaft) 1967, p. 381 (orig. p. 39).
(9) L. Wittgenstein, Philosophische Grammatik, hrsg. von R. Rhees, Schriften 4, Frankfurt (Suhrkamp) 1969, p. 205.
(10) Wittgenstein 1953 § 1
(11) L. Wittgenstein, Preliminary Studies for the 'Philosophical Investigations', generally known as The Blue and Brown Books, Oxford (Blackwell) 1958, p. 17
(12) Eric Stenius, Wittgenstein's Tractatus. A Critical Exposition of its Main Lines of Thought, Oxford (Blackwell) 1960, p.212. Cf. my "Syntactic Metaphor: Frege, Wittgenstein, and the Limits of a Theory of Meaning; Philosophical Investigations 13 (1990) 137-153.
(13) D. Davidson, "The Logical Form of Action Sentences"; in: Davidson, Essays on Actions and Events, Oxford (Clarendon Press) 1980, p. 105-122.
(14) James Edwards, The Authority of Language, Tampa (Univ. of South Florida Press) 1990.
According to a workable definition by J.M. Soskice a metaphor is 'that figure of speech whereby we speak about one thing in terms which are seen to be suggestive of another'(1). A prominent feature of this characterization of metaphor (as of many others) is that the 'thing' spoken about seems to be taken as 'given', as an entity independent of the particular words chosen to speak about it. There is, so it seems, one definite object of discourse, in respect to which different linguistic approaches are possible. The linguistic means for grasping the object are multiple (and some of them are metaphorical), but the object stays the same. In the most basic case the object seems to be of a kind that one can hold in one's hand or point to with one's finger. Only when we say something about it, when we classify it with help of a predicate, does language seem to come into the picture.
A few years ago, for example, some people used to speak about a prominent German politician in terms of expressions like Birne, reif, etc.(2) This way of talking has more or less disappeared by now, but the same identical person is still around. We can point to him, introduce him to somebody, or speak about him. Or consider another example, which makes use of a different medium: Recently a poster that advertised science books was on display in German bookshops. It showed the slogan 'Wissen schafft Kompetenz'(3), together with a deeply impressive close-up portrait of an orang-utan, his chin rested upon his hand and his eyes thoughtfully gazing at the viewer. The poster shows a competent professional (e.g. a university professor) as an orang-utan. It shows one thing (a professor) in (pictorial) 'terms' which are suggestive of another. In both cases the 'thing spoken about' (in my examples: the particular politician and the generic 'competent professor') can exist independently of language (or at least of the metaphorically used symbol). It can be spoken about in many different ways, and there seems to be no problem in saying that it remains the same, regardless of the way it is treated linguistically.
In the case of the generic person, however, doubts may arise. Do 'generic persons' like the competent professor or the engineer exist independently of language so that we are entitled to say of the orang-utan picture that it it is taken to refer to her, to the respective 'generic person' (traditionally speaking, that the picture refers to the 'universal')? Or should we rather say that the picture metaphorically exemplifies a label like professor or engineer (i.e., it exemplifies an expression, a piece of language, and only via language may it refer to the respective linguistically classified 'things', depending on the viewer's choice of a label)? In the second case, it would be the expression professor that would constitute and so 'create' the genus. And this in turn would mean that the 'thing' that the poster refers to metaphorically is not altogether independent of language. But still, as long as verbal labels like professor exist, it would be independent of the particular metaphorical symbol(4). So we would still be entitled to say that with metaphors we speak of one thing (specified verbally as the professor) in terms which are seen to be suggestive of another (of the orang-utan).
Setting aside these complications, let us return to our first reading of the definition. If it does capture the whole story, the 'creative' side of metaphor, rich and important as it may be, seems to be restricted to what Frege called the realms of 'colouring' and 'sense' (Sinn): Although the use of new metaphors may reveal important facts, may bring new insights, and may lead to the discovery of new connections, it seems that the objects of our understanding (the references of our expressions) remain the same.
But do they? Looking e.g. at the history of science and its famous case studies, it is not clear what it would mean to say that science had always treated the same objects, even if it had used wrong theories to refer to them. Is phlogiston just a wrong conception, part of a wrong theory about (as we now know) really existing objects like oxygen? In that case we would have a wrong picture of an undoubtedly existing object. Or should we rather say that the stuff that had been called phlogiston never existed, that the object cannot be separated from the conception and that consequently a dismissal of a conception means the dismissal of an object? But then, how about oxygen, will it be gone one day, not because of an environmental catastrophe, but for 'purely linguistic' reasons, because we no longer speak about it in the way we are now used to do? Do we create these things linguistically, so that they disappear when our way of talking changes?
We can imagine two extreme views, the 'creation'-theory on the one hand, and the 'colouring'-theory on the other(5). The 'creation'-theory will hold that with the disappearance of a particular 'way of worldmaking' (to use Nelson Goodman's phrase) its objects, as inhabitants of that world, have necessarily disappeared also. When e.g. we find it impossible to reconstruct an ancient world, when we fail to make it intelligible for us, its objects (as these particular objects, in this particular relatedness to other objects and to the way of life of their times) do not exist for us. We read the old texts, but they do not lead us to the 'things themselves'. Many of us would want to say e.g. that the devil as someone who takes possession of a person does not (or should we say does no longer?) exist. The devil, even if he should enjoy a comeback some day, will never be what he used to be in the Middle Ages. With other ways of speaking having emerged, he seems to have disappeared.
But here the colouring-theorist will express hist doubt: is this really a purely verbal matter? Changes in 'ways of worldmaking', he claims, would be misdescribed as 'purely linguistic' affairs. He will remind us of Wittgenstein's 'language games'(6): In many or indeed most cases they surely involve more than just talking. And there seems to be no clear borderline between linguistic and more-than-linguistic changes. So there are some complications with the idea of a purely linguistic realm, and they might invite the position that in speaking metaphorically we do not create things linguistically, but we redescribe fixed and given things that remain the same. This is the position of the 'colouring theory'. It insists that it is impossible that an object should disappear for purely linguistic reasons. Instead, we must say: The 'phenomenon' that gave rise to a formulation will go on to exist, e.g. the strange personal behaviour that in the Middle Ages would have been spoken about (unfortunately and wrongly) in terms of one's being possessed by the devil. Only our description of the phenomenon changes. That does not exclude the possibility that the phenomenon might disappear for other reasons, like the kind of hysteria so prominent in the times of Sigmund Freud and so seldom found today. But here the reasons are not a change in the description, but, say, a change in the habits of child-raising. The phenomenon, if it does, will disappear for other than linguistic reasons.
Of the two answers, the negative one of the colouring-theorist might seem to have the stronger support: objects cannot come and go as linguistic habits change. This position admits that the colouring, the particular features, the way an object is 'given' and is related to other objects may change with the change of descriptions. But the existence or non-existence of objects, so it claims, surely must be independent of our descriptions. In terms of traditional philosophical doctrines one might say: Admittedly, as far as the 'thing in itself' is concerned, only God may be able to perceive it fully and correctly, but at least about the existence or non-existence of an object we can judge for ourselves, regardless of our many changing and perhaps never absolutely successful attempts to give adequate descriptions. Existence, so it seems, must be something in the 'outside world', and our linguistic attempts to classify objects, to describe them, etc. are provoked by the resistance we feel of something outside our will, by the often hurting symptoms of reality, of 'real life', in contradistinction to 'mere talking'. So the definition quoted at the beginning of my exposition seems to stand up against the challenges: Metaphors speak of objects in different terms, but there are no 'metaphorically created objects'. There is only a colouring of something given, but no metaphorical creation.
This position might be further strengthened by directing the attention to the difference between talking about a given object in terms of certain expressions on the one hand, and talking about inferred entities in terms of a proposed model on the other. With this difference in mind, a person opting for the colouring-theory will readily admit that often models are given up in the course of the development of a theory. And in this process, hypothetically postulated objects will be given up as well; they are no longer believed to exist. In such a case, certain objects will turn out to have been fictions. But, one will insist, their hypothetical status has been clear all along, and neither the reasons for their having been assumed, nor the reasons for our giving them up have been merely linguistic. There have been experiments, data, and activities like successful or unsuccessful attempts to act on the basis of the model.
This additional commentary on the role of terms for inferred entities once again shows the basic assumptions of the colouring-theory: There are objects on the one hand, and linguistic activities on the other. The linguistic activities may change, but this will not change anything on the side of the objects. The same is true for inferred entities: we may be wrong when we suppose that a certain object exists in a certain region of outer space, but by changing our theories we do not create or annihilate objects. The model of the object here is the stick in the child's hand, and the model of speaking about is whether the child calls it a horse or a gun. The place of metaphor is the predicate expression: we 'speak about one thing in terms which are seen to be suggestive of another'. We speak about the stick, however we might call it.
At this point I would like to take a look at the subject-position of sentences and at a phenomenon which (following Eric Stenius(7)) I propose to call syntactic metaphor. This step is meant to help destroy the view that only predicate expressions (functioning as predicates or as descriptions, like in the Rome of the East) are places for metaphors. To explain what I mean, I first take a look at the late philosophy of Wittgenstein.
One of the points in Wittgenstein's step by step development of ever more complicated language games in the Philosophische Untersuchungen is to show that all forms of language arise out of particular language-games, tied at first to quite particular purposes. Some of these purposes might be found in all human cultures. Among them there may be the purpose of sorting out the rotten pieces of fruit from the good ones, and this may lead to our familiar object-concept form which is so useful in handling 'middle size dry goods', as Charles Taylor ironically formulates(8). This form in turn invites the view discussed in the first part of my paper, namely, that identical 'objects' are treated in different terms, some of which are metaphorical. So I agree with Wittgenstein when he claims that even the distinction between object and concept starts out as a quite specific distinction, having its place in a quite specific language game.
The next important step that the development of Wittgenstein's model language games is meant to show is that once a structure has been developed, it can be projected (and typically is projected) into fields of use that are different from the field it originated in. And projection belongs to the same family of linguistic activities as metaphor. We not only say the table is brown but also John is sad or the meeting is closed. We not only speak of the baker's bread but also of the baker's arm, the baker's car, and the baker's wife. We use the same syntactical form for different semantical relationships. Such processes of projection are well known in the lexical case. That we are able to go on meaningfully in the use of a term, i.e. to apply it with communicative success to cases that are not repetitions of the original situations and often cannot have been anticipated by the community of language users, is a basic fact about our linguistic competence and the first step to metaphorical uses of language in the traditional sense. What I want to stress here is that not only words, but also structures like the genitive or the subject-predicate-form are projected, and that this also is something very familiar to all language users.
To characterize this projectability of structures, I propose to revive a term that Eric Stenius had coined thirty-five years ago in his book about Wittgenstein's Tractatus: He speaks, as I have mentioned, of 'syntactical metaphors'. He describes a sentence containing such a metaphor with the words: 'Its sense does not have the form suggested by its logical syntax, but nevertheless this syntax seems to be the best syntax we can give it.'(9) As an example Stenius cites Frege's famous sentence 'Der Begriff Pferd ist kein Begriff'(10); or, more simply, he could have used the sentence 'Pferd ist leicht gewinnbar'.(11) Here the speaker wants to say something 'about a concept' (that it is easy to explain), but he uses (Frege says: he unfortunately has to use) a sentence 'about an object'. In this process, the predicate expression horse moves to the subject-position of the sentence; so it looks as if there were a new 'object of discourse'. A sentence-form is 'metaphorically projected' to a realm where it serves a new purpose, where we have no objects (in Frege's sense) we are talking about. Taken literally, it is wrong, but there is no better form available, and the speaker is successful with it in her attempt to communicate. As in the cases of metaphorical expressions of the familiar kind, an old 'form of words' is put to a new use.
To give a clearer idea of what is involved here, I would like to add another, simpler example which is inspired by discussions in Frege's and in Wittgenstein's writings, but is (in the form I use it) not actually found there. Imagine a simple language game of the kind we find at the very beginning of the Philosophische Untersuchungen: Building material is ordered with the help of predicate expressions; other predicate expressions are available to classify pieces according to their colour or size; the numerals have been learned in such a way that it is possible to give an order like five bricks; there are expressions like this slab, serving as expressions for objects (so the language game has a particular version of the object-concept-form, showing in a sentence like e.g. these slabs are broken); and the step from orders to statements has been made. It is a part of this little language game that the numerals have so far only been used for orders or statements of the form five slabs, or four columns, i.e. a numeral has never appeared at the predicate or subject position of an utterance. And now imagine that a speaker 'misuses' an expression of the category of the numerals by putting it at the predicate-position and uttering these slabs are five. He metaphorically projects the form these A are B (where A and B mark the positions of predicate expressions) to a realm where it has not yet been used. From the point of view of the rules that had defined the game so far, this step is a violation of rules: The speaker uses the form of classifying objects for a speech act that is not a case of classifying, and he uses a count-word to form an expression that is not expressing an act of counting (but, as we may say, the result of such an act). This, I think, is a clear case of a metaphorical use of a sentence structure, of a 'syntactic metaphor'.
And now we can see that a similar step can be taken by putting a word for counting at the subject-position and uttering e.g, fourteen is even or five is odd. In both cases undoubtedly there is a violation of rules, as long as we look from the perspective of the language games established so far. But on the other hand we can be quite confident that the linguistic community will understand both types of utterances. This confidence derives from our past experience with our hearers' reactions to our utterances involving jokes, irony, or puns. Basically, a hearer will find projective steps of these kinds, unusual as they are in the particular case, no more difficult to follow than the projective step of using the word slab not only for a piece of building material, but, say, for a slice of bread.
You will remember that Frege (in his Grundlagen der Arithmetik) took a more conservative view(12). He argued that since a sentence of the form these slabs are five expresses a truth claim, we have to find out which is the object that the expression five says something about, and he works on a solution that shows that it is a concept that such a sentence says something about. Likewise, he struggled hard to explain in which sense 'sentences about numbers' (with a numeral in the subject position) can be seen as sentences about extensions of concepts, and that these extensions are objects of some sort. But from the Wittgensteinian perspective I have tried to sketch, it is unnecessary to search for such objects. Wittgenstein has given up the idea that all cases of truth must necessarily be cases of an object falling under a concept (or, as Frege proposes, a concept of first order 'falling in' a second order concept). For Wittgenstein, it is enough to see and understand the projective (metaphorical) step that puts numerals in predicate- or subject-positions (or a concept-word in subject-position, as in Frege's 'Pferd ist leicht gewinnbar').
So when we adopt Stenius' position and see these projections of syntactic structures as cases of metaphor, do we finally have an undisputable case of metaphorical creation? When in this way new subject-expressions come into use (in our example this is the 'name of a number' or a 'concept taken as an object'), are these cases of 'worldmaking' that have led to new kinds of objects? And how about the alternative of 'real' or 'only linguistic'? Is it so clear and intelligible as it seemed to be at the beginning? Does it make sense to ask, whether numbers are real or only linguistic?
As might be expected from the hints given earlier, I think that the alternative is too narrow. On the one hand we might say that numerals are words and that talking about odd and even numbers in some way means to talk about numerals. Now we can say that this means that we talk about language. So in this sense all remains inside the realm of language. But still it is not true that our creations are 'merely linguistic'.
The reason is that in the language game of counting we handle 'real', hard, non-linguistic 'objects'. Our counting can be correct or incorrect, correspond to the facts or not, and these facts are not something 'only linguistic'. And in that way, talk about our means of handling 'hard facts' is not 'merely linguistic' either. Thus I would like to conclude that the syntactic metaphor here has led us to new objects, objects that before the invention of counting and of the projective step of talking 'about' numerals clearly did not exist. Thus these objects have not been 'given', and the new metaphorical step is not only that they are spoken of in terms suggestive of another object.
Before I finish I would like to take a brief look at the philosophy of psychology. In particular I want to ask whether the expression state of mind, at least in some of its uses, can be regarded as standing for a 'metaphorically created object' of which we, in spite of its linguistic creation, would still (as in the case of numbers) not want to say that it is 'unreal' or 'only linguistic'. I think that considerations of the kind indicated so far can shed some light on what Wittgenstein is trying to get at when he uses an enigmatic formulation like the one I will quote in a moment. The context is the following: Wittgenstein thinks of a situation in which he silently beckons in order to make a person N. come over to him, and then of a later situation in which he explains what his beckoning has meant. He says: 'Man kann nun sagen, die Worte "Ich wollte, N. solle zu mir kommen" beschreiben den damaligen Zustand meiner Seele, und kann es auch wieder nicht sagen.'(13)
The traditional view, the one against which Wittgenstein is arguing, takes the words beschreiben and Zustand meiner Seele as in important respects similar here to their use in phrases like to describe an object (e.g. a stone or an inferred entity like a star), and the state of affairs on the highway after the tornado had hit it. According to this view, a 'state of mind' should be seen as an entity which exists independent of language. In such a view its status as an 'object of discourse' is not derived from the ability of its expression to appear in the subject-position of a sentence. So a 'state of mind' is perceived as having its own existence; the person having it can describe it or leave it alone or change it, i.e. she can behave towards it in a multitude of different ways. Accordingly, to describe here stands for an optional activity.
In contrast to this view one might hold that the phrase to describe one's state of mind is figurative in a way comparable to the phrase she left her in the lurch: As there is no spatial entity called lurch that could be approached from different directions to rescue the person left there, there is no entity called the state of x's mind that could in principle be approached by a number of different activities of which describing is only one. The occasions on which we say of someone that she is describing her state of mind are, according to this alternative view, situations in which we talk about moves in ongoing language games. Certain legitimate and meaningful kinds of moves in language games we happen to call describing one's state of mind although literally there is neither a 'description' in the narrow, object-oriented sense, nor a 'state' that the speaker knows of when she engages in this language game.
So according to this second view the move in the language game does not derive its meaningfulness from the independent existence of an entity called the state of her mind which is the object the speaker is talking about. Instead, a transfer of meaning takes place in the opposite direction: Our talk that seems to be 'about' an entity has a well-defined place in our verbal exchanges, and the seemingly descriptive entity-talk derives the meaning it has from this communicative function. As in the case of numbers one could say that we have an entity that has been created linguistically. But here too it is easy to see that the relevant language games are interwoven with 'forms of life', with practices like asking for motives, telling dreams, speaking about hopes, etc. In human mental life, nothing is only linguistic.
I would like to stress Wittgenstein's point, that the case in question is not one of inferred entities. Although we have good reasons, supported by a considerable amount of experience, to claim that 'states of mind' can be severely disturbed by damaging the brain, our talk about the mind is not an imprecise way of talking about states of the brain. Ich habe das Gedicht im Kopf(14) is as a metaphorical expression as I know the poem by heart. To somebody with low skill in knowing her feelings or finding her way in conceptual puzzles it wouldn't be good advice to read books about the brain. What she has to be trained in is the practice of certain language games, and this training can take place during psychotherapy or in a philosophy class.
Wittgenstein's positive answer is: We speak about ourselves as persons when we are talking about our 'states of mind'. In the course of doing so, we are telling stories, are answering questions; and we use the expression state of mind. When we see its grammar correctly, there is no reason to avoid the expression. But when we fail to see its figurative character, when we e.g. falsely take it to stand for an entity or state of affairs in our brain, then 'one may not say so'.
So as a conclusion I would like to say: There indeed are metaphorically created objects, especially if we include 'syntactic metaphors' as means of metaphorical creation. They have their place in language games, but it is misleading to call language games only linguistic. They are closely interwoven with 'forms of life', and the particularly human forms of life and their objects are so deeply embedded in language that we cannot imagine them to exist without any language.
This paper was read at the conference 'Metapher and Rational Discourse', Trinity College, Dublin, May 1995. It will soon be available in a book with the same title, eds. Debatin, Jackson, Steuer (Niemeyer Verlag, Tübingen, Germany)
(1) Janet Martin Soskice, Metaphor and religious language, Oxford, Clarendon 1985, p. 15. - I would like to thank Frederick Ferre for his careful reading of this paper and his helpful suggestions.
(2) Pear, ripe.
(3) Knowledge brings (or produces) competence, the first two words being a pun on Wissenschaft, i.e. science.
(4) Nelson Goodman, Languages of art, Indianapolis, Hackett 1976.
(5) For convenience I am using the term 'colouring-theory' to include both, the positive or negative 'colour' that an expression confers to an object, and what Frege called the realm of sense, i.e. the way an object is 'given' to us.
(6) Ludwig Wittgenstein, Philosophische Untersuchungen / Philosophical Investigations, New York, Macmillan 1953.
(7) Eric Stenius, Wittgenstein's Tractatus. A critical exposition of its main lines of thought, Oxford, Blackwell 1960. Cf. Hans J. Schneider, 'Syntactic metaphor: Frege, Wittgenstein, and the limits of a theory of meaning', Philosophical Investigations, 13 (1990), 137-153, and H. J. Schneider, Phantasie und Kalkül, Frankfurt, Suhrkamp 1992.
(8) Charles Taylor, 'Theories of meaning', in: Taylor, Philosophical papers vol. I, Cambridge, Univ. press 1985, 248-292.
(9) Stenius, loc.cit. p. 212.
(10) The concept horse is not a concept. Gottlob Frege, 'Über Begriff und Gegenstand', Frege, Kleine Schriften, ed. Ignacio Angelelli, Hildesheim, Olms 21990, p. 170.
(11) Horse is easy to explain. This is my simplification of the sentence taken up by Frege from Kerry 'der Begriff "Pferd" ist ein leicht gewinnbarer Begriff'; loc. cit. p. 169.
(12) G. Frege, Die Grundlagen der Arithmetik. Eine logisch-mathematische Untersuchung über den Begriff der Zahl. Centenarausgabe, mit ergänzenden Texten kritisch herausgegeben von Ch. Thiel, Hamburg, Felix Meiner, 1986
(13) 'One can now say that the words "I wanted N. to come to me" describe the state of my mind at that time; and again one may not say so.' Ludwig Wittgenstein, Philosophical Investigations / Philosophische Untersuchungen, New York, Macmillan 1953, § 662.
(14) Literally 'I have the poem in my head'.
I would like to begin with a story: In his fascinating book 'On the origin and development of a scientific fact' (1935) Ludwik Fleck reproduces an illustration from a book on human anatomy of 1642. It shows the female reproductive organs as found by the straightforwardly empirical method of cutting open a corpse. Especially it shows a small tube or 'ductus' that no anatomy book of today tells us anything about; we are sure that it does not and did never exist. Fleck then explains that the ancient theory of an analogy between the male and the female reproductive organs led the investigators to expect with certainty to find that particular tube; and sure enough they indeed found it and depicted it accurately in their textbooks.
What Fleck is doing here is to protest against the positivist programme of his Vienna contemporaries like Rudolf Carnap in a way quite similar to the protests of Karl Popper and similar especially to what later Thomas Kuhn would have to say about the structure of scientific revolutions. Kuhn acknowledges Fleck's influence, and with Kuhns book in mind we can understand how Fleck came to speak of 'the origin of a scientific fact': He shows that it is necessarily a framework of concepts and conceptions that informs even the most basic empirical investigation. There are no 'brute' scientific facts, outside any conceptual framework, and consequently there is no way of just reporting, just mirroring brute facts. It follows that in order to fully understand a 'scientific fact', it is necessary to understand how it was produced. Of course, one cannot say that scientific facts are 'invented'. But metaphorically speaking one has to acknowledge that every such fact has two 'parents': one is conceptual, here we are on the active side; and the other one is empirical: after we have knit our conceptual net, we are on the passive side and we have to wait and see what we have caught.
My motive for organizing this panel is the contention that to reflect on this connection between conceptual steps and empirical findings is helpful and indeed necessary in the context of this conference for two reasons. Firstly, if it is true that even in a science like Anatomy conceptions and concepts strongly influence and even produce scientific facts, it is clear that in the studies concerned with languages and cultures this will be the case to an even larger degree. The idea to possess an abstractly rational methodology with the help of which one can just mirror the facts is clearly an illusion. To be unaware of this illusion and to cling to what we in the West have come to regard as scientific methods would lead to a kind of imperialism: to the establishment of a specifically Western perspective under the wrong name of scientific rationality.
So the empirical scientist has to critically ask him- or herself: To what degree and in what particular way am I producing the facts? Does my kind of production result in a distortion, bocoming visible when my results are compared to how the people under investigation themselves describe what they are doing? And if so: Am I not thereby systematically producing misunderstandings between cultures? Having formulated these questions, the next step is to look for a possibility to avoid scientific imperialism. Is there a way to be aware of the fact-producing side of conceptual frameworks and still work for the goal of a common, a shared account of cultural differences?
So my first motive in planning this panel was to investigate the prospects of a third possibility: a shared account without the illusions of objectivity. It is a third possibility as compared to two other views: The first is the eurocentristic 'imperialism' already described. It is undefensible and so no real possibility any more. The second possibility is more popular; it is what I would like to call a post-modern 'tribalism', i.e. a general relativism. It seems to be a natural consequence of Fleck's claim that there is not one truth, but a number of competing conceptual perspectives which to some degree 'produce' quite different 'facts'. But relativism is an unattractive position also: While imperialism rests on an epistemological illusion and is ethically undefensible, relativism is a position of resignation: it isolates perspectives and discourages our curiousity to learn from people and from ways of life we are as yet unfamiliar with.
My second motive for considering the issues of this panel relevant for this conference is what I see as the tribalism in the young field of Pragmatics itself. If we as members of this Association want to have some shared view of the unity of our field instead of a great number of unconnected paradigms and empirical findings, it is advisable to try to understand how the diverse empirical results are connected. If what has been said about the fact-producing character of science is true, such an understanding can be accomplished by (1) explicitely stating how the empirical results depend on particular antecedent conceptions and concepts and (2) how these conceptions are connected to pre-scientific perspectives and intuitions. To be able to discuss the relative merits of these conceptions seems to me to be a necessary condition for overcoming the tribalism of the field. So there is some parallelism between the problems of interdiscplinarity and the problems of intercultural communication.
Since I myself am working as a philosopher, I would like to mention that the concerns of this panel as I have outlined them are philosophical in a specifically western and specifically modern understanding. They are concerned with the theory of knowledge, and one of the traditional occupations of this branch was to search for a universal method to distinguish valid from invalid knowledge-claims. The result was expected to be a methodological monism and universalism: Carnap e.g. expected that there would be found one method for all the sciences and one understanding of 'science' for all rational creatures. In the course of the development of Carnaps philosophy of science important distinctions were introduced. And some of these distinctions can be used to describe the limits of monism and universalism. Of special interest in this respect is the distinction between conceptual and empirical questions.
The most basic case of this distinction is the division between questions of logic on the one hand and questions of fact on the other. At the beginning of this century it was conceived as a kind of revelation to see the truths of logic no longer as very general truths about the world, but as truths following from the structure of language. If we now take 'logic' in a broad sense to include the doctrine of concepts, we can equate 'logical' and 'conceptual' truths. To give very trivial examples: That every rose is a flower is a conceptual truth, comparable to the truth that it is not the case that both a and non-a are true; once the meanings of the relevant words are fixed, these sentences cannot but be true. And in a more sophisticated but basically similar way every terminology produces conceptual truths. In Behaviorism, e.g., we can say that every response answers a stimulus; and this is a conceptual truth. On the other hand it is an empirical truth that some roses are yellow or that some yellow roses can be found in Europe.
As a next step we have to acknowledge that the grid or net (contrary to what Frege and the early Wittgenstein had thought) is not forced upon us by 'the nature of the facts'. Fleck here spoke of 'active connections' to highlight the fact that we as investigators are actively doing something and that normally there is a situation of choice: In principle we can decide between a number of different conceptual approaches. We have seen that this is so even in a science like Anatomy; and in the humanities we can easily imagine that e.g. religious customs might by one investigator be treated as superstitions, and by another as based on deep human experiences that are worthy of serious further investigation.
That there is a certain amount of freedom, a relative independence of the framework chosen to conceptualize a field of experience, fits nicely to prominent claims of the tradition of pragmatic thinking: Peirce in his famous 'pragmatic maxim' had stated that the meaning of a word is the way in which its utterance can guide our actions; this foreshadows Wittgenstein's later formula that 'meaning is use'. And if we succeed to free ourselves from too narrow fixations in our understanding of 'rationality' and 'scientific method', we can see that normally we can approach a situation with quite different goals in mind. Different goals lead to different actions and different linguistic actions constitute different concepts.
One possible reaction to this epistemological situation is what I have called 'tribalism': Each researcher opts for a paradigm of his or her own choice, a traditional school of thought, a specific conceptual framework, and tries to keep all his or her interests within the confines of this limited field. She tries not to take notice of other 'tribes', except by competing with them for funds and resources, i.e. for external success. An early version of this reaction was Carnap's 'principle of tolerance': Since he presupposed that no rational argument is possible to evaluate different frameworks, we have to be tolerant towards scientific languages, waiting (in a Darwinian spirit) for the practical results. The tribes compete on the practical level, and some in the long run will turn out to be fitter than others.
It may be seen as an advantage of this choice that it can lead to a relative and limited peace of mind, an undisturbed determination to try out one specific path with rigour and consistency. If you just follow the paradigm you have inheritied from your teachers, you will not be disturbed by the huge amount of other activities going on around you. But on the other hand a blindness for new perspectives, for relevant findings in other fields, an unability to explicitely locate one's own approach in the broader field of competing and complementary approaches, surely in the long run will be a disadvantage. And philosophically it is unsatisfactory from the start: You do not understand what you are doing.
So a second and more satisfactory reaction to the end of the idea of the 'mirror of nature' (Rorty) is the explicit acknowledgement that paradigms and perspectives should not just compete, but should be discussed, compared, evaluated. This means that the researcher not only has to be sure to follow a 'clear' or 'logical' methodology, but one that can be defended as relevant and/or as less distorting for the 'objects' under investigation than a competing approach. Questions of this kind are located on the methodological level; they cannot be answered inside a specific paradigm, but have paradigms as their subject matter.
In addition, they are philosophical questions in a more ambitious, ethical sense, as soon as considerations about the goals of a particular scientific approach are included: Why do we want to know what we are planning to investigate, and how do our methods of investigation fit to that end? Since questions about conceptual schemes are questions of meaning and since from a pragmatic perspective questions of meaning are questions of action, the ethical question 'what is the proper approach' cannot be excluded from methodology. In this way the philosophy of Pragmatism and the considerations called 'pragmatic' are good starting points to investigate the details of the 'third approach', avoiding imperialism as well as tribalism.
I now turn to the question how in the humanities investigations can be conducted that are on the one hand open to rational argument (and insofar proceed according to a 'scientific method' in a broad sense) and that on the other hand do not constitute a danger to the cultural identity of the people under investigation. If we now speak of 'scientific method', this term cannot mean a neutral, objective method of collecting independently existing 'facts', e.g. by the method of prepared questionnaires that sociologists have favoured for some time. Fleck and Kuhn have shown this even for the natural sciences. On their field often at least a common goal exists, namely that of controlling a particular natural process. Insofar as it is clear what a control would consist in, it is often possible to agree upon what is a success and what is a failure on the way to to this goal. Since in the humanities our goal is (or should be) to understand, not to control a certain domain, the criteria for success are far less obvious; quite different views can claim to have understood something, so the criteria for success are themselves subject to disagreement between different investigators. Furthermore, in the case of the humanities not only the investigators have particular (possibly contrasting) perspectives on their field, but also the people under investigation have their own perspective (or perspectives). So the possibility of competing views with competing definitions of the goals of understanding is given from the outset.
I would like to argue now that a pragmatic approach that sees language as a purposeful activity, embedded in other activities, is the most promising route when we are looking for a way out of the dilemma between scientific imperialism and tribalistic relativism. The special methodological point I would like to make here is that both the language of the people under investigation and the language of the investigator should consciously and carefully be reflected upon, and that both languages should be perceived according to the broad, pragmatic understanding of what a language is. So my claim is: We have to be Pragmatists not only in regard to our objects of investigation (a point unnecessary to stress on this conference), but also in regard to our own language of research.
What exactly is to be gained in this way? I think that a careful pragmatic reflection on her own language will enable the investigator to relate particular conceptual frameworks customary in her scientific community to their pre-scientific roots. This means that she can relate them to their place in the history of her own culture: to folk-theories, to pre-scientific and scientific verbal and nonverbal practices, to the goals of such practices and to matters of evaluation. A discussion on this level will make possible a comparison of paradigms, an evaluation of their relative adequacy. These evaluative perspectives are communicated and criticized by the narratives in terms of which the identity of the investigator is constituted, i.e. the specific understandig of what it means to be a member of that particular culture. One such narrative is the story of mankind as enlightened and freed from superstition by means of western science. Normally, as members of that culture, we take very much of this for granted; we live in a paradigm without reflecting on its history or its place in the broader cultural context. The dangers of this unreflected use of the customary methodologies are (as we have seen) the distortion or even production of facts, with the result that the people we are trying to understand cannot recognize themselves in our descriptions. Misunderstanding and conflict follows, or a polite way of not being interested or of confirming the expectations of the investigator for no other reason than to avoid embarrasment.
But if we do reflect on the pragmatic roots of our own conceptual setups, we have a better awareness of the particular, culturally located character of the approach taken. We have abandoned the illusion that the process of investigation is no more than a neutral collecting of facts. If such an awareness is indeed present on the side of the investigator, as an awareness of the pragmatic side of his or her own language, this will have a positive effect on the resources available in the case of conflict: A disagreement on the side of the investigated subjects with how we describe them, can now from the start be seen as being possibly a disagreement in very basic moves at the roots of quite different systems of language-games. On a smaller scale this can also be the case in interdisciplinary disagreements, so our experience with interdisciplinary discussions can to some degree guide us here.
What I would like to argue for can also be formulated in the context of the theory of meaning. The pragmatic approach to language is aware that the object of understanding cannot be restricted to something 'in the mind' of the subjects; something like the 'meanings' of their utterances that would only have to be 'translated', to be transported (as a definite, identical 'content') from one medium of expression (e.g. the Japanese language) to another medium, like English. If meaning is use, to understand meaning is to understand use. But to understand use is to understand a broad field of practices, verbal and non-verbal, the boundaries of which cannot be outlined in advance: they can include the most basic attitudes to biological human needs (food; sexuality), encompassing perspectives on life as a whole (religion), or very elaborate and specialized activities that are characteristic for only one culture (like, as far as I understand, 'the way of tea' in the culture of our hosts).
So meaning and acting are closely and in complicated ways interwowen; they produce the manifold conceptual grids of the different cultures and subcultures; it is not the case that 'the world all by itself' produces a special (the 'logical') grid that could be taken as a universal instrument for all kinds of investigation. It follows that the work we invest to advance understanding between two cultures cannot be started from a neutral point of view. The work to be done does not have the nature of a comparison in the sense of a measurement, in which two items A and B are measured against a third, neutral item C. Instead, it is a hermeneutic endeavour: A person or a group of persons with one specific 'horizon' approaches another person (or group) with a different horizon, and whatever understanding can be achieved by the two parties involved does not guarantee the possibility of transfer to a third one. Metaphorically speaking, what we can reasonably aim at are contrastive 'cultural grammars', in which one language is described and made sense of from the particular point of view of a given second language. There is no neutral cultural grammar that could be a means of description for all particular grammars. Certainly the language of Western science is no such grammar.
In what sense, then, can we speak of a method, a way on which to proceed would justify to expect some progress towards understanding? Are not the matters that have to be treated so complex and so dependent on the individual circumstances of the people involved, that there is no hope for any systematic non-imperialist study? Will not an individualistic form of tribalism be the only result, so that instead of our middle-size tribes, gathering around their different paradigms, we will have huge numbers of investigating individuals, each of them with his or her strictly personal views?
At this point I would like to take up a hint given by Charles Taylor when he speaks of the necessity to develop a "language of perspicious contrast" and a proposal put forward by Jef Verschueren to give the concept of 'adaption' a central place in the definition of the field of pragmatics. I think it should be possible to apply what we know under the title 'the history of ideas' to the project of writing a 'contrastive cultural grammar'. The development of certain concepts, their change during time, can be made intelligible by seeing it as a history of adaption to changing situations. The changes in the situations will be partly natural, but to a great part cultural, so the term 'adaption' does not indicate a reductionist, biological perspective. The subject of adaption would be the cultural identity of a group: it changes and at the same time is preserved.
If the goal is a contrastive history of ideas I think it is most promising to begin on the level of ontogenesis, i.e. with the story of how in a paradigmatic case a newly born person becomes a member of the culture he or she is born into. The project would be to describe from the explicitely reflected point of view of one cultural tradition how a person becomes a member of a second (the investigated) cultural community. This story can be told as a story of adaption. Since every human being has biological needs, there is a starting point on which the situations in different cultures are similar enough not to pose unsolvable problems. But then very soon the needs of the infant are interpreted in particular ways, and he or she has to adapt to the particular cultural surroundings, acquiring new cultural needs, learning new ways to satisfy them, and so on.
When Taylor speaks of a 'language of perspicious contrast', I imagine that the description of one course of adaption, being characteristic of one particular culture, can be given with constant explicit reference to the investigator's own way of adaption. The investigator should not try to give a neutral (e.g. behavioristic) description of the adaptive process, but should explicitely compare the options of her own pragmatic system of 'language + action' to the options taken by the culture described. In the beginning, when the situations of small children will be compared, we will probably be able to see different options as answers to comparable or even identical needs. But it is clear that the longer the adaptive process has been under way, the more the needs themselves are deeply shaped and indeed created by the particularities of the culture considered. So the project of writing a 'contrastive cultural grammar' does not entail a functionalism of a Darwinian kind; cultural needs are not epiphenomena of biological needs. This is what makes the project so complicated. But I see no other way than that of writing comparative histories of how a biologically human being adapts to the 'language games' constituting his or her culture, if the description produced should preserve cultural identity and at the same time should be helpful on the way to understanding. The perspective of adaption can secure the common ground; the contrastive element can avoid the relativism of a disinterested 'anything goes'.
As we know from the 'history of ideas' such a project can also be undertaken on a phylogenetic, i.e. on a historical level in the usual sense. Here it plays the role of helping us to understand our own situation as members of one culture. It is necessary in places where different ideas and traditions flow together in such a way that their relations are unclear or even contradictory. And it can be argued that this is the situation we are currently in in the field loosely described as 'Pragmatics'. Here again the perspective of adaption is helpful, understood as the question 'to what needs is a certain proposal (method, theory, research strategy) an answer, and how are these needs related to more basic, e.g. to pre-scientific needs?' Here again not only the investigated language is treated in a pragmatic perspective, but also the language used in investigating it.
But these are more abstractions and generalizations; I think that it is now high time to see what the 'field workers' can do with ideas such as these. There may be a wide gap to be bridged and the philosopher might have to learn that certain ideas he likes are illusions. But I am convinced that without constantly trying to keep in touch with the wider perspectives there will be either tribalism or imperialism or a mixture of both.
[After Post-Modernism Conference. Copyright 1997.]