Aller au contenu
login
arrow_backRetour aux issues
santifer/career-ops #3223

Native-speaker check wanted: the human-in-the-loop row in 11 languages (a940028a)

ecoDébutant documentation help wanted good first issue

descriptionDescription

I merged `a940028a` today, which restored the absolute human-in-the-loop wording in eleven translated READMEs. **I don't speak nine of those languages**, so I want the wording checked by people who do. ### What happened Every README's features table carries the project's central promise: the system never submits an application. Eleven translations had weakened it to "never submits **automatically**", which is a different claim — it leaves room for the system to submit some other way. The drift appeared independently in eight languages and always in the same direction, because adding the adverb reads naturally to a translator. Arabic was worse: `النظام لا يقوم بالتقديم التلقائي أبدًا دون إذنك` said "never submits automatically **without your permission**", turning an absolute guarantee into a conditional one. ### Why I could fix it without inventing anything **Each README's own FAQ already stated the guarantee correctly** — only the table row had drifted, which suggests the two surfaces were translated at different times. So in seven languages the fix was deleting one word, and in the three that needed a clause added (`de`, `ru`, `ar`) I copied it verbatim from that same file's FAQ. Every word is its original translator's. ### What I'm asking for Deleting a word can't change meaning, but it **can** leave a sentence that reads stiffly, and copying a clause from a FAQ into a table row can land in the wrong register. That's what I can't check. If you read any of these, please look at the human-in-the-loop row and say whether it sounds natural. A comment is enough; a PR only if you'd rather. | language | file | changed | |---|---|---| | العربية | `README.ar.md` | clause replaced from its own FAQ | | Deutsch | `README.de.md` | clause replaced from its own FAQ | | Русский | `README.ru.md` | clause replaced from its own FAQ | | Français | `README.fr.md` | one word deleted | | 日本語 | `README.ja.md` | one word deleted | | 한국어 | `README.ko-KR.md` | one word deleted | | 简体中文 | `README.cn.md` | one word deleted | | 繁體中文 | `README.zh-TW.md` | one word deleted | | தமிழ் | `README.ta.md` | one word deleted | | Português (BR) | `README.pt-BR.md` | one word deleted | | Español | `README.es.md` | `aplicación` → `candidatura`, accent fixed | **Arabic is the one I'd most like a second pair of eyes on**, because it needed a real replacement rather than a deletion and it's the language where I'd least notice an awkward result. ### The rule this row now carries Each row ends with an invisible marker that states the constraint at the point of editing: ```html ``` The three named hedges are the three that actually occurred. If your language has a natural fourth way to soften it, say so and I'll add it: the list is only as good as the failure modes we've seen. No deadline and nothing is blocked on this. The wording is live and defensible as it stands; I'd just rather it were also natural.
codeOuvre sur GitHub