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

1. The Context Principle

The tangle Moore’s paradox, a problem which first came to be widely discussed in the 1940s, is well described as resting on an insufficient understanding of the import of a “context principle”. Moore’s paradox, in G. E. Moore’s own description, was that it is paradoxical, that it should be absurd for me to say, in an assertive way, that “I believe it is raining and it is not raining” (Moore 1993: 208).

The most commonly cited context principle is ascribed to Gottlob Frege: the principle that a word has meaning only in the context of a sentence or rather, that the meaning of a word should be considered only in the context of a sentence. A more general version of a context principle in circulation is that the meaning of a sentence (or any linguistic sign) is given by its context. Frege’s context principle is often taken to suggest a fact: that the smallest meaning-bearing element of language is ‘the proposition’ (or sentence, Satz). From this fact, it would follow that the proposition or sentence carries its sense within itself as opposed to words. Words, it is thought, can be used in many different ways – to take an example from Wittgenstein’s Tractatus, “Green is green” is not by necessity an identity statement, but may just as well be used to say that Mr Green is not feeling well or has a particular political inclination – depending on the context (or the particular intention by the speaker in that context).

Frege’s principle was introduced as a methodological rule. It was not a discovery about the proposition. Frege’s aim was to create a script or language for the description of mathematics, it was not to pose a semantic theory or a theory of meaning in the sense of today’s analytic philosophy of language (cf Stenlund 2002: 15).

A context principle challenges the line drawn between semantics and pragmatics by connecting the propositions (as signs or collections of signs) to the contexts of utterance which give them their particular sense. Calling upon a context principle is a challenge to compositionality, according to which signs carry their meanings in themselves and bring them into the sentence by combination according to rules (which also contribute to the sense as a whole).

There is a mistaken view of the import of the context principle in circulation, according to which the context principle gives rise to an extended compositionalism: if meaning is not to be found in words, it must reside in sentences: sentences are the smallest meaning-bearing entity in language. This view underlies Donald Davidson’s extension of the context principle: “only in the context of a language does the sentence (and therefore the word) have meaning” (Davidson 1967). This picture of a language has a set theoretical background: it is a set of sentences.

Cora Diamond has developed a Wittgensteinian version of Frege’s context principle: words do not carry their meaning in themselves, but the structure of rules governing the composition of sentences does not either (Diamond 1981/1992). According to Diamond, nonsense is not something one can construct by combining parts in the wrong way, because this would entail that the words in some way must carry the meaning within them into any context – in violation of the context principle. Lars Hertzberg has shown that this point can be extended to sentence-like structures – a particular sentence does not necessarily mean the same in any context. For example, “Caesar is a prime number” may work perfectly well as a judgment by a better about a pet turtle participating in a race. This is an extension of the context principle to a situation with a speaker having something to say – in contrast to Davidson’s extension to an abstract concept of ‘language’.

2. Moore’s Paradox

The context principle in Diamond’s and Hertzberg’s is a criticism against the distinction between that which is said and that which is implied, or what is said as opposed to saying it. To claim that the context principle would leave the proposition or content of an utterance undefined or undetermined would backfire: that would entail the very claim that an invocation of the context principle aims to counter: the supposition that it is, or must be clear what “the proposition” as such would be, that we could understand what a proposition necessarily means by encountering it in isolation, without context, without use. The point of the context principle is exactly to take into account the lack of clarity in relation to the proposition as a philosophical-theoretical (“semantic”) concept.

It may be true that I went to the cinema last Tuesday, but that I do not believe that I did. Nevertheless, Moore worries, it seems impossible for me to express this very “same proposition”: “I believe it is raining and it is not raining”. Someone else, however, could express it about me, and it could be true. This is Moore’s paradox: that it would be absurd to express a proposition which in Moore’s terms is “perfectly in order” and might very well be true.

One of the central ingredients of Moore’s paradox is the idea that there could be a proposition schema void of context (‘I believe that p & - p’), of which we (the philosophers) once and for all could judge whether it is “in order” or “assertible” or not. The particular example in the paradox (“I believe it is raining and it is not raining”) is taken to be clear because the schema is in order, and the problem is the application of the instance of the use of the schema. One could describe the paradox, about which Moore wrote that “it is paradoxical that it should be absurd to say it in an assertive way, although the sentence is perfectly in order”, as the dilemma that he wants to take the sentence to be in order although he realizes that it, at the same time, is problematic. And it cannot be both.

The context principle plays on two levels when it comes to Moore’s paradox. On the one hand, it seems that the sentence schema is faultless and hence that there is an acceptable proposition. But a proposition is not acceptable as such, only due to the schema – the context principle is the insistence that the schema does not determine whether a string of signs can be used to communicate something (“it’s content”) or not, but that a context is needed. The problem in Moore’s paradox is that we do not know what the content would be, what it would be for me to both believe that it is raining and at the same time claim that it isn’t.

However, it is possible to come up with a perfectly working use of the schema: suppose that I am at a party and I keep feeling behind my chair with my hand, and I explain my behavior to you: “I believe I have my purse with me but I don’t have my purse with me”. This is an example in which a context provides the schema a use and it shows that a schema as such is not sufficient to determine whether there could be meaningful use of a particular “sentence”. The “context” – in this context – is the situation, a surrounding, the description of an event in the context of which a sentence of the sort proposed could be used. And as soon as a context is given, the oddity of the sentence, the purported non-usability of it, disappears. This is one of the interesting features of Moore’s paradox – another being the “contradiction-likeness” of the sentence itself, ordinarily placed in focus by formally inclined philosophers of language.

The sentence is a blackboard construct (not something expressed in any potential “natural environment”) and that feature is stunning. But when the blackboard context (also a sort of context which poses as contextlessness) is exchanged for a real one, the Moorean feature disappears.

3. The Philosophical Term “Believe” Carrying Its Meaning in Itself

Another level on which a context principle could come to play is in relation to the words or terms in the Moorean sentence (the Moorean sentence I call the problematic one – it is no longer Moorean in a context where it works). In order to achieve a contradiction-like feature to appear in a sentence like “I believe it is raining and it is not raining” the unit under scrutiny must be taken to consist of two connected parts. The “and” hence is taken to be a formal conjunction, which makes both sides valid simultaneously.

The perspective in which the problem presses includes a special philosophical-terminological use of the word “believe”. The idea of a single meaning of the word “believe” which it has to have regardless of context of use is required to arrive at the contradiction-likeness: “I believe it is raining” has to retain its meaning through all contexts. It is taken to be the epistemological expression of knowledge as a state-of-mind, the opposite of “It is not raining” which would be “the expression of the belief” that it is not raining. (One example of this special use of “believe” is found in Searle 1969, cf Malcolm 1992.)

Now, does the use or assertion of a proposition always entail the expression of that belief? A “yes” here is the commitment to the terminological use of “belief”. (It is not an observation but a stipulation.) What is the alternative? The relation between the word ‘believe’ and something expressed (a sentence) is not necessarily constant, but in the terminologized version of belief, the exclusion of ambiguous or multiple uses is required. ‘Believe’ can be used to modify a statement: to express emphasis, soften commitment. To describe someone as believing that it is raining is not necessarily to ascribe him belief as a state of mind or a particular piece of knowledge. When I say “I believe it is raining” it may be about me or about the rain outside, depending on the context. “Belief” in a case where the Moore paradoxical sentence is taken to be contradictory is part of the description of assertion as the expression of belief, not of the phenomenon itself.

One road to a better understanding of Moore’s paradox could be to see that we are asked to determine once and for all whether the sentence will work or not, and that is to invoke the schema and the proposition (or its alleged given parts) as bearing their meanings in themselves.