Wiktionary, a Wikimedia project, serves as a collection of independent dictionaries with definitions in various languages. Currently the largest one is the English-language Wiktionary, which can still be a useful middleman if you want to translate some concept from one language to another (even if it is not English).
One of the things it stores are proper noun, including those of various geographical objects - mountains, lakes, rivers, plains, bays, seas, cities, villages, forts, etc. Each entry may be categorized in one of multiple languages (since most scripts are used by multiple languages). The English entries can have translation tables, that include translations in different languages.
Would it be possible for a two-way relationship between the projects:
OSM can get data from Wiktionary with the help of the translation tables (from English) and entries in different languages (to English)
Wiktionary can get translations of names that are filled in OSM, but not yet into Wiktionary
This will require quite a lot of automation which can be a hassle, since both projects seem to have strict policies regarding bot accounts (the English Wiktionary has about 10 active bots, not sure about OSM).
Actually, you are right, that will be even better. As an upside, Wiktionary can also tag Wikidata’s Q items at definitions and possibly in the future sync the translations directly from there.
The English Wiktionary has a policy on when to include a toponym that requires attestation in at least three sources. It usually isn’t difficult to find these sources manually, but automating the process would be more difficult unless you focus on specific subsets of geographic features. At that point, you’re probably joining multiple sources besides OSM.
Any geographic feature that satisfies Wiktionary’s criteria for inclusion would likely satisfy Wikidata’s notability policy. So you can create a Wikidata item with the same sources, then link the OSM element to it using wikidata=*. Many OSM data consumers will automatically consult the item’s labels in various languages to use as translated map labels.
The long-term goal is for Wiktionaries to incorporate lexicographical data into their entries. No existing OSM-based data consumer is using lexicographical data yet, but it could be useful for grammatical information that’s difficult to model elsewhere.
In the meantime, the English Wiktionary can link to a Wikidata item or lexeme. Wiktionary could also potentially cite OSM to satisfy any attestation requirements. The two-way process you describe seems less likely, if only because a tag linking a toponym to a Wiktionary entry would be very redundant – inherently the same as the name=* itself.
From my understanding that is supposed to be the case for all definitions, but is more like a goal. A lot of definitions lack any provided source (e.g. Asia Minor), and some have a single one (e.g. Athens).
Also I was not talking about linking Wiktionary entries but rather “scraping” the data in the translation tables periodically.
P.S. Thanks for the link to the Wikidata lexicographical data, I did not know of the project.
Yes, the policy is aspirational in general. However, as we often see in OSM, an automated process for bulk additions will attract more scrutiny. The Wiktionary community may not be prepared for an influx of entries with few or no citations. Similarly, Wikidata’s lexicographers get very annoyed whenever someone imports reams of lexemes without senses. (Fortunately, sense definitions can be generated automatically, as on Wiktionary.)
Mappers already go in the other direction on occasion. People have been manually copying names from Wikipedia, Wikidata, and dictionaries into OSM for decades now. Sometimes the very first thing a Wikipedian wants to do is “synchronize” all the Wikipedia article titles or Wikidata labels of a given country or city with OSM name tags.
Unfortunately, this often leaves formatting mistakes and perpetuates mistakes in underresourced languages that they can’t read and know nothing about. We don’t have good quality controls around these problems because our project is optimized around the idea of contributing what you know yourself rather than hearsay and guesswork. An automated process would unfortunately exacerbate this issue, as there has been a certain amount of guesswork on the wikis too.
Wikidata’s lexicographical data project is really cool, and I hope data consumers can find novel uses for it in conjunction with OSM!
Figuring out general directions or locations of places if you are abroad (whether for work or as a tourist).
Good luck figuring out the spelling of some French city or town just by hearing it (Aix-en-Provence is a great example). Same if it will be written in a script you cannot write in.
On the other hand translations in another language may keep some meaning (in which case it will not help you e.g. Golden Sands) or as in most cases, the “translation” will just be transliteration into another script (bearing in mind that it may depend on language also). This means if you hear some name of a place, by approximating the its sounds in your language and using the appropriate letters, you might be able to search for it and find it.