Response

The weakest passage in this retrieval is the Herbert, which is not Herbert building a temple at all but Herbert's name attached to a joke about Puritans disliking holidays — a text that operates entirely through the social occasion of its author's surname. It is the least relevant to the stimulus's question about poems that function without a speaker present. But its weakness is instructive. The stimulus asks about texts designed to operate like architecture, indifferent to who inhabits them, and the retrieval has returned almost nothing that actually does this. What it has returned instead are poems about the memory of such structures — poems that mourn or reconstruct or hypothetically rebuild the temple from outside it. Hardy's shawm mourning its own "holy calm" in "Sabbath sanctitude" is not liturgical poetry; it is an elegy for liturgical poetry, spoken by a museum exhibit. Clare's shepherds have "vanished all" and their old music is left "like a vagrant bee, / For summer's breeze to murmur o'er, and die" — Clare. The songs persist as fragments precisely because no one is left to perform them, but Clare's poem is not one of those songs. It is a named poet grieving their anonymity. The retrieval, in other words, has located the nostalgia for the temple rather than the temple itself. This is not a failure of search. It may be a fact about the canon: what survives into a collected-works tradition is not the anonymous liturgical text but the authored poem that looks back at it.

Keats is the most honest about this. The "Ode to Psyche" arrives "too late for antique vows, / Too, too late for the fond believing lyre" — Keats. The temple whose holy forests and sacred fires once operated without requiring a named devotee is gone. What Keats proposes instead is to become the temple himself: "So let me be thy choir, and make a moan / Upon the midnight hours" — Keats. The list that follows — "Thy voice, thy lute, thy pipe, thy incense sweet... Thy shrine, thy grove, thy oracle" — is an inventory of liturgical furniture, every item preceded by "thy" as if Keats is cataloguing what the temple contained. But the whole structure depends on the first-person verb: let me be. The temple has been converted back into a performance. The architecture requires an agent after all. This is the precise inversion of the stimulus's Kafka observation about forms so robust they survive the substitution of their contents. Keats's form is so fragile it requires the poet to substitute himself for every element — choir, shrine, oracle, prophet. He is not building a temple. He is performing one, alone, at midnight, and the performance is what makes the poem an ode rather than a hymn. The hymn does not need Keats. The ode needs nothing else.

What the stimulus is actually looking for — the anonymous carol, the inscription, the text that operates regardless of who speaks it — is absent from this retrieval and probably scarce in the corpus as a whole, for reasons that are structural rather than accidental. The canon as it has been collected and digitised is a canon of authorship. Anonymous texts survive in anthologies as curiosities, their anonymity treated as a deficiency (author unknown) rather than a design feature. Hardy comes closest to the alternative: his dead quire sings "words of prayer and praise / As they had used to sing" — Hardy — and the poem's power is that the quire functions after death exactly as it functioned before, because the liturgical form never depended on the singers being alive. The form is the agent. But Hardy frames this as phantasy, as haunting, as something that requires the supernatural to explain. He cannot quite believe that a form could operate without persons. Neither, it seems, can the rest of the tradition. The corpus registers that the temple exists. It keeps writing poems about standing outside it.