The English-Learning and Languages Review Æ Homepage
Language, Linguistics and Philosophy
following article first appeared, in substantially the same form as here, in
the journal Cogito, Vol.13, No.3, Winter 1999.
There, the introduction refers to an article by the same author, entitled
“Language and Philosophy”, that appeared in Cogito, Vol.5, No.3, 1991,
in which Wittgenstein’s “picture-theory” of linguistic meaning is
discussed. This later article is, in
some ways, a follow-up to the earlier one, in trying to provide a more general
bridge between linguistics and philosophy.
In this version, the titles of works are given in full in the body of the
text, rather than by date-reference to the bibliography. The article is reproduced here by kind
permission of Carfax Publishing, Taylor and Francis
This article is an attempt to bring together the worlds of philosophy and linguistics in a general way by providing some principles that would underlie any formal analysis of language.
I would justify the need for such principles by pointing to the increasing overlap between the two disciplines in the twentieth century, particularly the latter half, where you find grammarians justifying their theories through appeals to philosophical ideas about the nature of language, and philosophers claiming that some philosophical issues are part and parcel of the linguistic forms in which they are stated.
I could not pretend, however, that the ground rules I shall be suggesting here would satisfy every current theory about language, since there are deep disagreements both within and across the two disciplines. My own sympathies will become clear as the article proceeds.
Here and there I have supplemented my text with quotations from the works of Wittgenstein. I have done so not with the intention of validating my own ideas thereby or of explaining Wittgenstein’s, but rather with the hope that the lucidity of his words may clarify mine.
The topics that are covered are: metalanguage, linguistic self-reference, the definition of a sentence, the principles of analysis, meaning and context, analysing meanings, analysing sentences, understanding meanings, the independence of sentences and their meanings, the purpose of linguistic analysis, the limitations of terminology, the idea of linguistic acceptability, the reality of language.
One or two of the ideas in the final section may be found expressed in a different and longer form in two recent articles (see References), in which I consider the claim that the structures of human language are genetically innate.
Writing or speech that has language as its topic has no special status. There is no ‘metalanguage’. When language is the topic, there is not a different type of relationship between the topic and the language used in writing about it from that between any other topic and the writing about that. The relationship is not between language and language, but, as in all other cases, between author and topic.
The fallacy of metalanguage is discussed by J.W.Miller ( A definition of the thing). He writes, for example: "A higher language can be a language only as it secures a still higher formulation. That formulation saves it from being an object....But no assertion of the sort ‘This object is a language’ is ever possible."
Whereas people can say things about themselves, words cannot. What you understand directly from a piece of language, therefore, is not anything about it, such as its function; what you understand from it is what it means. Nothing in language can be its own topic. For, if it must be true that one and the same thing can be referred to by two identical sentences, this would be impossible if each of two identical sentences referred to itself and not to the other one. A more formal explanation may be found in Wittgenstein’s Tractatus after the statement "No proposition can make a statement about itself" (3.332).
The whole discussion about self-referring sentences could be disposed of if it can be agreed that, when we say that some words refer to something, that is just a misleadingly shorthand way of saying that, through those words, some person refers to that thing. It can no more be said that words can refer to themselves than that they can have wishes. Wittgenstein argues (Zettel, 691) that the correct response to anyone claiming that some sentence containing the words "this sentence" is self-referring is to ask "Which sentence?". I would extend Wittgenstein’s criticism and ask also "Who said so?".
Although schoolchildren may be given grammatical conditions to help them write well-formed sentences, it is better for linguistic analysis to have a more elastic definition based on meaning, so that something taken as the expression of a complete idea may be called a sentence. Let us therefore define as sentences the divisions of a piece of writing or speech into its smallest parts that each make complete sense. The function of a sentence is to produce its meaning, which it does when it is understood by someone.
To see how this definition might work, take the conversation: A: Where was the duck? B: On the pond. There, I would call B’s reply, On the pond, a sentence, whereas if the reply had been The duck was on the pond I would not say that the last three words were a sentence.
A sentence may be thought of either as something that has occurred uniquely in a particular context or as something that has occurred, or might occur, in more than one context. We need not think that there are, for that reason, different categories of sentences, but what we must understand is that sentences are real and are things we hear, speak, read and write, and are not abstractions of speech or writing, as if they were some purer form of their realization. Language is what it is and should not be imagined different to satisfy a linguistic theory about it.
It may be useful to state here that, for the sake of logical consistency, I shall always be using the word “different” of separate things between which you wish to distinguish and the word “identical” of separate things between which you do not wish to distinguish. I shall not be using “identical” to refer back to one and the same thing.
An aim of analysis is to assign comprehensible terms to the resultant parts. With the proviso above about “different” and “identical”, we can give the main principles of analysis briefly as follows: one and the same thing cannot be described by different terms in one and the same description, and things you wish to regard as different from each other cannot be described by the same term; so a term describes only what you apply it to and you cannot say that terms you give to things you regard as different are identical terms.
The function of any part of a sentence cannot be known without your knowing the whole sentence and the context in which it was understood. If a sentence has topics, what it says about them is what the meaning, or part of the meaning, of the sentence is. Part of such a sentence that simply refers to a topic has no meaning. So, for example, in the sentence William defeated Harold, there are topics (for example, the person ‘William’) and what is said about them is, in whole or in part, what that sentence means, but in that sentence the word ‘William’ has no meaning. Except where a word is itself a sentence, it has no meaning. ("A proposition is articulate. Only facts can express a sense; a set of names cannot" - Wittgenstein, Tractatus 3.141).
The meaning of a sentence is produced, not by an accumulation of the functions of its parts, but by their combination in the sentence as a whole. The function of a part of a sentence can be known, or deduced, only when the sentence has produced its meaning. Nothing that can be part of a sentence, therefore, can have an inherent function unless it occurs only in one sentence that always has the same meaning. If we imagine that the understanding of a sentence is a sort of process or event that produces results and that those results constitute the meaning, then even if the difference in results is only in the number, rather than the type, then we can still say that something that can be part of a sentence can have a different function on different occasions.
Just as a part of a sentence has no function in producing meaning except in the whole sentence, in the same way a sentence has no meaning except in a context; and so no sentence has an inherent meaning unless it can occur only in one context. A sentence that is different from another is capable, depending on the context, of having a meaning that is different from the other’s.
For one sentence to be in a different context from another is not grounds for saying that the sentences are different from each other and, since no sentence has an inherent meaning, identical sentences might have different meanings in different contexts, and one and the same sentence might have more than one meaning in the same context, whether it was intended to or not by the speaker or writer of it.
There is a distinction in kind between expressing the meaning of a sentence in some other way and describing or analysing the meaning of a sentence. For linguistic analysis, descriptions of meanings must be expressed using the terminology of concepts. They cannot be descriptions of the physical states of those who understood the sentence.
To provide different descriptions for all possible meanings within language cannot be an aim of linguistic analysis. The meanings of some sentences may be imagined as analysable into concepts. The description we may give to one such part can be called its term. The purpose of the analysis will determine what can and should be given terms and which parts are to be regarded as identical with each other. Identity and difference among terms will correspond exactly with identity and difference among parts of meanings.
Understanding a sentence implies the consequences of perceiving it. A sentence’s meaning therefore is all the direct effects of perceiving the sentence. Strange though it may sound, therefore, we may say that meanings happen in the brain but, although we might imagine the descriptions of meanings as descriptions of neural occurrences, it is certainly not possible at this time in history to make such descriptions that could be understood as being those of the meanings of sentences. I suspect too that it will never be feasible.
Whereas concepts are not real things, sentences themselves are real and perceptible. The analysis of a sentence must leave no doubt, therefore, about the spatial or temporal arrangement of the parts it describes, whereas the description of the meaning of the sentence will simply show how the concepts are related to each other.
To provide different descriptions for all perceptibly different sentences cannot be an aim of linguistic study. Since the terms will depend on your understanding the meaning of the sentences, it may be that you will wish to regard some perceptibly identical parts as different from each other or perceptibly different parts as identical with each other, in which cases they will respectively have different terms or share the same term.
The descriptions of sentences are independent of those of meanings and vice-versa. This can be inferred from what has been said above, but may be summarized as follows (though perhaps in a way not suited to ordinary language): unless no sentence has a meaning that can be described identically with that of a different sentence, a sentence whose meaning has the same description as that of a different sentence can have a different description from that of the other sentence.
In the analysis of a sentence (not of its meaning) anything may be taken as one part of it, provided that one and the same part is not taken as more than one part for each description of the sentence. So, one part could contain, divide, be contained by and be divided by other parts. You cannot regard a part of a sentence as different from itself; and so in any one analysis you cannot give it different terms. The term you give to a part of a sentence will show what you wish to say about that part for that description of the sentence. Different descriptions of the same sentence, where different terms may be given to the same part from one description to the other, cannot be joined together to make a single description of the sentence.
The guiding principle of semantic analysis is that there are no inherent meanings. Languages have variety and they change - probably because humanity is various and changes - and while senses may seem to stick to words for a time, there can be no predicting how words are used. ("One cannot guess how a word functions. One has to look at its use and learn from that." - Wittgenstein, Philosophical Investigations, 340.)
Synchronistic analysis can serve some purposes, but cannot give the true picture. There has, in the latter half of the 20th century, been a blindness among some influential writers on language to the chanciness, necessary inefficiency and aesthetic causes that make language what it is, and this blindness has come from a drive to rationalize the nature of language by tying a theory to it. Yet theories are not what language-study needs. There is a huge fund of linguistic data, and our better understanding of it will come, not from the straitjacket of a theory, but from percipient and illuminating explanations of how it comes to be as it is. The mistaken view I have referred to above has led to the idea that your understanding of a sentence is compelled by the structures of language. The correct view, in my judgment, is that it is a result of the interrelation of many of your experiences up to that moment.
In this context the word ‘understanding’ is used neutrally; it does not exclude ‘misunderstanding’. When the two are contrasted, that brings in the intention of the writer or speaker, and the meaning of the sentence then could be defined as the intention of its originator. Interesting though the problems are that this raises, it seems better to take the sentence as that which brings about what it expresses rather than as that which is brought about by what it was intended to express.
In the definition of meaning, therefore, we are concerned neither with objective descriptions of physiological realities nor with descriptions that may arise from subjective reflections about ourselves as the originators or receivers of sentences. What is more important to clarify is that, although we commonly speak of a sentence as ‘having a meaning’, there is, of course, nothing in a sentence other than what it physically consists of. A sentence does not contain its meaning. This may seem a trite thing to say; yet some writers on language appear to assume that meanings inhere in sentences, particularly those who claim that meanings and sentences can be shown, through procedures of analysis, to be interdependent. This claim reveals a mistaken view of the nature of language and of how language is understood. The correct view, as it seems to me, is put by Wittgenstein in the Tractatus, albeit in considering a hypothetical language whose structures are limited to the unambiguous expression of possibilities of physical reality. ("The proposition includes all that the projection includes, but does not include what is projected. A proposition does not actually contain its sense, but does contain the possibility of expressing it.... A proposition contains the form, but not the content, of its sense." - Wittgenstein, Tractatus, 3.13).
Formal semantic analysis requires comprehensible comparisons among meanings. There must therefore be sufficient generalization together with an exactitude of terminology. Formal semantic analysis is therefore limited in its scope. We may be able to be exact about generalizations when the topics are external physical objects; we are less likely to achieve much precision when they are emotions, and formal analysis of this sort is almost certainly going to tell us nothing of what is important in the meanings of works of literary art. This is not to deprecate formal analysis, but simply to indicate what its limitations must be.
The independence of sentences from meanings implies a necessary independence of their analyses. The two may be set against each other, or side by side, to see how the language works, but any fusion or linking of the two must be theoretically ill-founded; nor can there be predictions from one to the other. This principle seems to me well expressed by Daneš (Travaux Linguistiques de Prague): "... strict differentiation of the two levels (syntactic and semantic) is indispensable. That does not mean a separation of levels, but only a methodological step which enables us to ascertain their systemic interrelation."
This independence does not imply that a sentence and its meaning are separate phenomena. Meaning is implicit in understanding, just as number is implicit in counting; and it is the physical form of the sentence that is the object of your understanding. Understanding a sentence therefore is the same as understanding what it means. Our everyday usage would be more illuminating if, rather improbably, we talked, not of ‘the meaning of the sentence’ but of the sentence ‘being its meaning’, rather like saying ‘virtue is its own reward’. Some influential theories of the fairly recent past have held, by contrast, that there are logical connections, that can be formulated, between form and meaning. While, indeed, your understanding of a sentence will determine how you analyse it, that is a far cry from saying that the form of the sentence is derivable from some other structure coming from your rationalization of the sentence’s meaning.
On the assumption that the point of linguistics is our understanding language better, we must count it as illogical to say at any time that it is because we know we have got closer to understanding it perfectly that we know we now understand it better. It would be illogical therefore in deciding between explanations in linguistics to say that one was better than another on the grounds that it gave a better account of the nature of language. That could not correctly be a reason, but only a logical consequence if you were right. Since language is not hidden, we are not seeking in linguistics an accurate description of it, since that would merely be the closest copy. What we are seeking are descriptions that help us to understand it better, or to understand its nature. The two aims in linguistics are accuracy of knowledge and the best way of considering that knowledge. The two feed each other, but are to be judged differently: the former can be tested empirically and the latter is judged according to how illuminating or useful you find that interpretation.
The terms describing parts of sentences carry no implications about the functions of those parts towards producing the meaning of the sentence. Grammatical terms in the analysis of sentences are only labels: they do not imply anything about states of affairs in reality, and are connected with semantic concepts only insofar as some are derived from what sometimes could be or could once have been one of their functions.
So, for example, the grammatical terms Singular and Present have only that connection, and not a necessary one, with the concepts of singularity and present time. We must not infer from the mnemonic origin of some grammatical terms what the functions might be of whatever the terms are assigned to.
So, to describe a part of a sentence as Adverbial Clause or preposition or vowel or Third Person Plural Pluperfect Subjunctive Passive, or whatever else, does not define its function ("Grammar does not tell us how a language must be constructed in order to fulfil its purpose, in order to have such-and-such an effect on human beings. It only describes and in no way explains the use of signs." - Wittgenstein, Philosophical Investigations, 496).
It is not for grammar to turn up its nose at anything. So long as you are confident that a sentence was meant (and that includes lies, quotations and woolly jargon) and that it means something, then, although there might be much to say about it, such as why you like or dislike it, how it could be expressed differently, when and to whom it would be thought suitable or unsuitable, one thing you cannot say about it is that it is ‘acceptable’ or ‘unacceptable’ as a sentence of the language, though it may be acceptable or unacceptable for one reason or another to this person or that. There are no criteria, however, for making such a judgment on purely linguistic grounds. This notion of linguistic ‘acceptability’ has been invoked by some grammarians who offer the circular argument that a sentence is acceptable if it conforms with the ideal set of rules that eventual complete success in the grammatical analysis of the whole language would produce.
Of course, you can have preferences about all kinds of linguistic usage. It is only human and entirely healthy that judgments should be made about the use of language, and the expression and discussion of how one’s language should be used and how it should develop must be a vital part of civilized culture. One hopes too that learners of a foreign language (and of their own) will be taught what are considered the best forms of that language. There is, however, nothing in language itself from which we can decide about such things.
The acceptability of a sentence as part of language depends on the intention of the originator of it: if the sentence was meant, it is part of language. We cannot therefore accept a concocted or invented linguistic example as part of language unless it was intended to produce its meaning in some real context. You may use such examples to teach someone something about a language, but they will not be a real part of language from which you can draw any linguistic conclusions.
The concept of language as something abstracted from the actual use of it is mistaken. Language is no more than the words that people have said, heard, written and read. Grammar is not something in itself; it is not any part of language, nor does it shape it in any way. Grammar is just ideas about language for talking about it from a certain standpoint - from a grammatical angle. This ought not to need saying; yet much of the linguistics of recent times has presented things otherwise: that grammar is something to be tested against what happens in language or even against what might happen.
That deterministic view of grammar is in line with the notion that there is some biological constituent of human beings that makes it inevitable that human language-use is as it happens to be. This is the theory that linguistic structures are genetically innate. It must be said that there is as yet no physiological evidence to support such an idea and how one would recognize a correspondence between a grammatical structure and a genetic one has not yet, to my knowledge, even been hinted at. It is at present just a speculative attempt by some people to rationalize what they see as certain regularities across all languages. The linguistic evidence that drives some people to propose a genetic cause for it strikes me as unsurprising, and I am at present content to regard it as what humans are likely to do because of how they have developed rather than as what they are predisposed to do because it is part of their development.
Since language is such a vital part of human culture, it must follow that how we regard it is vital too, and that we should not treat ideas about language as merely a question of intellectual fashion, or assume blindly that bad ideas are bound to provoke good ones in revolt. There are principles to be argued over, one of which I hold to is that language should not be analysed as if it were an abstract construction that allowed possibilities, some of which have been realized, but that it is something in which there are no possibilities apart from what has actually happened in it.
Finally, for those who still think that meaning can be objectified, here is one last quote from Wittgenstein:
"The sense of the world must lie outside it." (Tractatus, 6.41).
BULLEY M. (1991). Language and philosophy. Cogito, Vol 5 No 3.
BULLEY M. (1997). A wild gene chase. English Today, Vol 13 No 3.
BULLEY M. (1998). Not to be found in the brain. English Today, Vol 14 No3.
DANEŠ F. (1964). A three-level approach to syntax. Travaux Linguistiques de Prague, Vol.I.
MILLER J.W (1980). The definition of the
WITTGENSTEIN L. (1961). Tractatus
logico-philosophicus, tr. Pears and McGuinness.
WITTGENSTEIN L. (1953). Philosophical
L. (1967). Zettel, tr. Anscombe.
The Editor welcomes your comments or contributions to discussion of this article.
The English-Learning and Languages Review Æ Homepage