Maybe someone can help me with this question: I stumbled about several places, in a major city, where the
name:xx=
tag was filled with a literal translation of the subject’s name.
My understanding would by that the localized named tags would contain the name of the subject that is commonly used in that language, but not a direct translation.
However, there are apparently users that have been doing these “translations” for years and many stuck for a long time. I also didn’t find any clear guideline on how the
name:xx=
should be used; the wiki is very vague about the subject.
Is there any consensus about this?
Example of what I mean:
name=Кра́сная пло́щадь
name:en=Red Square
=> This makes a lot of sense, because the place in Moscow is normally referred to by English speakers as the “Red Square”, and not as “Krásnaya plóshchaď”; just as “Moscow” isn’t referred to as “Moskva”.
However:
name=Place de la Concorde
name:en=Square of Unity (???)
=> This doesn’t make sense. While “Place de la Concorde” can be translated as “Square of Unity”, an English speaker wouldn’t use that term to refer to the specific place in Paris, and others wouldn’t understand it.
This is an an actual problem for clients/apps/renderers which use the “local” locale by default, as it produces a map with meaningless names.
I am coming to the conclusion that with OSM names, and maybe all tagging in OSM, the more you look into it the more of a can of worms it is. My first attempt at handling names with a nod to internationalization came up with a different set of issues, see https://retiredtechie.fitchfamily.org/2019/12/27/names-are-hard-in-openstreetmap/
Getting back to your specific observation, I agree that a literal translation of the name is not a good idea and that it just adds noise to the database that the renderers will have to figure out how to filter out. If I knew the name:xx value is a literal translation and also know that the xx language doesn’t have a common name for the feature then I’d remove that tag.
p.s. Out of curiosity I plugged Кра́сная пло́щадь into https://www.deepl.com/translator and see that it translates to “Beautiful Square”. I had not known that before.
I agree with n67 that it is a can of worms. Big, three-headed worms carrying guns and knives and having a bad disposition.
For me as someone who does not understand anything but the latin script, I want all non-latin names transcribed when I view the map. But when I’m in the place, it would be a good idea to have it in a way the locals will understand. I’m sure I would be understood better to ask for directions to Krásnaya plóshchaď than to the Red square.
But transcribed using what set of phonetics? I’d prefer the phonetic/alphabet mappings used in English but others might prefer Italian, Spanish, French, German, etc. That in itself is a strong reason to simply have the name field use the local language and leave the name:xx tag blank.
Then a renderer could use an appropriate library to create a phonetic equivalent in the language the renderer is targeting. The Germans have a Postgresql plug-in that purports to do that. See https://github.com/giggls/mapnik-german-l10n
That tag confirms the common name used in that language is the same as the “local” name. The consumer should decide without assumption on what kind of the name it is, and supply the phonetic name separately. On the OSM side, separate tag for transliteration (or transcription?) and “translation” of the literal meaning (instead of localization) would be needed.
I don’t understand why you mentioned phonetics in other language. This is an issue for the “local” romanization