Repository path: workshop/translations/miserere/R07-v1/translation.md · rendered 2026-09-09

T-miserere-R07-v1 — Bécquer, «El Miserere», the vision, under R07 (fluency)

Source: as T-miserere-R08-v1 — the same 668-word span, the same copy-text. Regime: R07-fluency v1.0, rules F1–F10, frozen 2026-07-28. Interpretive rulings IR1, IR2, M1–M6, and the American-English declaration: ../rulings-R07.md, committed before the source was opened for translation, as R07's S073 binding condition requires.

Contamination: suspected, and the reason is not the published comparator. No published English of this tale was read before this rendering; the bounded Bates exposure in provenance.md §5 contains no English inside the span. The suspicion is of the lead against itself: this rendering was written after T-miserere-R08-v1, in the same session, on the same 668 words, by the same hand. Note (bhb) measures lead-against-self at up to 41 contiguous tokens — above anything this project has recorded against a published human translation — and no design has ever priced it. The self-overlap is measured after the freeze and recorded in ../dependence.md.

The direction of that bias, registered before the numbers exist. Writing the plain arm second can only pull it toward the elaborate one. Every prediction the study limb registers — that the R08 arm is read as the source-follower, and that it is read as the more elevated English — is a prediction that the two arms are far apart. Contamination between them therefore works against the design and not for it.

The translation

[P1] The last stroke of the bell had not yet died away, fading from echo to echo, and its ringing could still be heard trembling in the air. Then the whole church began to light up by itself: the stone hoods over the statues, the marble steps of the altars, the cut blocks of the arches, the pierced railings of the choir, the carved clover leaves along the ledges, the black uprights of the walls, the floor, the ceiling. No torch, candle, or lamp could be seen anywhere that would account for the strange glow.

[P2] It looked like a skeleton, whose yellow bones give off the phosphorescent gas that glows and smokes in the dark with a restless, frightening blue light.

[P3] Everything seemed to come alive, but with the jerking motion that electricity forces on a dead body, twitches that mock life — a motion that comes all at once and is more horrible even than the stillness of the corpse it shakes with a force no one understands. The stones settled back onto one another. The altar, whose broken pieces had been lying about in no order, stood up whole, as though the mason had just given it the last stroke of his chisel. The fallen chapels rose with it, and the broken capitals, and the huge shattered rows of arches, which crossed and looped through one another at random until their columns made a maze of red stone.

[P4] Once the church had been rebuilt, a far-off chord began to be heard. It could have been mistaken for the hum of the wind, but it was really a body of deep, distant voices that seemed to come up out of the earth itself, rising little by little and growing clearer all the time.

[P5] The bold pilgrim was beginning to be frightened. But his passion for everything strange and marvelous was still fighting his fear, and it gave him courage: he got up from the tomb he had been sitting on and leaned out over the edge of the chasm where the torrent jumped among the rocks and fell with a steady, terrifying roar. His hair stood on end.

[P6] Then he saw the skeletons of the monks who had been thrown from the church wall into that gorge. They came up out of the water badly wrapped in the rags of their habits, their hoods pulled close; under the folds of the hoods, the dark sockets of their skulls stood out against their fleshless jaws and white teeth. They caught at the cracks in the rock with the long fingers of their bony hands and climbed until they reached the top, and as they climbed they spoke the first verse of David's psalm in a low, hollow voice, with a heartbreaking sound of pain:

[P7] ¡Miserere mei, Deus, secundum magnam misericordiam tuam!

[P8] When the monks reached the porch of the church, they formed two lines and went in, and knelt in the choir, where they went on chanting the verses of the psalm in a louder and more solemn voice. The music kept time with their voices. That music was the far-off rumble of the thunder moving away and muttering after the storm had broken up; it was the moan of the wind in the hollow of the mountain; it was the steady noise of the waterfall on the rocks, the drip of seeping water, the cry of a hidden owl, the scrape of restless creatures. All of that was the music, and something more besides, which cannot be explained and can hardly even be imagined — something that sounded like the echo of an organ backing the verses of that enormous hymn of repentance by the psalmist king, in notes and chords as enormous as his terrible words.

[P9] The ceremony went on. The musician watching it, spellbound and terrified, felt he was outside the real world, living in that strange country of dreams where everything takes on shapes that could not exist.

[P10] A violent jolt shook him out of the daze that had taken hold of every faculty of his mind. His nerves jumped under the force of an overwhelming feeling, his teeth chattered, he shook with a trembling he could not stop, and the cold went into the marrow of his bones.

Translator's log

Frozen before any evaluation of this rendering was designed. Coverage codes per R07 §5, the strict test only — a rule names the feature and exactly one live option survives it. The exclusion-test rate is not computed, per the S073 binding condition; the mean number of live options per logged site is reported instead.

29 logged sites. Strict D at 25 of 29 (86.2%); P at 3; S at 2. Mean live options per logged site: 2.10. No INT flag — R07 has no such code.

A note on that 86.2%, because it is the highest strict cov(R07) this project has recorded and the reason is not that the rules got better. RS-20260730h §5 puts the four published strict rates at 11.4% / 5.6% / 6.7% / 12.7%, and RS-20260801 §11 found that the strict D codes cluster on F1, F5 and F7 — the rules that name a property of English and say nothing about the source. This span is a Latinate source read by a rule set with F1 and F2 in it, so almost every contested site is a register site of exactly that kind. The figure says the span is unusually favourable to this statistic, not that the rule set decided more than it usually does. Reported here, and reported again wherever it is cited. internal-judgment-only.

What F1–F10 did not reach. M2 (tenses) and M3 (the Latin) fired and are the two S codes. M6 (proper-noun knowledge) fired at Rey Salmista and was ruled in advance — no gloss, in either arm — so it is not a logged decision here. M4 (lexical steadiness) bore at versículo ×2 and ara ×2 and was ruled in advance to be held to one English word each; also not logged.