Ludwig Wittgenstein famously wrote that "meaning is use". This was written in Section 43 of the Philosophical Investigations. I will here discuss larger translation problems relevant not only to this passage, but to Wittgenstein's work in general.
Man kann für eine große Klass von Fällen der Benützung des Wortes "Bedeutung"--wenn auch nicht für alle Fälle seiner Benützung--dieses Wort so erklären: Die Bedeutung eines Wortes ist sein Gebrauch in der Sprache (PU §43).
A preliminary translation of this passage might run as follows:
For a large class of cases in which the word "Bedeutung" is employed--though not for all--one can define this word in this way: the Bedeutung of a word is its Gebrauch in the language (PI §43).
I would like to comment directly on the German text. As indicated by my preliminary translation, there are two terms in this sentence which can be perfectly rendered in English only after we have explained some of the concepts associated with them. I will focus on the translation of these two terms and in so doing will also be able to describe a number of the important features of this text.
The German word "Bedeutung" has the difficult job of representing two different semantic concepts that are conveniently represented in English by two different words. In English, these words are "meaning" and "reference". In its first sense, "Bedeutung" corresponds to the generic English term "meaning". We can speak of the Bedeutung of "Apfle" in the same way that we might speak of the meaning of the word "Apple". And here when we say "meaning" (in English) we are not talking about what object, reified or not, the word "Apple" refers to. We are talking very simply about the meaning of the word "apple". There is also the second sense of "Bedeutung" in which it corresponds to a more complex use of the English term "meaning". In this sense of "meaning", we construe the meaning of a word along the lines of the object to which the word refers. This second sense of "Bedeutung" is best translated into English as "reference", though there certainly are many cases in which the English word "meaning" does the same job. For example, consider the case of the hardcore logical atomist who argued that the meaning of the word "Apple" is precisely the object to which it refers. Sometimes the English "meaning" takes on the meaning of the English "reference" just as the German "Bedeutung" attempts as well to express both concepts.
Of course, one of the reasons why it is so difficult to translate Wittgenstein on this point is because he is concerned with demonstrating the transparency of the very concept of meaning. And so, when we translate "Bedeutung" into the English "meaning", we must remember that "Bedeutung" is being defined in the passage that I have quoted; he writes: "One can define it in this way . . .". This indicates that Wittgenstein is very concerned with overthrowing any holdover philosophical assumptions concerning meaning that we may have inherited from Frege, Descartes, and Plato. The word "Bedeutung" certainly fulfilled a function before Wittgenstein wrote this text, and he is well aware of that. But, it is ordinary this function with which Wittgenstein is concerned, and not the extraneous (as he sees it) semantic concepts that have been attached to this basic concept by his philosophical forefathers. Wittgenstein is, I would argue, concerned with "Bedeutung" in the first more general sense. After discussing the German term "Gebrauch" we will be in a better position to see this.
The word "Gebrauch" in this passage is translated by Anscombe as "use". This translation is fine, properly understood. There are also good reasons for using the English word "application" here. Both words express the same concept if we make sure to recognize that both words have the capacity to refer to either a state-description or a dynamic-description. These two terms, as I use them here, are my own and I will return to them shortly. First, though, I need to clarify my usage of "use" here.
I suggest that it will be good to think of "use" and "application" as two-place predicates by which we are mapping some term to something which it is (to something that is it). Since I will be using it as a two-place predicate from here on it, I will use "use of" rather than "use" in order to help keep this clear. "Use of" is a two-place predicated of identity as I am here using it. Construing "use of" as a predicate of identity may seem strange at first, but it is really quite philosophically familiar on further thought. "Use of", as I construe it, bears numerous affinities to Aristotle's "telos" (function). Aristotle thought that a thing was defined by its function. A knife is something that's purpose is to cut. This was, for Aristotle, the definitoin of the term "knife". There are also certain affinities between my conception of "use of" and Heidegger's description of objects. A hammer is one tool in one context, but another tool in a different context for the simple reason that it is inolved in our projects in separate ways in the separate contexts. In one context, the hammer is a thing-to-drive-nails while in another it is a thing-to-remove-nails and in a third it is a simple weapon. I point out these affinities merely to show the reader that construing "use of" as a two-place predicate mapping identity is neither unfamiliar nor obviously unsound, though it may seem so at first glance. Having adequately discussed my use of "use of", I can now return to the above-given terms of "state-description" and "dynamic-description" and describe both them and their place in Wittgenstein's definition of "meaning".
When I speak of a state-description, I am talking about the sense of "use of" which might lead us to think of "Bedeutung" as defined by the compiled set of all past instances in which that phoneme sequence was uttered. A state-description involving "use of" implies, then, that use is being employed in the sense of mapping some abstract object to all of its prior instantions. If we interpret "Bedeutung als Gebrauch in der Sprache" in this first sense of "Gebrauch" we arrive at the conclusion that Bedeutung is constituted by all past occurences in the language. The Bedeutung of "apple", being its use in the language, can be mapped to all past occurences of "apple" in the language. These past occurences define "apple" and are identical with the word's place in our communication (I use "place in our communication" to avoid the more philosophically difficult "meaning").
When I speak of a dynamic-description here, I am talking about the sense of "use of" which might lead us to think of "Bedeutung" as defined by its relationship with other utterances in a language. The word "Bedeutung", as all other words and sentences, has certain relationships with many (or perhaps all) other terms and sentences understood by its speaker. "Bedeutung" has relationships with "Wahrheit" as well as with "apple" as well as with "That apple is red.". When we say that the Bedeutung of "apple" is its use in the language in the sense of dynamic-descriptions, we are saying that the Bedeutung of "apple" is constituted by the relations it bears with the remaining extension of the language that we were dealing with when we said that it was defined by its use 'in the language'.
I propose, then, to interpret Wittgenstein as using "Bedeutung" as pertaining to meaning, not just to reference, and "Gebrauch" as pertaining to relationships, not to the quantitative compounding of past occurences. A translation-cum-explanation of the passage I quoted above might be:
For a large class of cases in which the word "meaning", as both sense and reference, is empoyed--thought not for all--one can define this word in this way: the meaning of a word is defined by the relations it bears with the other terms and sentences encompassed in the language being used by the speaker (PI §43).
I will now further explain the two choices in translation I have here made.
I choose "meaning" rather than mere "reference" to represent "Bedeutung" for the simple reason that Wittgenstein's claim is otherwise puzzling. If he is taken to be saying only that words refer to the relations they have with other words etc., then what he is saying is difficult to understand unless we are also in a position to see his argument concerning the larger extension of "meaning". It is philosophically unsound to claim that reference is a relation, but that sense is not. If reference is fixed by relations, but sense is fixed by, e.g., some reified entity, then we are in a semantic position entailing absurdity. If "apple" refers only to a relationship, but it is made sensible in terms of an abstract, then we must wonder how it is that "apple" refers to a relationship at all.
I choose the second sense of "use of" for similar reasons of charity. One can't tenably hold, I think, that the meaning (as both sense and reference) of a word is all prior instantiations of that word. Certainly, past usage plays a role, but present usage is much more important. What would it mean for a word to refer to all cases of its past use? What sense of "reference" would we be talking about here? This is especially poignant when we consider such words as "today". Does its present use refer to the way it was used yesterday as well as the way it was used this morning? The most pertinent problem for this definition of "use of", however, is that it gives us a definition of meaning that is viciously circular. If meaning is prior use, what is the content of that prior use to which we are mapping meaning? It can't be prior use again, and then again, for the regress here is clearly evident. If defined in this first sense of "use of", meaning turns out to be a vacuous concept for the reason that we are mapping the maning of a term back to a set of its past occurences in which the word was, then, just as meaningless as it is now.
That the second sense of "use of" is not subject to this problem of vacuousness can be seen by looking at the way in which a word has relationships. If we map meaning to relationships, we are dealing with something that is present (and a presence). We aren't mapping back to something in the past which, by definition, must be mapped again to something prior. We are mapping to an actual set of relationships that a word has right now. Further, we must take these relationships as self evident, and not as non-existent and vacuous, for the simple reason that our words do have relationships and we do understand them. One asks, "Two and two make what?" and the child ecstatically replies, "They make four!". One utters "Hello." with certain sentential expectations in mind, and more often than not these expectations are fulfilled and the addressee of this utterance says "Hello." as well.