This point came up in another discussion, and I thought it might be worthwhile to say something about it in its own topic. The following is the recommendation and guidelines I follow for SC.
In academic practice it is common to indicate supplied terms in translations with [square brackets]. I recommend always avoiding these, as they are unscientific, misleading, and ugly. They detract from the text and convey a misleading sense of false precision.
If square brackets are to be used at all, they should restricted to critical editions of texts, where the editor has added words to someone else’s manuscript. But in such cases it is, of course, better to use a clear semantic markup. On SC we only use square brackets on the specific request of the author (or occasionally in inherited texts that we haven’t processed fully).
The basic problem is that in translation one is constantly adding or omitting words. This is entirely necessary and normal. There is nothing about the process of translation that says it must be done word for word, unless one is specifically preparing a word-for-word glossary, which is not really a translation anyway.
The reason for this should be obvious to anyone who knows more than one language: meaning is not mapped precisely on words, but often emerges at a higher linguistic level.
Consider just one example, Ven Bodhi’s translation of SN 27.3. As always, his translations are taken, not to criticize, but as an example of best practices.
As so often, the Pali text is abbreviated (with no variants that I know of):
Yato kho, bhikkhave, bhikkhuno … pe … abhiññā sacchikaraṇīyesu dhammesū”ti.
Now, as in any translation, you can’t do this word-for-word. No-one would translate kho. And pe and ti are represented, not with words, but with punctuation: ellipsis and quote marks respectively.
But making it as literal as possible, we would have:
“When, mendicants, a mendicant … in regard to things realizable by direct knowledge.”
Note that here, despite attempting literalism, we have added or expressed idiomatically many things.
- The text is capitalized.
- Punctuation is added.
- Yato, which literally means “from when”, is rendered more idiomatically as “when”.
- The indefinite article “a” is added.
- The phrase “in regards to” represents, not any words in the original, but the locative plural ending -esu.
- Similarly, “by” represents the instrumental sense of abhiññā (which here is found in an idiomatic form in the Pali rather than the more standard abhiññāya).
- The two words “direct knowledge” represent a single word in the Pali, a somewhat inadequate rendering for ñāṇa when prefixed by abhi-.
Ven Bodhi has:
When a bhikkhu has abandoned the mental corruptions in these six cases … [his mind] becomes wieldy in regard to those things that are to be realized by direct knowledge.
Obviously the extra text has been added to make the abbreviated portion readable, which is fine. But why is [his mind] singled out as supplied text? It seems completely arbitrary. Why not indicate [has abandoned the mental corruptions in these six cases] and [becomes wieldy]? Maybe there’s some reason for it, but it is not at all apparent.
The reader would naturally assume that all these phrases are actually found in the text, and only the words in square brackets are added. This is obviously incorrect, and in any case, adds nothing to their comprehension of the text. Thus the use of square brackets here, and commonly elsewhere, merely creates false precision.
The reality is that we are adding or removing things in every sentence, and so we should. If you want to compare what is in the translation with the original, you have to do it by consulting the original: there is no other way. That is why on SC, we are developing segmented translations, so anyone can see exactly what the source text says.