From the ALWS archives: A selection of papers from the International Wittgenstein Symposia in Kirchberg am Wechsel

1. Origins

When Wittgenstein came to Cambridge in 1929 and started doing philosophy again after about ten years, he was mainly faced with the problem of colour incompatibility. In the Tractatus he argued that a claim such as: “Two colours are at the same point at the same time in our visual field” (cf. Wittgenstein 1989, 6.3751) is logically impossible. And since elementary propositions were defined as mutually independent, it followed that he probably regarded sentences such as “A is green” and “A is red” as not being elementary expressions. The symbolism of the Tractatus did, however, allow a logical conjunction of such expressions containing particular colour statements about certain points in our visual field. Because such conjunctions are not “mirrored” by anything in reality, this causes an asymmetry between what can be said and what is possibly the case. In Wittgenstein’s early writings, the concept of possibility was determined by the truth-functions and logical constants and therefore fully independent of what is said in those expressions. In one of his unpublished manuscripts, Rush Rhees remarks the following: “In LPA the Möglichkeiten were represented by the logical constants. And this means, I suppose, that they were represented finally by the logical constants which give (form?) the Grammatik jeder möglichen Beschreibung. In the Bemerkungen the ‘Grammatik der Beschreibung der Tatsachen’ is found or festgestellt in the Phänomenen. And these do not have the kind of systematic unity which LPA gives to the logical constants and the Logik/Kalkül der Wahrheitsfunktionen. […] This Phänomenologie was part of the recognition of the diversity of Systeme, the diversity of possible grammars: and so of possibilities. It had an immense importance in connexion with the notion of ‘logical possibility’ and ‘logical impossibility’.”

So, what we can claim first, is that the symbolism of the Tractatus was mistaken, or at least incomplete for it could not prevent propositional conjunctions that do not have a possible equivalent in reality, or to put it in Tractarian terminology, that do not picture a possible state of affairs.

Secondly, Wittgenstein had to reject a central feature of his elementary propositions, i.e. their mutual logical independency. In January 1930, he told Moritz Schlick that in the Tractatus he had introduced rules for the syntactic use of logical constants without considering that those rules might have anything to do with the internal structure and connexions of sentences, e.g. in cases such as “Blue and red are in one and the same point”. He now regarded such logical products as invalid. Therefore, those rules would only form part of an extensive syntax, he did not know about at the time of the Tractatus (cf. Wittgenstein 1993, 74). In his notes of Wittgenstein lectures between 1930 and 1933, G. E. Moore also points out that it was with regards to elementary propositions and their particular connexion with truth-functions that Wittgenstein had to change his views most: “His present view [1932] was that it was senseless to talk of a ‘final’ analysis, and he said that he would now treat as atomic all propositions in the expression of which neither ‘and’, ‘or’, nor ‘not’ occurred, nor any expression of generality.” (Moore 1993, 88)

Thirdly, as the remark also shows, Wittgenstein rejected his idea of a logical analysis of propositions within a truth-functional calculus, as the one and only way to get to elementary propositions that exclusively consist of names and show when a sentence has sense or in other words, that display what it means to say something, as opposed to a combination of senseless or nonsensical signs. Part of this dismissal was due to Wittgenstein’s misunderstanding of this kind of analysis as if it were like a chemical or physical one that detects hitherto hidden things and which could construct a theory of elementary propositions analogous to the principles of mechanics (cf. Wittgenstein 1993b, 210).

2. Transitions

Confronted with the problem of colour incompatibility in early 1929, Wittgenstein now claimed that what we need is a purely phenomenological theory of colours which only deals with what is really perceivable, excluding any hypothetical objects such as waves, cells etc. (cf. Wittgenstein 1991, 273). In his Some Remarks on Logical Form he consequently demanded “a logical analysis of actual phenomena” whereby those phenomena must be represented by numbers which therefore have to enter into elementary propositions themselves: “A simple example would be the representation of a patch P by the expression ‘6–9, 3–8’ and of a proposition about it, e.g., P is red, by the symbol ‘6–9, 3–8 R’, where ‘R’ is yet an unanalysed term (‘6–9’ and ‘3–8’ stand for the continuous interval between the respective numbers).” (Wittgenstein 1929, 166)

Without going into details about Wittgenstein’s idea of a numerical representation, for our purposes it is only important to point out that both his concept of logical analysis and of elementary propositions did turn away from his Tractarian ideas. Wittgenstein’s postulated elementary connexion between colour statements in terms of numbers, he had now introduced, obviously showed the possibility to construct propositions beyond the truth-functional calculus. Rhees remarks: “The phenomenological account of colours – the representation of the grammar of colours, for instance – may be something like a complete analysis of colour propositions. But it is not the sort of thing you have in a truth-functional calculus” (Rhees, unpublished). So, if this account of colours as they are given in immediate experience is the result of an analysis, it is not the kind of operation that went with the truth-functional calculus. This means, for instance that we would not yield the result that two colours – say red and green – cannot occupy the same point in visual space by applying the analysis of “This is red” and “This is green”.

Wittgenstein now spoke of primary propositions and their relation to hypotheses, a connexion that was not anymore guaranteed by any truth-functional nexus between propositions and their truth arguments or as Rhees once put it in discussion: The connexion between sense data propositions and hypotheses is not formed by truth operations, as the truth functions of elementary propositions are. They have a logical connexion with hypotheses – but not through truth operations. And the way in which hypotheses depend upon sense datum propositions for their connexion with reality and so for their sense, is not like the way in which truth functions depend upon elementary propositions in the Tractatus. So it was again the internal relation which became central in Wittgenstein’s new conception of a phenomenological language. During this period in 1929, he obviously thought that this connexion between language and reality was guaranteed by our immediate experience and the way in which immediate experience can be understood. The “primary language” he probably regarded as an account of sensory data which would verify or falsify hypotheses. In his Philosophical Remarks, he stresses this point by arguing that talking about sense data and immediate experience means to look for a non-hypothetical account (cf. Wittgenstein 1991, 283).

What Wittgenstein needed, however, to guarantee the necessary relation between language and reality was a type of proposition that would be immediately evident and he thought that propositions describing our immediate experience could fulfil this condition due to their apodictic character, a feature that, by the way, also distinguishes them from atomic propositions in the Tractatus.

Let us now briefly look at two criticisms that are connected with his conception of a phenomenological language:

The first is offered by Wittgenstein himself, when at one point he spoke of propositions which could be verified once and for all, and were not facets in the verification of an hypothesis, as something like surfaces which were not the surfaces of bodies (cf. Wittgenstein 1993b, 221). This would hold of any description of ‘immediate experience’. If such propositions are really propositions, they do not seem to lead on to anything beyond themselves. But the figure of a mere surface does seem to suggest that he did not think they were proper propositions – although they had an important role in the verification of hypotheses. If it makes no sense to say that they might be false, then they certainly cannot be elementary propositions. Rather, they are the means whereby a proposition may be connected with reality. If they were only surfaces, they would not belong to a meaningful description. This means that what we call “the description of immediate experience” belongs to hypotheses and their verification. They are so to speak not self-contained expressions but means to connect language with reality.

Secondly, Wittgenstein’s account of primary signs as signs that cannot be misunderstood might have been connected with the idea that he thought that sensory impressions themselves might function as symbols which means that independent of any conventions they contain a grammar that guarantees the demanded immediate evidence, a grammar of what can be imagined or thought. This kind of language would thereby exclude any nonsensical combinations of mental pictures. Phenomenological language is then characterised as a symbolism that warrants immediate understanding. It seemed that Wittgenstein did not, however, clearly distinguish between descriptions of phenomena and descriptions by phenomena and Rhees once told me that he might have sometimes meant the one sometimes the other, presumably due to the fact that he had still not clearly distinguished between the bearer of a name and the meaning of a name. Another quite famous critique also supports this assumption. When Frank Ramsey talks about the idea of acquaintance in connexion with Wittgenstein, he makes the following interesting remark: “Another is the argumentation about acquaintance with before leading to the conclusion that we perceive the past. […] It turns on a play with ‘acquaintance’ which means, first, capacity to symbolize and, secondly, sensory perception. Wittgenstein seems to equivocate in just the same way with his notion of ‘given’.” (Ramsey 1990, 7)

In one of his unpublished manuscripts, Rhees remarks the following: “The confusion comes, presumably, in treating ‘das Gegebene’ als das eigentliche Zeichen oder das eigentliche Symbol, a kind of limes to which we approach as we see what various equivalent ‘conventional’ symbols have in common: ‘Das, was verunreinigt worden ist’.” And: “It sometimes seems as though grammar would be what is common to all languages which say the same thing: as though such a language if it could be expressed at all, would be the expression of a grammar. But it is hard to see how there could be a grammar without signs. Is this the point of the suggestion about a grammar of Vorstellungen? Then you have the difficulty of whether this grammar could be the same if you changed all the words; how would you know which word stood for what?” This remark obviously shows one of the fundamental problems lying in the assumption that a phenomenological language could entirely do without signs and where immediate experiences could themselves work as symbols. For at least it seems that the combination of possible phenomena could not tell us how to use such a language i.e. form the grammar of our “secondary”, ordinary language. In other words, how could immediate experience tell us whether we describe immediate experience correctly, unless there were some rules for the use of words in such descriptions? These rules would presumably be independent of any immediate experience. This criticism also points to the way, Wittgenstein eventually saw the relation between a phenomenological language and the grammar of our everyday language.

3. Consequences

In November 1929, Wittgenstein remarks that a phenomenological or primary language is not anymore what he aims at (cf. Wittgenstein MS 107, 205–206, 1991, 51) and just a few weeks later he argues that he was wrong in assuming a primary language as opposed to our ordinary “secondary” language (cf. Wittgenstein MS 108, 29). But what his idea of such a language shows is that he had to introduce a new type of proposition that were different from the Tractarian conception of elementary propositions and logical analysis and that shall assure the internal connexion between language and reality.

In his Big Typescript, Wittgenstein then headed one of the chapters “Phenomenology is grammar” and what he meant was that phenomenology could now determine the range of all possible description.

This new idea becomes best obvious in his discussion of a grammar of colours. To illuminate his approach, Wittgenstein chooses the example of a colour octahedron, which he regards as grammatical and not psychological, like in cases where we investigate say coloured after images (cf. Wittgenstein 1991, 51–52). In fact, the configuration of colours in terms of a geometrical figure helps us to depict the grammatical rules of our colour concepts, i.e. why it makes sense to talk of a reddish blue but not of a reddish green. (The octahedron is, by the way, only one way of arranging our colour concepts.) In other words, a particular geometrical arrangement does not represent any empirical proposition or as Rhees puts it: “We are talking about colours, of course. We are not saying that one sense datum is between two others, which would mean‚ ‘spatially between’ I suppose. To see a colour is to see, what colour it is. And this is, where the concept comes in” (Rhees, unpublished). If colours did not have the particular configuration they have, they would not correspond to what we call “colour”. This shows the erroneous assumption that another arrangement than the one given were logically possible: “We should not know what was meant by ‘seeing red and green in the same place’. Nor can we ask whether experience agrees with the order of colours represented in the colour-octahedron, for instance. If they did not have this order they would not be what we call ‘colours’” (Rhees, unpublished). What is of course important to mention in this context is the fact that within a spacial representation, some of our ordinary expressions would have a different semantics, for instance, if we say that one colour is “between” two others this is not to mean “between” as in “between two chairs”, or if one colour, say R, “contains more” blue then another colour Y, this just means that R is nearer to blue than Y. But you cannot say that there is more blue in this colour than there is red in that, or that it is nearer to blue than that is to red, for we are not talking in terms of space, although the use of a geometrical figure might mislead one to think this way.

To close the circle of argumentation and give at least a clue what Wittgenstein meant by saying that phenomenology is grammar, let us shortly come back to our original problem of colour incompatibility. When Wittgenstein introduced his concept of a grammatical rule at the beginning of the 1930s, it became most clear in his claim that propositions such as “Red and green cannot be in the same visual spot”, “There is no such thing as reddish green” etc. are not descriptions of our immediate experiences. They cannot be the result of our immediate experiences but rather they form our grammar of colour concepts, or, generally speaking, they rule the use of our ordinary language. To negate such propositions is not anymore to assert a contradiction. What it means instead is that it contradicts a rule that belongs to our grammar of colour concepts.