Structure and intention in language: a reply to Knapp and Michaels.
This is not a trivial mistake in their reply but colors everything that follows. Thus they write, claiming to represent my view, "it makes sense to regard interpreters as having to choose between interpreting those marks as the token of a sentence type and as the products of Wordsworth's intentional action." But on my view there is no way you could interpret the metaphorical speaker meaning if you did not already know the literal sentence meaning. That is, if you want to understand the speech act performed in the utterance, you can't choose between interpreting the marks as a token of a sentence type and as the product of an intentional speech act, because the way the speaker produces an intentional speech act - literal, metaphorical, or otherwise - is by producing a sentence that has a literal meaning quite independent of any speaker's particular intentions.
In spite of all the debates in linguistics and the philosophy of language there is a widely accepted picture, some parts of which I was exhibiting in my article: languages, such as French and English, are conventional or rule governed. The rules determine both what counts as a sentence (syntax), and what sentences mean (semantics). Speakers use these meaningful syntactical objects to perform intentional speech acts. Sometimes the speaker meaning coincides with the preexisting sentence meaning, sometimes not (pragmatics). But if complex communication is to be possible at all there must be sentences with shared conventional meanings that exist prior to the performance of this or that speech act. Languages are human creations and therefore dependent on human intentionality, but it is not the case that every sentence requires an intentional utterance in order to be a sentence. Knapp and Michaels wish to challenge this picture by denying that there are any sentences or sentence meanings that exist independently of particular historical speaker's intentions. I cannot see that they give any valid reason whatever for challenging the traditional account. Indeed I believe there are a number of significant errors in their reply to me, but for the sake of brevity I will confine myself to pointing out the following half dozen:
1. In their original article they said that a sequence of marks could not be a sentence, was not even part of "language," unless produced intentionally. In their reply to Wilson, they retreated from this view, and now they want to go back to the original view. But it is still a mistake. On the standard definition of sentencehood, sentences are purely syntactical objects. They are defined as well formed formulae according to the formation rules of the language. Sentences could not be identified with actual intentional productions of strings because a language such as English contains an infinite number of sentences but there can be only a finite number of actual human intentional productions.
There are a number of decisive objections to defining sentencehood in terms of actual human intentions, but this is the simplest and most decisive. Any natural language contains an infinite number of sentences, but there is only a finite number of actual human intentions. Of course it is always open to Knapp and Michaels to come up with some nonsyntactical definition of "sentence," "word," and so on. But what is their account? And what is the point? You still need some way to describe the output of the infinite generative capacity of natural languages and as a matter of logic that cannot be done in terms of actual, and therefore finite, human intentions.
2. They claim, "to treat a set of marks merely as a string of sentence tokens is to treat it in such a way that the number of its meanings becomes, quite strictly, infinite." Their argument for this is that there is an infinite number of actual and possible languages in which the marks might be sentences with different meanings in the different languages.
But this conclusion does not follow from the premise. A sentence is always defined relative to a specific language, and relative to a language a sentence always has a restricted range of meanings. There is a minuscule number of strings that count as a sentence in more than one language. You will see how rare this phenomenon is if you try to think of a string that is both a sentence of English and a sentence of German, say. In any case, the possibility of inventing new languages has nothing to do with the fact that any given sentence has a restricted range of literal meanings relative to the language of which it is a sentence. Given that there is a finite range of actual languages, and sentences have meaning relative to the language of which they are a part, it is just false to say that the number of meanings of any string is "strictly infinite."
3. In support of their claim quoted in part two above, Knapp and Michaels say that a string is only a sentence of a language if it is intended by its producer to be a sentence of that language. According to them, you can't tell what language a sentence is a sentence of without knowing the intentions of its producer. Indeed it turns out "to be a necessary condition of any set of marks' being a sentence token of any sentence type in any language that it be intended by its producer to be a token of that type."
But on the standard account of sentencehood, this is just false. A parrot for example might produce a string which is in fact a sentence of English (for example, it satisfies the formal conditions for being a sentence token of English) without the parrot having any of the relevant intentions at all. The parrot does not even have the concepts of English, tokens, types, and so on. Again, it is always open to them to provide an account radically different from the standard accounts of well formed formulae that go from Frege to Chomsky, but they provide no such account and no motivation for abandoning the standard accounts. Furthermore the quoted passage above is an odd claim for them to make, given their overall position, because it looks as if they are granting the existence of types and language, apart from specific speakers' intentions. What do "type" and "language" mean in the above passages?
4. In a passage that is very confused they write as follows: "Perhaps Searle would want to say that what gives the wave-poem its meaning is not the author's intention to follow some particular set of rules (since there is no author) and not the interpreter's ability to follow some rules (since the interpreter is not creating the meaning) but the fact that the marks themselves have followed the rules. But if the rules are rules that marks can follow, then they aren't rules at all - they're laws of nature."
Well, to begin with, formal structures don't follow rules, but they do satisfy or conform to rules. They meet conditions set down by rules. And this has nothing to do with "laws of nature." Everything in nature conforms to laws of nature, if things didn't conform to laws of nature, the "laws" would not be laws. So the choice is not between following rules and following laws of nature. Rather the question for any putative sentence is: does it satisfy the formation rules of the system? If it does, the string is a well-formed formula (that is, in natural languages, it is a sentence) if not, not.
5. An obvious objection to their view is that computers without intentions generate sentences. In response to this, in a long footnote, they claim that the sentences generated by the computer must be programmed, it's really the programmer's intentions that make the string a sentence. But this will not do. Suppose my cat wanders over the keys of my computer keyboard(1) and out comes a perfect English sentence: "The chair is made of wood." By any of the standard criteria this is a sentence of English.
6. They have some fairly heavy going about strings that depart in varying degrees from the syntactical or orthographic norms. There are indeed lots of cases where a string departs from the ideal norm in various ways; and in real life, as they correctly maintain, we will appeal to the producer of the sentence to find out what he or she intended, what sentence they were trying to produce. But it does not follow from this fact that the notion of sentence itself is defined in terms of intentional productions. In a way, the example shows just the reverse, we are puzzled by the deformed cases precisely because we have an antecedent notion of the ideal and we don't know whether to treat the deformity as approximating this or that ideal sentence or word.
To conclude: Why does all this matter? Who cares what a sentence is? Well, if you don't get the details right you won't get the overall picture right. And if you don't get the overall picture right, you are going to make a lot of mistakes down the line. The overall picture is this. Languages are human institutions and thus are intentionalistic through and through. But it is a mistake to confuse the intentionality of the system with the requirement that every element of the system be the product of an actual human intention. The intentionality of the system is precisely such as to allow the existence of an infinite number of sentences, and this in turn implies that something can be a sentence without having been produced by any actually existing historical human intentions. The sentence is the standing possibility of an intentional speech act. That is what sentences are for, to talk with. Knapp and Michaels have it exactly backward. On their view, only the speech act can create the sentence, for without the production in the speech act nothing is a sentence. I do not believe that on such a view you can give a coherent account of the relations of syntax, semantics, and pragmatics. But in any case they have not tried to do so nor tried to come to terms with the difficulties. This is illustrated by the mistake in their very first sentence. They think that when a speaker intentionally utters a sentence it "acquires" a speaker meaning.
I take it that we are all three, Knapp, Michaels, and myself, agreed that the way to understand any actual speech act or string of speech acts, literary or otherwise, is to try to figure out what the author meant. But I believe that in order fully to understand this valid point, it is essential to keep clear the distinctions between languages as formal systems of syntax and semantics on the one hand, and the intentionality of speakers' meanings on the other.
(1) I owe the example to Herman Cappelen.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||25th Anniversary Issue; response to Steven Knapp and Walter Benn Michaels in this issue, p. 669|
|Author:||Searle, John Rogers|
|Publication:||New Literary History|
|Date:||Jun 22, 1994|
|Previous Article:||Reply to John Searle.|
|Next Article:||Postmodern love: questioning the metaphysics of desire.|