Since the 1940s, the great majority of teachers of poetry have been telling their students that what authors intended to say is irrelevant to the meaning of literary works. All they have to pay attention to is the words that appear on the page. There is, as the saying goes, “nothing outside the text.” To wonder what authors might have been thinking about would be an irrelevant imposition on the free play of meaning. It would be an example of the “intentional fallacy.” With this consensus among literary critics in place, the tables were set for the legendary “Death of the Author” announced by Roland Barthes in 1968.
How did authors become personae non grata? It all started with a philosophical trend in the anti-mentalistic Forties, when prominent philosophers were hoping to solve many of their favorite problems by getting rid of Descartes’ concept of mind. As a result, anti-Cartesian doctrines like logical positivism, pragmatism, and behaviorism were in vogue, and mind-unfriendly philosophers like Gilbert Ryle and A. J. Ayer came to the head of the class.
At the same time, literary critics were feeling confined by the pronouncements of many prominent poet-critics, especially T. S. Eliot, who had the temerity to attach footnotes to the poem of the century, The Waste Land, pinning down its recondite sources and allusions. So eager were the critics to keep extra-textual evidence of meaning from limiting their interpretive freedom that they banished authors’ intentions from the works themselves. Literary works became examples of “literary language,” exhibiting a new-found liberation from clarity and singleness, an exhilarating inability to exclude any possible meaning. The ambiguity and multiplicity of literary language became the basis of a new defense of literature, highlighting its distinction from the allegedly too rigid jargons of science and commerce.
Later trends like structuralism and deconstructoin only deepened the confusions. I have addressed them elsewhere, and I am not going to revisit the issues here.1 But I do want to assert the antidote: Poems and literary works are not mere texts but utterances made by authors for a specific audience in a particular context. And this is true of ordinary language and its speakers as well.
So here is an example of a mere text that is not yet an utterance—“It’s midnight in Paris.” Speakers of English can recognize its standard use. They know under what conditions it would be true and appropriate to pronounce. But without a speech situation, it is strangely rootless.
What if you encounter this sentence as an actual utterance but without access to the context. Say you are at a café in the US and you hear someone at the next table say “It’s midnight in Paris.” Now, on the basis of the mere text, you have no idea what the point of this utterance might be. Is it a reminder that it’s too late to make a phone call to Paris? Has somebody said it’s too early for drinks, evoking the humorous response that, since it’s already midnight in Paris, it’s not too early to have one here? Is somebody identifying their favorite movie, Woody Allen’s Midnight in Paris? You don’t know. All you have is the standard, conventional meaning of the text, which gives you no idea of the point of the utterance. The speaker-meaning and context that make this an utterance rather than a mere text are completely opaque. But there must be a point that the speaker had in mind; otherwise not a word would have been said.
Such examples from the domain of ordinary speech, showing the difference between sentence-meaning and speaker-meaning, should be enough to show that texts without a grasp of the intentions behind them cannot be adequately understood—that the need to cope with intentions is not a fallacy. But it was more difficult to demonstrate this using poetry. The classic debates about authorial intention often depended upon thought-experiments asking, for instance, how we would respond to famous poems if we learned they had been generated by chance—by monkeys with typewriters, by waves washing up on shore. But now we have a new resource, AI-generated texts that have every appearance of being actual poems even though they represent no intentions and have no context. They are poetic texts but not poetic utterances. Here’s a free-verse ChatGPT poem for which I provided the title as a prompt.
How to Educate a Frog I. Pond Life Begin at the water’s edge, where cattails sway and lily pads offer platforms for discourse. Here, in this aquatic classroom, observe the curriculum of insects and algae— lessons not in books, but in ripples and croaks. II. Observation Watch the dragonfly, iridescent teacher, its flight a lecture on aerodynamics. Note the beetle, a study in armor and utility, and consider the tadpole—your own youthful reflection— its metamorphosis an education in becoming. III. The Art of Waiting Sit still, eyes like twin moons open to the sky and the canopy above. Learn patience from the heron, its unwavering focus a testament to intention. Remember: wisdom often comes to those who wait. IV. Communication Master the language of croaks and chirps, each utterance a word in a moist dialect. Speak not just to your kin but to reeds and fish, to the very water that sustains you. Expand your lexicon with ripples and splashes— the syntax of survival. V. The Limits of the Pond Contemplate the pond’s edge, that finite boundary, and what lies beyond—fields, roads, unimaginable terrains. While you are a creature of water and shade, let curiosity tempt you toward the unknown. After all, even a frog can dream of oceans.
As we saw in a previous post (“The Automated Pleasures of Hypocrisy”), it is clear that in no time my old friend Chad Geppetto can spit out a perfectly plausible facsimile of a humanly authored poem. He can mimic the metaphor-rich language of poetry, speaking emotional rather than literal truth, constrained by the laws of pleasure rather than the canons of logic. He carries out fanciful premises with unflinching consistency and lucidity, accomplishing exposition, development, and closure as if he knew why such patterns are effective, anticipating likely reactions to his messages, and striking a convincingly appropriate tone, all while leaving no obvious signs of digital handiwork (or of digits without hands!). This poem also shows that Chad can bring all kinds of specialized knowledge to bear, and further experiments make it clear that he can impersonate characters and mimic their points of view. He does something like what poets do—learning from many examples how to compose a poem—but he does so, remarkably, with no access to the world that prompted those poems or the feelings that they depend upon. Unlike the words of W. B. Yeats, his words will never be “rhymed out in love’s despair.”
“How to Educate a Frog” has a further feature of some AI poems—numbered sections with informative headings that appear without being prompted. Such sections are common in the history of English poetry, but they seem old-fashioned today, suggesting that, despite being prompted toward free verse, the machine is rather conservative in its models. AI poems do have a somewhat contemporary feel, though, because the machine picks up the whimsical character of today’s magical-poetic logic. It mixes registers with ease, combining the technical and the everyday with Audenesque abandon, coining phrases like “aquatic classroom,” “iridescent teacher,” and “syntax of survival.” Best of all is the frog’s “moist dialect.” These sallies would be felicitous in a humanly-authored poem.
It can’t be denied there are pleasures here. The humanly prompted scenario is carried out to comical effect and the prospect of hearing the frog’s “moist dialect” gives the batrachian singer a certain charm. The problem is that, not being an utterance, the poem has no point beyond what is explicitly stated. Its meaning is all on the surface. The machine uses the imperative voice directing the education of a frog but with no deeper purpose. Who, after all, would want to educate a frog anyway?
A human being writing this poem would have to be pointing to something beyond that fanciful idea. In genuine poetry, the most absurd premise still leads to some conclusion, or at least entitles us to look for one, but here there is no possibility of going beyond the explicit content of the text. (It must be said, though, that the machine does take a crack at a closing lesson, putting its money on a heart-warming frog-to-prince maxim: “After all, even a frog can dream of oceans.”)
Some types of poem—love poems, for example—have an obvious generic purpose; their motive is preset. But any poem worth reading will have ambiguities and unstated implications, and a fine poem will have a host of these. AI poems do not. And English professors, teaching poems as if they were mere texts, shouldn’t have been able to derive a point from them either. But they did and still do so—incessantly. Often the point they draw is how ambiguous language remains once you rule out the author’s intention.
Of course we could look for possible implications in any text. If you encountered “How to Educate a Frog” in a magazine you might find it comically didactic. You might wonder, Are frogs giving us a lesson here? Is this a genuine instruction for living, valid for human beings too? Or is it an ironic comment upon our need to find lessons everywhere? Is the ending genuinely hopeful and playful or does it mock our love of open horizons and happy endings? For a humanly authored poem, there would be answers to these questions, or the questions would at least be relevant, making uncertainty part of the experience of the poem. But with a machine-poem this is not the case. Ambiguity is beside the point because there is no point beyond the one explicitly articulated in the word-stream.
This, in my view, is what the “Death of the Author” would have felt like if poems really could be mere texts without an author behind them. The text is all you get and there is “nothing outside the text.” Anything that isn’t said just isn’t said. Many literary critics argued that texts without authors would be less constrained and offer a multiplicity of meanings. They were particularly keen to discover ironies, which depend upon the contrast between what is being said and what is meant. But poor Chad’s poems show us that mere texts without authors and a speech situation can rely only upon the standard meanings of words and the context of surrounding sentences. They can’t have irony because there’s nothing for the literal meaning of their sentences to contrast with.
This may seem like a dry, technical point and, to be fair to my professorial colleagues (if in a somewhat backhanded way), even though they claimed to be ignoring authorial intention, they mostly wound up being faithful to it in their fashion, with ritual apologies for breaking the taboo. But the “Intentional Fallacy” and “Death of the Author,” in my view, contributed to making the study of literature a forbidding and strange endeavor, a densely theoretical enterprise. And that was unfortunate for everyone except tenure candidates with a desperate need to go beyond the obvious.
I am reminded of what a great student of mine named Herbie, a Math and English major at one of CMC’s sister colleges, said to me some years ago: “In your class we actually discuss the poems.” An unexpected compliment! What she meant was that in some of her classes the texts were a mere jumping-off point for theoretical divagations. Indeed, in many classrooms, literary works become models not for experiencing the power of literary language but for escaping from its power by demystifying or “deconstructing” it.
Earlier I mentioned some of the short-term factors that gave anti-intentionalism a start—anti-Cartesianism and the irritating authority of poet-critics. But there were also longer-term tends involved, trends that continue today. The anxiety about authorial intention, the too close tethering of poetic meaning to the author’s mind, was part of a centuries-long reaction against Romantic psychology, whose defenders took poetry as the pure expression of great men’s souls. Poets, in the Romantic view, were the “unacknowledged legislators of the world,” as Percy Bysshe Shelley put it. Invested in the poet, Ralph Waldo Emerson tells us, “there is a great public power, on which he can draw, by unlocking, at all risks, his human doors, and suffering the ethereal tides to roll and circulate through him.” The results can be remarkable. The poet is “caught up into the life of the Universe, his speech is thunder, his thought is law, and his words are universally intelligible as the plants and animals.”
This is opening one human door too many. And if Emerson’s conception of the poet is excessively elevated, the tying of poetry too closely to the author’s psyche also threatened to bring the art down to the level of the poet’s personal life, burdening it with multitudes of sins. Unlike authors, texts engage in no extracurricular activities. They do not misbehave beyond the page. So, in a number of ways, authors had it coming.
For me, the antidote to the mutually aggravating set of enthusiasms and anxieties about poetic language and its relation to authors is the realization that poets do often speak as themselves but that when they do so they speak poetically. To speak poetically is to speak with the freedom that comes with the recognition, shared between audience and poet, that giving pleasure is the primary and essential aim of the utterance. Speaking poetically is a lot like speaking wittily; you can say things you obviously don’t mean literally as long as your audience picks up the fact that you are not being serious. Similarly, poetic license authorizes you to endow your speech with artificial patterns of sound and structure, make grand pronouncements, articulate temporary moods, try out opposing ideas, impersonate real or imaginary characters, and mock, cajole, and caricature other peoples’ points of view. In poetry you can address absent people, dead people, imaginary people, imaginary objects or real objects that couldn’t understand you even if they were present, and you can do all of this with varying degrees of seriousness or whimsy. Add to this the freedoms that go with storytelling in general and you begin to see why the poet’s voice can’t be muffled into mere text.
In his poem “The Choice,” W. B. Yeats writes
The intellect of man is forced to choose Perfection of the life, or of the work, And if it take the second must refuse A heavenly mansion, raging in the dark.
In these lines Yeats is obviously speaking with long-meditated, self-conscious profundity. If you are addicted to paradoxes and the treachery of language, you can look for ironies in the tension between force and choice, but that hardly seems to be Yeats’s point. Language, for him, is not an obstacle. The only trouble with language is that it’s so hard to forge into poetry.2
Yeats is a thinking poet with a grand cosmic-historical system behind him, and not to take his statements about the world as statements about the world is simply to miss the point of what he is doing. There is, of course, a cost to this. You cannot endorse the value of Yeats’s poem as yet another demonstration of the complexity or elusiveness of literary language. You have to engage with the value of what he is saying.
Things are slightly different with Philip Larkin. At the end of “This Be the Verse,” he writes
Man hands on misery to man. It deepens like a coastal shelf. Get out as early as you can, And don’t have any kids yourself.
The tone of this utterance is not as easy to pick up as in the case of Yeats. The first two sentences of this final stanza add geological gravity to a poem which starts with coarse humor about how one generation inflicts its pain upon the next (“They fuck you up, your mum and dad”), but how serious is the advice that comes at the end? Reading it with other Larkin poems in mind, I take it as a humorously vernacular and self-consciously inadequate response to a serious situation. Larkin is speaking poetically but in a different vein from Yeats. He is striking an inadequate but situationally understandable pose, expecting his readers to recognize that.
The upshot of this discussion is that questions about tone and meaning can’t be answered by applying general principles about texts and literary language. What is required is the careful weighing of words chosen by an author in context. Poetic language does not manifest a generic refusal of statement. Poets use it to play with the powers of statement and all the other things language can do, and it is up to us readers to figure out what they are up to on a poem-by-poem basis.
My most substantial treatment of it is in my book The Varieties of Authorial Intention: Literary Theory Beyond the Intentional Fallacy (Routledge, 2017).
This does not keep Yeats, of course, from exploiting the freedom of speaking poetically. In one poem he imagines himself divesting his bodily form and entering the pure art-world of medieval Byzantium (“Sailing to Byzantium”). In another we find him recharging his imagination by lying back down in the visceral realm of the bodily excitements, “the foul rag and bone shop of the heart” (“The Circus Animals’ Desertion”).