Essay on the Principles of Translation

What it is about

The first systematic treatise on translation in English. Tytler defines a good translation by three general laws (ch. I). It gives a complete transcript of the ideas of the original. Its style and manner are of the same character as the original's. It has all the ease of original composition. The laws are ranked: manner may be given up to keep the sense, but never the sense for the manner, and neither for a fancied elegance (ch. VIII, on D'Ablancourt). Additions are allowed only when they are tied to the thought, and omissions only of accessory, redundant ideas (ch. III). Ch. V shows failures of manner: Castalio's Latin Bible pours nine sentences of Genesis 1 into one period. Ch. XI separates the "general idioms" of a language, such as adjective order, the progressive he is writing or to before the infinitive, which any translator exchanges at once, from the particular idiomatic phrases of conversation. Those need a corresponding idiom (rem acu tetigisti, "you have hit the nail upon the head"), and they must not belong to the wrong age or country (Echard's Terence).

Ideas

• I1. The three laws as a lexicographic order over the residual. When the encoder in tools/translate chooses between candidate chunks, it ranks them by world-level agreement first, the author delta's fit second and the stack's fluency third, and never trades a higher level for a lower one. In normal mode, report the three in that order: world agreement, delta bits, stack bits.
• I2. Tytler's split between general and particular idioms is the split between the standard encoder and the lexicon. General idioms are rules of the target language: order, aux + participle, to + infinitive. They belong in profile.rs and en/src/lin.rs, the same for every author. Particular idioms are multiword concepts in meta.rs. Test it this way: an override in an author delta that is only a general idiom of the language should move into the standard.
• I3. Additions and omissions weighted by role. A dropped meta token costs more when it is a core argument (nsubj, obj, iobj) than when it is an accessory (advmod, amod). Score the reader's "dropped" class by relation in measure.rs, so that dropping an accessory looks different from dropping a sense.
• I4. Castalio's error is a boundary error: the sense is kept, but the sentence and clause boundaries of the original are lost. Count sentence and clause boundaries in the meta-world of each verse and report their agreement between authors. This is a cheap first step towards boundaries as features (open problem 4).
• I5. Idioms of the wrong age. Label each delta override with the period or register of its author, and add a control that applies an archaic author's overrides to modern text. The school-talk delta getting hath or unto would show this mismatch as red.

For Mova

Tytler's ranking is the order the translator's measures should be read in. I2 bears on the standard and delta split in profile.rs (open problem 6, the standard leaning to the Bibles). I3 and I4 bear on the reader residual classes in measure.rs.

Topics: Machine translation

Cite

The paper itself, not this card. BibTeX from the library's record of the work.

@book{tytler1791essay,
  title = {{Essay on the Principles of Translation}},
  author = {Tytler, Alexander Fraser},
  year = {1791},
  url = {https://www.gutenberg.org/ebooks/64890}
}

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/tytler-1791-essay-principles-translation/