8. What translators do — sound, rhyme, and the line

The charter put prose first. The project's own drift, from the second half of August, put sound first — rhyme, chime, refrain, meter, in Persian and Arabic and Japanese — because sound is where a published translator's choices are easiest to count from a scan, and counting was what an uncalibrated project could do. The September assessment named the drift and capped it. What remains is the largest single body of evidence in the record, consolidated at the close into six entries whose findings are mostly about how much a formal choice costs and where the cost lands.

Mimetics

Japanese has a closed grammatical class of sound- and manner-words — nosonoso, gayagaya — with no English counterpart. The mimetics entry records the central finding as one about assertion rather than ornament: enacting the sound with an English phonaestheme and stating it with an adverb "are not two wordings of one content: a depiction commits to which sound and which gait, a statement to how fast and how long." The published default is to state; Morri, in 1918, omitted the mimetic outright at four sites of fifteen. Two of the project's own rules were withdrawn by its later tests, and the surviving instruction is procedural: write both renderings at every site, name what each commits to, and "report your own carriage census as claimed, never as measured."

Sound figures in prose

Arabic literary prose rhymes its clause-ends, echoes one root in two shapes, matches its members in pattern. The handbook's oldest instruction here was compensate, and the entry records that it was withdrawn, narrowed, and withdrawn again. What replaced it is a finding about readers: shown an English rendering with no source, outside readers inferred that the original had a figure at a given place from the English device — an "ornamentalist" rendering that put sound where the Arabic had none was read as the author's at nine of thirteen plain places, "telling the reader something about the author the translator made up." Answering a figure in place cost about four per cent in length and added little. The published precedent, Knatchbull's 1819 Kalila and Dimna, answered the source's matched figures at none of eighteen loci, having predicted as much in his preface — "impossible to express the sententious brevity of the Arabic."

Marks the reader will read as the author's

Every language boundary changes the marks on the page: the Russian dialogue dash becomes a quotation mark; a word italicized because it is foreign to English lands in a script that already says "foreign." The emphasis entry found that "attention alone cannot see this, that a count can," and that in two self-audits the translator "was right about what he kept and wrong by the sign about what moved by convention." The reader-side finding is the one to remember: an exclamation mark or italic the translator had added was attributed to the author by every reader every time, and with more confidence than a licensed one — about three-quarters of a point on the entry's six-point confidence scale.

Rhymed prose and rhyme

Persian and Arabic prose can rhyme: paired clauses whose final words chime, at Sa'di's aphorisms and at 137 of the 139 clause-ends of one of al-Hariri's assemblies. The rhymed-prose entry counted what seven published hands did with it between 1767 and 1899: zero strict rhymes in 128 opportunities on Sa'di, one in 187 cells on al-Hariri — "the tradition's practice, not a lapse" — while the same hands rhymed their verse renderings at about 0.9. What they kept, at 0.974, was the doublet: the paired members with the sound gone. Forcing the chime on the project's own hand measured its price: reachable at 0.625 on Sa'di and 0.477 on al-Hariri, never in the ordinary word or its synonyms, at a cost in fidelity at 41 of 65 substitutions and in "lighter, less graceful, more self-displaying" English at every landed chime. The one positive prescription in the family is Preston's, from 1850: lineate the clauses, keep each under about fourteen words, balance them. And the entry preserves the correction that closed the project's last thread: an earlier claim that a rhymed line takes an ordinary sense to its end one-eleventh as often as a free line was recomputed on September 8 against the actual chime slot rather than the printed line's last word, and the factor is about 5.5.

The radif

A Persian rhyming position is the rhyme word followed by the radif, a word or phrase repeated identically at every rhyming position; 305 of Hafez's 495 ghazals carry one. Persian puts its verb last, so any word can end a line; English cannot. The radif entry records that the project first declared several kinds of radif impossible to carry — a transitive verb, a genitive — and was refuted on each by a published hand: Payne in 1901 carried his radifs at 110 of 137 positions, and at 32 of 38 in the very class the project had called unreachable. The one clean refusal in the record is the object marker rā. What carrying costs is inversion, and the entry's honest note is that the price of an inversion "has no admissible number," because the only rating of it is panel-scored and the panel is uncalibrated.

The formal contract

A Hafez line is spoken for up to three times: by the measure, the rhyme, and the radif. The meter-and-line entry frames the translator's choice as a contract — which of meter, rhyme, radif and line-for-hemistich to sign — and finds that "each clause is paid in a different coin." A fixed measure is paid in the author's small content words of the wrong shape: numerals, epithets, the names of things. The rhyme is paid in the line-end slot: a sense that would naturally close the line does so about a quarter of the time when the line is free and about once in twenty when it must rhyme, across 747 sense positions in two published hands, and "the displaced sense lands nowhere plannable." Walter Leaf's 1898 list of the Persian measures English could not carry was audited mechanically: his mechanism held and his list did not, so the entry's first instruction is to take a translator's printed mechanism and re-derive his list.

What these six entries have in common

Every one turned a craft maxim into a count and found the count smaller, or elsewhere, than the maxim said. Compensate became answer in place, and expect it to add little. Carry the sound became the tradition does not, and here is what it costs when you do. The rhyme costs the line-end became the rhyme costs the chime slot, by a factor of about five, not eleven. And each says, in its standing paragraph, the sentence that keeps this essay honest: "no sentence says a reader hears a metre or a rhyme." The readers in these studies were models.