Cataloguing my books has gotten me dipping into volumes I’d forgotten all about, and yesterday it was Paul Goodman’s Speaking and Language (1971), which I bought and eagerly read in 1974 (it’s full of annotations) but hadn’t looked at in years. Goodman was (as Edward Said said in his perceptive NY Times review) “amateurish and utopian,” and here he takes a thoughtful amateur’s look at language and the attempts of linguists to corral and analyze it. He makes a lot of mistakes and says some silly things (the margins are full of my penciled question marks, “Huh?”s, and corrections—it’s Verner’s Law, not “Werner’s”), but he also had some very interesting and perhaps useful things to say, and I’ll quote a couple here. From Chapter III (p. 41 in my Vintage paperback):
Most often words do not fail a speaker; rather, he wrenches the words a bit and communicates. This does not mean that the constant supra-individual code is unimportant; on the contrary, it is all the more indispensable. Unless the speakers know the code well, they do not hear the modifications. Bloomfield speaks of “the fundamental assumption of linguistics, namely: In certain communities [speech-communities] some speech-utterances are alike as to form and meaning.” But it is how the speaker varies the code—by his style, the rhythm and tone of his feeling, his simple or convoluted syntax, his habitual vocabulary—that is his meaning, his meaning in the situation, which is all the meaning there is. This should be a platitude, except that it tends to be denied or brushed aside by linguists.
And the end of Chapter VII, “Constructed Languages”:
My bias as a man of letters is that it is best to do linguistics like natural history or art criticism, reasoned but a posteriori, rather than like mathematics, as is the current style. Present-day linguists and nineteenth-century philologists have made too much of a big deal of prediction, predicting the forms that will occur. It is soul-satisfying to have one’s prediction confirmed, but it is not terribly important except in sciences that are applied, as physics to engineering, and where the consequences must be controlled. The chief use of humanistic studies is to explain, to understand, to appreciate. And in linguistics we want to make sense precisely of novelty, unique appropriateness, history, and even accident—they are expected factors.
Let me make a close analogy. In literary criticism it is possible to define literary genres and predict from them. But in analysis it will be found that only hack works conform to the genre. Powerful works are sui generis; they sometimes set themselves absurd conditions and carry it off—consider a really crazy work like Handel’s Messiah. This does not mean that powerful works are incomprehensible; on the contrary, most (not all) excellent works are rather transparently demonstrated. But we cannot know one before it has been invented. Similarly, it is possible to take a statistical average of speech events and abstract a structure from it that has routine application. But this deceptively makes us think we understand something when we don’t; examples of excellent speech may not fall in the average—and I have been arguing that they are likely not to. The most intimate speech, the most convivial speech, the most expressive speech, the most poetic speech are likely to be “deviant.” But they are not deviant; they can be reasoned a posteriori.
Thus, I think that Roman Jakobson’s testy remark is wrong-headed, that “idiolect is a perverse fiction . . . everything in language is socialized.” The issue is not whether speakers have a private language—of course they do not—but whether good socialization (and good society) does not require spontaneity, concreteness, and invention in the intercourse of its members.
I miss Paul Goodman and his tough-minded humanism; we need more amateurs like him (and perhaps fewer specialists who can’t communicate).