What is English if not a language?
Develops a distinction between language as something we have knowledge of and language as something we have beliefs about. This distinction is formulating using the philosopher's notion of an intentional object of belief. Together with the distinction between collective (= mutual belief) and distributive judgements, the argument seeks to dissolve false controversies between idiolectal (Chomsky) and sociological approaches to language.
In an essay The Misunderstanding of Newspeak, Roy.Harris surveys some of the mistakes which arise from Orwell's supposition that there is something called "the English language". But if there is no such thing as the English language, what are we doing when we use the expression "the English language" or, alternatively, What is English if not a language? And the same question for French, German, Tagalog, Hopi and four or five thousand others. My answer is that a language is, in the first instance, an object of belief not an object of observation or inference from observation, though linguistic objects of the latter sort exist as well. Let me expand on this.
An anthropologist, whose male native informant claims to be the reincarnation of a monkey, will record the informant as believing that he is the reincarnation of a monkey. But when the informant claims to be a native speaker of Hopi, the anthropologist will record him as a speaker of Hopi. If the anthropologist is an anthropological linguist, then he or she will go on to try to construct a grammar of Hopi. The problem which then always arises is that no two informants' versions of Hopi are identical. Linguists have, historically, responded to this problem by reconstructing the concept of a language in diverse but generally unsatisfactory ways (surveyed in chapter 3 of my book Language in Mind and Language in Society (1987)). The implausibility of these reconstructions, some of which bear all the fantastic hallmarks of attempts to.save a discredited theory (I am thinking of William Labov's probabilistic "community grammars") ought to suggest that something is fundamentally wrong in the general strategy. And indeed it is. The cause of all subsequent difficulties is the initial assumption that if someone says they are a speaker of Hopi then, in a straightforward sense, they are. If we recorded the claim to be a speaker of Hopi in the same way as the claim to be a reincarnation of a monkey, the difficulties disappear. To turn the Eurocentrism of this paragraph round, imagine now a Hopi anthropologist studying people who claim to be speakers of English.
Let 'S' designate any one of these people. Then for any S who claims to be a speaker of English (for simplicity, a native, monolingual and male speaker), a first-shot revised anthropological transcription of the speaker's claim is represented by (1):
(1) S believes he is a speaker of English.
This is hardly exciting. It becomes so if you are prepared to grant that (2) represents the logical form of S's original claim:
(2) S believes, of the language he speaks, that it is English.
In philosophical terms, "English" appears in (2) as the intentional object of a speaker's belief and it appears opaquely: from the fact that some S believes he speaks English it does not follow that he believes he speaks Engelska, since he may not know that "Engelska" is the Swedish word for "English". In contrast, "the language" appears extensionally and transparently in (2), so that if the language S speaks is actually (say) Jamaican creole, then we can substitute "Jamaican creole" or "creole de Jamaique" for "the language" without affecting the truth of (2). As this illustration perhaps suggests, the way things have now been reformulated allows that the language S speaks and S's beliefs about that language can vary independently of each other. This is a major theoretical gain. For it dissolves the linguists' problems with the "inherent variability of languages". The new formulation in (2) allows for the possibility that "the language he speaks" may vary from individual to individual - that is, in the linguists' terms, it may always take the form of an idiolect. But, surely, it may be objected, there must be some connection - indeed, an intimate connection - between the language S speaks and his belief as to what it is, for otherwise we should expect to find cases like (3) , but we don't
(3) S believes, of the German he speaks, that it is English.
First, it is by no means clear that we do not find cases like (3). After all, it is commonly said that some children of West Indian origin have encountered problems in the British educational system because they think they are speakers of English but their teacher's don't, Their teaachers rather think of them as speakers of creoles with insufficient resemblance to or mutual intelligibility with English to be counted as English (see Viv Edwards, The West Indian Language Issue in British Schools. Routledge Kegan Paul 1979).
Second, if you consider how, developmentally, speakers might acquire beliefs about which language they are native speakers of, it is intuitively obvious that they do not acquire them by inspecting the language they speak and matching it with samples of a named language. Rather, they are told which language they speak in the context, for example, of overhearing speakers of a foreign language for the first time ("They are speaking French. We speak English"). The belief is acquired on authority and, consequently, is a mutual belief: I believe I am a speaker of English because you believe I am, and I believe you believe you are a speaker of English, and so on.
Those two remarks made, it is nonetheless true that our language (or Knowledge of Language in Chomsky's phrase) and our beliefs about our language interact in complex ways. Without this interaction there would not be a politics of language - and there clearly is, as the endless encounters between professional linguists and indignant letter-writers attests. These interactions require very careful mapping, but that is only to say that the intentionalist approach to languages as objects of belief suggests a detailed research program in which linguists, psycholinguists and sociologists of language might be able to engage collaboratively without promptly falling out with each other.
First, of course, speakers have many beliefs and mutual beliefs about language in general and their own language in particular, many of which will be well-founded enough to allow them to be confident in guiding children and foreigners to acceptable linguistic practice in a given community. However speakers are constantly liable to over-confidence and to mistaking the source of their authoritativeness in any particular instance. Compare, for example, my well-founded belief that in certain cultural contexts it is not acceptable to use double negation ("I haven't got no money") with my intuition that (4) is ungrammatical:
(4) Who did you see the woman that met in town?
Now I am confident that readers of this essay will also judge (4) ungrammatical, but that does not entail that we are bound together by mutual beliefs (cultural knowledge) from which we authoritatively judge the ungrammaticality of (4) For in this instance it is possible that we judge alike because (4) shows the violation of a principle of universal grammar (I have taken (4) from David Lightfoot's The Language Lottery, MIT 1982, where it is used to illustrate that possibility). That we then all judge (distributively as the philosophers say) that (4) is ungrammatical is radically different from our judging (collectively) that double negation is unacceptable. In the former case, our "judgment" is actually an intuition which provides indexical evidence about properties of a nonintrospectible mental structure; in the latter case, our judgment is introspectibly related to our beliefs. Now in everyday practice we simply do not distinguish these two sources of our authoritativeness, and generally get along well enough without doing so. Sometimes, however, we are led to judge for others when we have no grounds either in shared beliefs or underlying structures. Thus, I well remember telling a technical college student in Exeter that he could not say "comprises of", but only "is comprised of" or "comprises", only to be confronted on next encounter with advertisements from every estate agent in town specifying "the house comprises of ....". Unwittingly, I had pitted my idiolect against local practice, the very error against which professional linguists like Peter Trudgill have rightly campaigned (in his Accent, Dialect and the School. Arnold 1975 and elsewhere).
However - and this is the second point about interaction between Knowledge of Language and Beliefs About Language - what sustains prescriptivists against the onslaughts of the linguists is the fact that beliefs about language can not only affect our speech and writings but through time enter into the grammar of the language we speak itself. It is evident, for example, that in cases of what are called hypercorrection someone's speech is partly controlled by false beliefs they hold about correct or proper or desirable speech forms. In England, a Northerner knows that Southerners pronounce "grass"" with a glottal (throaty) aaa - sound and think that the way to speak properly is therefore to amend their own native labio-dental /a/ to the glottal sound ending up with "gaaas" for "gas". (And how do you pronounce The Daily Telegraph?).
Through time, practices which for one generation are controlled by their beliefs can enter into the next generation's knowledge of language. This is because children do not hear that a particular piece of speech has been produced by adaptation of a native form; they hear only the output, the mental processes involved are unobserved. Consequently, they may come to produce the final form without the mediation of belief - based adaptive or transfer rules (the term "adaptive rule" is due to the Danish linguist Henning Andersen; "transfer rule" to the British linguist, Richard Coates). The best sort of answer to Roger Scruton's misleading reification of "a language structured by gender distinctions" (see Roy Harris's article for the context of this) is such facts as that my three year old daughter says "he or she" when she doesn't know the sex of a referent ("When our new baby is born, he or she will sleep in my room" - this is right, isn't it?). And her advantage over me is that she doesn't have to employ an adaptive rule to produce this form. (I should add that I have no views on what policy to adopt in a language such as French which really does have gender, le bebe being masculine. It is a matter of controversy among linguists what the relation is between grammatical gender (masculine, feminine) and sexual gender (male, female). Scruton seems to be unaware of the difference, which allows people over there to say "J' espere que le bebe sera une fille" without self-contradiction).
Third, and finally, and despite everything I have said, it is clear enough that the idiolects of speakers who believe themselves to be speakers of the same language do indeed cluster enough for the belief to be highly plausible. For most practical purposes, it is true to say that over there they speak French while over here we speak English. Some like Pierre Bourdieu (see his Ce que parler veut dire. Fayard 1982) would probably say that this clustering owes everything to the practices of modern national educational systems in reproducing national languages. Yet that is by no means self-evidently true, though it may be true for languages like Swedish, Danish and Norwegian which have been deliberately developed away from each other. It may be that this clustering can just as well be attributed to the operation of innate constraints on language development; Chomsky puts it this way in Rules and Representations (Blackwell 1980):
"I see no reasonable alternative to the position that the basic reason why knowledge of language comes to be shared in a suitably idealized population (and partially shared in actual populations) is that its members share a rich initial state, hence develop similar steady states of knowledge" (p.80).
Much current work, especially in the mathematical theory of language learnability, is devoted to fleshing out this claim, seeking to show how if the parameters of universal grammar are set in particular ways, specific language structures become highly accessible from a given linguistic input. But whatever isomorphism between speakers' knowledge of language and their beliefs about their language exists, it should not be allowed to obscure the major differences between these two orders of reality. The price of obscurity is that the arguments between professional linguists and indignant letter-writers will remain, as they have so often been, at cross purposes.
Written 1983 or 1984, lightly revised. Not previously published. A similar essay under the same title appears in J D Johansen and H Sonne, editors, Pragmatics and Linguistics. Festschrift for Jacob L Mey on his 60th Birthday, pp 137 - 140. Odense University Press, Denmark.