A Chinese monk who knew a little Japanese writes a letter of introduction for a visiting teacher. He needs to put the man's surname — Otomo — into Chinese characters. There are characters that sound like O-To-Mo, so he uses those. But Chinese characters carry meaning as well as sound, and the ones he picked happen to mean "wild, fierce, liberated." The teacher walks into temple after temple bearing a name that reads, on the page, like a description of a brawler. The real name, "Otomo," was itself a Japanese rendering of a Sanskrit Buddhist term — Kalyana Mitra, "spiritual good friend." So a gentle phrase about friendship, passed through two scripts by a translator working from sound, came out the far end meaning the opposite. Dukes uses this small disaster — it happened to his own teacher — to make a large point about how easily Buddhist texts get mistranslated, and why the errors are so hard to catch.1
This page collects Dukes's working theory of why translating Buddhist Chuan Fa and religious texts is a minefield, drawn from two of his footnotes.2 His core diagnosis is the phonetic-versus-literal trap. "Some Chinese sutras," he writes, "are translated from the original Sanskrit phonetically" — the characters were chosen to spell the sound of a Sanskrit word, not to render its meaning. A later reader who tries to translate such a term literally, taking the characters at their face meaning, "results in something far from its meaning."3 The ability to read Chinese is not enough; you need Sanskrit and Buddhist philosophical vocabulary to know when a string of characters is a sound-spelling in disguise. "Even native Chinese," Dukes notes, "do not always realize this."4
He gives three illustrations, each a different failure mode. The Otomo / Kalyana Mitra case is the phonetic-meaning collision — sound-chosen characters that also carry a misleading meaning.5 The Vasubandhu case is the identity error — Stefan Anacker's observation that "various scholars have attempted to make one Vasubandhu into two different persons and several Vasumitras into one," names splitting and merging in opposite directions.6 And the untraceable-title case — a deity-title Dukes spent "extensive researches" trying to identify — is his reconstruction method, where a name resists translation entirely and has to be rebuilt hypothetically from Sanskrit grammar.7
Picture two ways to write a foreign name in a script that is not yours. You can pick characters for what they mean and translate the sense — that gives you a meaningful but possibly clumsy rendering. Or you can pick characters for how they sound and spell the name phonetically — that preserves the pronunciation but the characters' meanings become noise, an accidental side-effect of the spelling. Chinese does both, and crucially, the two methods look identical on the page. There is no marker that says "these characters are a sound-spelling, ignore what they mean."
That is the whole trap. A reader who does not know a given term is a phonetic loan will read the characters for meaning, because that is what characters normally do, and produce a confident, fluent, completely wrong translation. The Otomo case shows the mechanism with cruel clarity: the monk chose the characters for sound (O-To-Mo), but they also meant "wild fierce liberated," so anyone reading them for meaning got a coherent-sounding falsehood. Coherence is the danger. A garbled translation that reads as nonsense gets caught; a mistranslation that reads as a vivid, plausible phrase — "the wild fierce liberated one" — sails straight into the record and stays there. Dukes's teacher "bore the name Fa Tao Meng equally with the correct translation of his name" for the rest of his life. The error did not announce itself. It just rode alongside the truth, indistinguishable to anyone who could not check the Sanskrit.
This page is a quality-control gate for the whole vault's handling of transmitted Buddhist material. Bodhidharma: The Revisionist Biography already shows the name "Bodhidharma" wobbling across scripts — Pu Tai Ta Mo, Daruma, Bodhitara, Tarabodhi; this page supplies the mechanism behind that wobble and the warning that some of those variants may be phonetic spellings whose apparent meanings are accidents. Manas Self-Looping and Pudgalabhava rests on Anacker's Vasubandhu scholarship — and this page flags exactly the identity-confusion (one Vasubandhu split into two) that Anacker himself had to clean up, so the vault knows the textual ground under its Yogacara pages was contested. The gift is a habit: never trust a translated Buddhist term, name, or title without asking whether it was rendered for sound or for sense, and whether the "person" behind it is one figure or several merged.
Dukes's richest illustration is a name he could not translate, and watching him fail productively is more instructive than watching him succeed. He had a deity-title — rendered Ka-chu-ri-ka-shi-ma — and "made extensive researches into this title in a search for possible alternative translations," only to find "it does not appear to be a Sanskrit or Tibetan title of any traceable Buddhist sutra."8 A dead end. So he switches methods. If it is not a known title, he reasons, treat it as "a pure phonetic" — a sound-spelling — and ask what Sanskrit words those sounds could represent.9
Now watch the reconstruction. He breaks the string into pieces and tries each against Sanskrit. Ka-chu-ri could be "dakini," a female celestial being. Ka-shi-ma could be Catura — one of the twelve spirits of the Buddha of Medicine — or kasina, a colored meditation device, or Garjita (thunderbolt), or Simha (lion).10 Then he tests whether the two halves might be one term: "Ka-chu-ri-ka-shi-ma- Acaryasimha (Acarya's Lion)."11 And he checks his reconstruction against grammar: in Sanskrit the endings "i" and "a" often mark a male-female pair, "so in this they appear to accord with Sanskrit grammar."12 This is the phonetic trap run in reverse, deliberately. Instead of being fooled by sound-spelled characters, Dukes assumes the title is sound-spelled and rebuilds the lost Sanskrit from the phonemes up, constrained by what the grammar will allow. It is honest method: he never claims to have found the answer, only to have bounded the possibilities — places, activities, birds, or "Acarya's Lion." The case study models the discipline the rest of the page only warns about.
You are checking a martial-arts history that confidently translates an old Chinese teacher's name as "the Wild Fierce One" and builds a whole romantic paragraph on it. Dukes's footnote is in your head, so the fluent, vivid meaning is exactly what makes you suspicious — coherent is where the trap hides. You stop and ask the one question that springs it: was this name written for sound or for sense? You look for a Sanskrit or Japanese original behind the Chinese. You find one: the characters spell a sound, and that sound transliterates a perfectly ordinary Buddhist term about friendship. The "Wild Fierce One" evaporates. You write in the margin: phonetic loan — meaning is an artifact of the spelling, not the man. Then a second name resists you entirely; no source matches it. You do not force a literal reading. You break it into syllables, lay each against a Sanskrit dictionary, and test whether the pieces make a grammatical pair — male "i", female "a" — exactly the way Dukes rebuilt his untraceable title. You end with three candidate meanings and no certainty, and you write that down too, because a bounded uncertainty is worth more than a confident error.
The evidence is two of Dukes's footnotes: Note 27 (the phonetic-vs-literal principle, the Otomo / Kalyana Mitra case, and the Anacker Vasubandhu observation) and Note 76 (the untraceable-title reconstruction).13 The translation principle itself — that phonetic renderings cannot be read literally — is uncontroversial philology, well-supported and not Dukes-specific. The Otomo anecdote is a practitioner's personal account about his own teacher and is unverifiable from outside, though it illustrates a real and common mechanism. [PLAUSIBLE — needs corroboration] on the specific Otomo / Kalyana Mitra etymology, which rests entirely on Dukes's testimony.
Tension, preserved: the same footnote that teaches careful skepticism also contains Dukes's counter-attack on other people's skepticism — his complaint that some writers dismiss Chinese manuscripts as "forgeries" merely because "one either has never actually seen a copy" or does not understand them.14 He is doing two things at once: warning that translation is hard and warning against using that difficulty to wave away inconvenient texts (notably the Bodhidharma / Wu Ti meeting, which he insists is not apocryphal). The page holds both without resolving the obvious risk — that "you just don't understand the manuscript" can shield a genuinely doubtful text as easily as it can defend a genuine one. This connects to the load-bearing Kongoryuji warrant: many of Dukes's decisive manuscripts are privately held, so his anti-"forgery" argument is also a defense of sources outsiders cannot inspect.
Open: is "Ka-chu-ri-ka-shi-ma" actually "Acarya's Lion," or one of the other candidates Dukes bounds (places, activities, birds, dakini)? He never decides, and without the original the title may be permanently unrecoverable.
Reading Dukes here, you feel two motives pulling the same footnote in opposite directions, and the page is sharper for not smoothing them. As a philologist he is rigorous and convincing: the phonetic trap is real, the Otomo case illustrates it cleanly, the title reconstruction is honest about its own uncertainty, and his use of Anacker shows he reads the critical scholarship. But as a lineage-holder he has a second agenda riding inside the first. The same expertise that lets him catch mistranslations also lets him argue that outsiders keep getting his tradition wrong — and from there it is a short step to "and that is why you should trust the insider's reading and not dismiss our privately-held manuscripts as forgeries." Mainstream textual scholars would converge with him entirely on the translation mechanics; Anacker, whom he cites approvingly, is mainstream scholarship doing exactly this work in the open. They would part company at the place where Dukes turns a real methodological caution ("you may not understand this manuscript") into a shield for sources that cannot be independently examined. The translation skill is genuine and checkable. The use he sometimes puts it to — protecting the unexaminable — is the practitioner-scholar's recurring move, and worth flagging every time it appears.
Bodhidharma: The Revisionist Biography — the biography watches a name wobble across four scripts; this page explains why names wobble and warns that the wobble is not neutral. On the biography page, "Bodhidharma" becomes Pu Tai Ta Mo, Daruma, Bodhitara, Tarabodhi — and the easy assumption is that these are mere spelling variants. This page forbids that ease: some may be phonetic loans whose apparent meanings are accidents of the characters chosen, and a literal reading of any of them could manufacture false biography. The insight neither page produces alone: source-criticism of a transmitted figure has to run two filters, not one. The biography page filters for legend-inflation (did the story grow?); this page adds a filter for translation-artifact (is this "fact" really a mistranslated sound-spelling?). A detail can survive the first filter and die in the second — a name that was never inflated can still be wrong because someone read its sound-characters for sense. Put together, they make the vault's history domain check every transmitted Buddhist datum against both failure modes before trusting it.
Manas Self-Looping and Pudgalabhava — that page's entire Yogacara argument leans on Anacker's translations of Vasubandhu, and this page surfaces the textual instability underneath that warrant. Dukes cites Anacker's own observation that scholars "made one Vasubandhu into two" and "several Vasumitras into one" — meaning the very author whose psychology the Manas page reconstructs was, for a time, a contested and possibly composite figure in the scholarship. The structural parallel is almost too neat to ignore: the Manas page is about a process that fabricates a false sense of a single persistent self out of a stream of moments; the translation history shows scholars fabricating a false sense of single (or doubled) authorial identity out of a stream of texts. The tension this exposes: the vault trusts "Vasubandhu said X" as if Vasubandhu were a stable unit, when the historical record had to work to establish that unit and partly failed (the Yogacara material itself notes his ideas were "considerably revised" between brothers). The joint insight — authorial identity in transmitted Buddhist texts is itself a reconstruction, not a given, and any page that says "Vasubandhu held" is standing on philological labor that could, like Otomo's name, have come out otherwise.
Sharpest implication: the most dangerous mistranslation is not the garbled one but the fluent one — a sound-spelling read for meaning produces a coherent, quotable, persistent falsehood that no amount of re-reading the Chinese will catch.
Generative questions: