Repository path: workshop/translations/rayo-de-luna/R04-v1/translation.md · rendered 2026-09-09
Page metadata (front matter)
| type | translation |
|---|---|
| id | T-rayo-de-luna-R04-v1 |
| status | active |
| created | 2026-07-27 |
| updated | 2026-07-27 |
| senses | accuracy, naturalness, style-correspondence, voice, cultural-mediation, consistency, purpose-fit |
| internal-judgment-only | true |
| provisional | true |
| translated-by | lead |
| contamination | suspected |
| links | workshop/regimes/R04-lead-close.md, workshop/translations/rayo-de-luna/source-es-full.txt, wiki/arms/ARM-typology-logs.md, wiki/findings/results/RS-20260727-log-typology.md |
T-rayo-de-luna-R04-v1 — Gustavo Adolfo Bécquer, «El rayo de luna» (opening)
Pair. Spanish → English. The project's first translation on the Spanish pair, and its tenth source language.
Why this exists, and why the ordering matters. This translation is the translation limb of S036's successor session, whose principal unit is ARM-typology-logs — an open coding of every frozen translator's log the project holds. It was drafted, self-revised, logged and frozen before a single one of the prior logs was reopened, so that this log can serve as a held-out saturation test of the decision classes derived from the others. A log written after seeing the coding scheme would be worth nothing as a test of it; the freeze is what makes the test possible. See RS-20260727-log-typology §5.
Correction, same day, to this framing paragraph only — not to the translation or the log, which stand exactly as frozen at b4cf674. The freeze commit said "the eighteen prior logs", taking the figure from NEXT.md. The corpus is twenty: NEXT.md and the arm page both undercounted (RS-20260727-log-typology §0). The number is corrected here rather than left standing, because a false count in a header is not protected by a freeze that exists to protect a translator's decisions.
Source. Gustavo Adolfo Bécquer, «El rayo de luna — Leyenda de Soria» (first published in El Contemporáneo, 12 and 13 February 1862). Public domain (Bécquer d. 1870). Text from Spanish Wikisource, El rayo de luna, retrieved 2026-07-27 via action=raw; cleaned of templates and markup and stored whole at ../source-es-full.txt — 3,552 whitespace-delimited tokens, 71 paragraphs — so a later span is taken from a fixed text.
Span. Paragraphs 0–13 of source-es-full.txt, 689 source words: the essayistic frame and the whole portrait of Manrique, ending on his speech to the stars. The narrative proper begins at paragraph 14 («Sobre el Duero…»), so the span is a self-contained unit and not an arbitrary cut. The remaining 2,863 words are not claimed, scheduled, or reserved: this is a single R04 freeze, not a serial work, and ARM-longwork is the project's only serial commitment.
Declared purpose (purpose-fit). A reading edition in period-idiomatic literary English of Bécquer's own moment — the register of an English writer of the early 1860s: lightly elevated, latinate where the Spanish is latinate, but not archaised beyond its date. No Wardour Street. The declaration is load-bearing here because Spanish→English is a near/near pair on TH-20260724-translation-distance-axes, and the near-pair failure mode the project has already documented is the cognate that is a false friend of register (goodness-senses.md, style-correspondence, S013). A declared period target is what makes those cognates decidable one way or the other instead of case by case.
Contamination: suspected, and the exposure is small, specific, and was written down before it was measured.
One published English translation is freely reachable and was downloaded before translating: Cornelia Frances Bates and Katharine Lee Bates, Romantic Legends of Spain (Thomas Y. Crowell, New York, 1909), Project Gutenberg #50044. Its body text was not read. One exposure occurred and is declared: while identifying which legends the volume contained, I printed its table of contents, which gave me the Bateses' rendering of the title — four words — before I chose my own. Nothing else of the translation was seen; the extraction of the comparator section for measurement was done by script, with only counts printed.
Bécquer is also a canonical Spanish author whose Leyendas are widely translated, so the general prior of remembered English is real and unquantifiable, independent of this session's exposure. That is what suspected records here. Per the standing rule (wiki/method-notes.md (bcd)), the measurement below was run after the freeze commit and is reported whatever it said.
Contamination, measured. Added after the freeze commit b4cf674, which contains the translation and the log below exactly as written before any comparator was extracted. Against Bates & Bates 1909, paragraph-bounded: longest common run 20 tokens, 24 shared 12-grams, against a null-control floor of 5 and 4 tokens on two other legends by the same translators. That is the largest lead-vs-published run this project has measured, and it must be read with three things attached. (i) It decomposes into six regions of 12-20 tokens covering 11.7% of the span, not one. (ii) The 20-token run contains a 7-token block twice, because the Spanish repeats a clause and both translators held the repetition: it is 13 tokens of distinct material. (iii) Eighteen of the nineteen decisions in the log below lie outside the shared regions - the runs sit where the log records nothing, which is where the pair left nothing to decide. Method, the FORCED-vs-remembered argument, and its limits: RS-20260727-log-typology §6. This artifact stays suspected and is admissible as practice, not as an independence measurement.
Cost. $0. Lead translation, no API call, never ledgered (charter §6, A4).
The Moonbeam — a legend of Soria
I do not know whether this is a history that seems a tale or a tale that seems a history; what I can say is that at the bottom of it there is a truth, a very sad truth, of which I shall perhaps be one of the last to make any use, given the turn of my own imagination.
Another man, with this idea, might have made a volume of lachrymose philosophy out of it; I have written this legend, which, for those who see nothing at the bottom of it, may at least serve to pass an hour.
He was of noble birth; he had been born amid the din of arms, and the unaccustomed blare of a war-trumpet would not have made him lift his head for an instant, nor shift his eyes a jot from the dark parchment on which he was reading the latest letter of some troubadour.
Those who wanted to find him were not to look for him in the wide courtyard of his castle, where the grooms were breaking the colts, the pages were teaching the falcons to fly, and the soldiers passed their rest-days sharpening the iron of their maces against a stone.
"Where is Manrique? Where is your lord?" his mother would ask at times.
"We do not know," his servants would answer. "He may be in the cloister of the monastery of La Peña, sitting on the lip of a tomb, listening to see whether he can catch some word of the conversation of the dead; or on the bridge, watching the waves of the river run one after another under its arches; or huddled in a cleft of the rock, busy counting the stars of heaven, following a cloud with his eyes, or watching the will-o'-the-wisps that shoot like falling stars across the surface of the pools. He will be anywhere rather than where everyone else is."
And in fact Manrique loved solitude, and loved it in such a way that there were times when he could have wished he had no shadow, so that his shadow might not follow him everywhere.
He loved solitude because within it, giving his imagination free rein, he forged a fantastic world peopled by strange creations, the children of his delirium and of his poet's reveries — for Manrique was a poet, so much a poet that the forms in which he might have shut his thoughts had never satisfied him, and he had never shut them in by writing them down.
He believed that among the red embers of the hearth there dwelt fire-spirits of a thousand colours, which ran like insects of gold along the burning logs, or danced in a luminous ring of sparks at the crest of the flames; and he would sit for hours together on a stool beside the tall Gothic chimney-piece, motionless, with his eyes fixed on the fire.
He believed that in the depth of the river's waves, among the mosses of the spring and above the vapours of the lake, there lived mysterious women — fairies, sylphs, undines — who breathed out laments and sighs, or sang and laughed in the monotonous murmur of the water: a murmur he listened to in silence, trying to translate it.
In the clouds, in the air, in the depths of the woods, in the fissures of the crags, he imagined that he made out shapes or heard mysterious sounds — shapes of supernatural beings, intelligible words that he could not understand.
To love! He had been born to dream love, not to feel it. He loved every woman for an instant: this one because she was fair-haired, that one because her lips were red, another because she swayed as she walked, like a reed.
Sometimes his delirium went so far that he would spend a whole night looking at the moon, which floated in the sky in a vapour of silver, or at the stars, which trembled far off like the changing lights in precious stones. On those long nights of poetic sleeplessness he would exclaim:
"If it is true, as the prior of La Peña has told me, that those points of light may be worlds; if it is true that there are people living on that globe of mother-of-pearl that rolls above the clouds — what beautiful women the women of those regions of light must be! And I shall not be able to see them, and I shall not be able to love them… What can their beauty be like?… What can their love be like?"
Translator's log
Written at translation time and frozen with the commit that first contains this file. Decisions and alternatives, not quality claims. Any evaluative sentence carries internal-judgment-only (front matter sets it for the page). This log is the held-out unit for ARM-typology-logs and was written before any of the eighteen prior logs was reopened.
Procedure. R04: read the Spanish whole (3,552 words, ~15 minutes of reading), selected the span at the narrative seam, drafted paragraph by paragraph, then made one self-revision pass against the source. Nineteen decisions recorded, numbered in the order they arose. No dictionary of Spanish→English was consulted; Spanish-internal sense-checking only.
D1 — «historia»/«cuento», the opening chiasmus. Spanish opposes historia (a true account) to cuento (an invented one), and the sentence is built as a chiasmus on that opposition. English "story" covers both terms and would collapse the figure into a tautology. Live options: (a) a history that seems a tale / a tale that seems a history; (b) a true story that reads like a fiction / a fiction that reads like a true story; (c) recast as "whether this happened or was invented". Chose (a). It keeps the chiasmus and the word count, at the cost of leaning on a sense of "history" — a true account of particular events — that was ordinary in 1862 English and is now faintly archaic. (b) is clearer and four words longer, and turns a figure into an explanation. (c) abandons the figure.
D2 — «en su fondo», twice. The phrase carries the essay's whole point and recurs in the next paragraph («a los que nada vean en su fondo»). Rendered both times as at the bottom of it, deliberately at the cost of idiom: English would more naturally say underneath the first time and see nothing in it the second, which would break the tie. Held the repetition.
D3 — «filosofía lacrimosa». Lachrymose is the cognate and is exactly the kind of near-pair cognate the declared purpose exists to decide. Spanish lacrimosa is literary but unremarkable; English lachrymose is markedly bookish and slightly comic. Rejected tearful (plain, loses the mock-grandeur that makes the sentence a joke) and weepy (right tone, wrong century). Took lachrymose: the deflationary joke needs the inflated word, and the declared 1860s register carries it. This is a decision made by the purpose declaration rather than by the sentence — I would probably have written tearful without it.
D4 — «Era noble», the paragraph with no antecedent. The subject is not named until three paragraphs later; the Spanish simply starts on him. Preserved the abruptness. But He was noble in English asserts moral nobility — a false friend, and one that would misdescribe Manrique, who is not presented as morally anything. Chose He was of noble birth, which is four words for two and slightly explanatory. He was a nobleman was the alternative and reads as a flatter opening.
D5 — «un punto». Ni apartar sus ojos un punto — an idiom of minimal degree. Took nor shift his eyes a jot. A hair's breadth was longer and more physical; at all was the flattening option. Jot is period-plausible and keeps the measure-word shape.
D6 — «la última carta de un trovador». Última is the standing ambiguity: last (final, dying) or latest (most recent). The portrait is of a young man reading current troubadour verse, not elegising a dead poet, so latest. Also read carta as letter rather than any documentary sense; no other reading survives the sentence.
D7 — Dialogue punctuation. Spanish sets speech with dashes and closes an interpolated attribution with a second dash. Converted throughout to English double quotation marks with the attribution outside. This is a silent normalisation to target convention; the alternative — keeping dashes — would make the page look like a translation, which is a legitimate choice under a different declared purpose and not under this one.
D8 — «vuestro señor». The servants address the mother with archaic plural vos and call Manrique señor. English has no way to carry the pronoun; your lord keeps the feudal relation in the noun and lets the pronoun go. Your master is the natural modern English and drags the scene towards domestic service.
D9 — «prestando oído a ver si sorprende alguna palabra». Sorprender here is to catch unawares — the dead are conversing and he hopes to overhear. Listening to see whether he can catch some word keeps the furtiveness and the tentativeness. Straining to overhear is tighter English and asserts effort the Spanish does not.
D10 — «los fuegos fatuos… como exhalaciones sobre el haz de las lagunas». Three items in one clause. Fuegos fatuos → will-o'-the-wisps (English has the folk item, so no mediation cost; ignis fatuus would have manufactured an opacity the Spanish does not have). Exhalación in Spanish is a meteor, and como una exhalación is also the ordinary idiom for like a flash — both senses are live and I could not keep both; took shoot like falling stars, which keeps the meteor and lets the speed idiom go. Haz is face/surface: took surface, rejecting face of the meres as a register the rest of the paragraph does not support.
D11 — «La Peña», twice. Retained the Spanish, uncalqued, at both occurrences (the monastery, and the prior). Calquing it (the Crag) would translate a place-name into a common noun mid-sentence; the retained form is opaque to an English reader in a way the Spanish is not, and I accepted that. Consistency across the two sites was checked deliberately — it is one of the few places in a 689-word span where a class could have gone two ways.
D12 — «hijas de sus delirios». Creaciones is feminine, so its offspring are hijas. English has no grammatical gender to motivate daughters, and daughters of his delirium would read as a chosen image rather than as agreement. Took children. This is grammar-to-lexis loss of exactly the kind TH-20260724 C1 describes — except that here the loss is total and costless, because nothing in the source is doing work with it.
D13 — «encerrar», twice in one sentence. Las formas en que pudiera encerrar sus pensamientos, y nunca los había encerrado al escribirlos. The repetition is the sentence's hinge. Held it with shut … shut in, accepting the slight awkwardness of the second, rather than varying to confine/enclose.
D14 — «se pasaba las horas muertas». A fixed idiom meaning hours on end. The literal the dead hours is available in English and would be atmospheric in a ghost-adjacent legend — which is precisely why it is wrong: it would import a Gothic colour the Spanish idiom does not carry, and would make Bécquer look more lurid than he is. Took for hours together. Recording the temptation because the false-friend mechanism here is idiom-level rather than word-level, and I have not seen that named on the near-pair side before.
D15 — «rumor… rumor que». Anadiplosis across a colon. Held it: murmur of the water: a murmur he listened to. The alternative was a relative clause, which loses the pivot.
D16 — «palabras inteligibles que no podía comprender». Read as printed in the Wikisource text. It is a paradox — intelligible words he could not understand — and some editions print ininteligibles, which dissolves it. I did not go looking for a variant edition to settle it, and I did not silently repair it: the paradox is entirely in Bécquer's manner and the emendation would be mine. Flagged rather than fixed. This is the one decision in the span that a later session could show to be wrong on textual grounds.
D17 — «¡Amar!». A bare infinitive as an exclamatory heading. Love! reads in English as an interjection or an imperative; To love! keeps the infinitive and sounds slightly stilted. Took To love! — the stiltedness is closer to the Spanish gesture than the misreading is.
D18 — «¡qué mujeres tan hermosas serán las mujeres de esas regiones luminosas!». The clause repeats mujeres and rhymes hermosas / luminosas. English can keep one or the other, not both. Kept the repetition (what beautiful women the women of those regions of light must be), which is clumsy, and let the rhyme go entirely. The alternative — smoothing to how beautiful the women of those luminous regions must be — is better English and deletes the only sound-figure in Manrique's speech. Note the asymmetry: nothing in this span made me trade the other way, and I did not have a rule for choosing; I chose per site.
D19 — Register drift, caught on the self-revision pass, not on the draft. The draft had accumulated whiled away, crevices, golden insects, luminous round — a mixture of Victorian-idiomatic and plain-modern that had no principle behind it. On revision I pulled the whole span towards the declared 1860s register and made three specific changes: whiled away → passed, crevices of the crags → fissures of the crags, golden insects → insects of gold (restoring the Spanish's postposed genitive, which English of that date tolerates). What I want on the record is the shape of the error, not the fixes: the register was drifting site by site while I attended to lexical problems, and the purpose declaration only became operative when I read the span as a span. The declaration was written before the draft and did not prevent the drift.
Not decided, and left visible. The title. El rayo de luna is the moonbeam, and I had already seen the Bates title in a table of contents before choosing (declared above). I took The Moonbeam because it is what the Spanish says, and I cannot certify that four words of prior sight played no part.