Libya: Tifinagh added to the primary `name` tag on ~1,250 places since July, mostly as transliterations of the Arabic name

Communities Africa
1

Hello,

I help build شوارع طرابلس / Tripoli Streets, a free civic reporting app used in Libya. It renders standard OSM raster tiles, so it shows the name tag as-is.

Over the past week four separate users have written to us about place names on the map, and one of them rated the app 2/5 over it — our only 2. In his words: «أسماء بعض المناطق مكتوبة بحروف غريبة وكأنها أمازيغية» — “some area names are written in strange letters, as if they were Amazigh”. A fifth asked us to make the map “like Google Maps”, which on inspection was partly the same complaint.

I want to be clear where I’m coming from before the data, because this is a sensitive subject and I don’t want to be misread.

This is not an objection to Amazigh names in OpenStreetMap. Tamazight has official status in Libya’s constitutional declaration. Amazigh names for Zuwara, Nalut, Yefren, Jadu, Kabaw, Ghadames, Ghat and Ubari belong on the map, and I’d like to see them rendered there. My question is narrower: which tag they belong in, and whether these particular values are names anyone actually uses.

What we found

Queried via Overpass on 2026-08-14:

  • name:ber is present on 1,557 of 1,838 Libyan places (85%). That includes 162 places inside Tripoli and 90 around Benghazi — cities where Amazigh is not the local language.
  • The distribution has no signal: 90% inside the Amazigh-speaking belt vs 82% outside it. If the tag marked genuine local usage you would expect the opposite.
  • 76.4% of name:ber values are a letter-for-letter Tifinagh transliteration of the Arabic name (87.5% at a ≥70% similarity threshold), not Amazigh endonyms.
  • Tifinagh has also been added to the primary name tag as a multi-script concatenation, e.g. ⵜⴰⵎⵓⵔⵜ ⵏ ⵡⴰⵜ ⵡⵉⵍⵍⵓⵍ زوارة Zuwara, with the Tifinagh first. 401 of 471 place labels in western Libya (85%) now carry Tifinagh inside name. This is what actually reaches renderers and end users.
  • Provenance: of 24 randomly sampled name:ber tags, 24 were first added by the same account, which made roughly 1,255 edits between 2026-07-07 and 2026-08-10. Janzur (جنزور) and the Tripoli neighbourhood حي السلام were tagged on 2026-07-07 — which lines up with when our users started writing in.
  • For contrast, the one tag that does look like genuine local mapping — name:jbn (Nafusi) — exists on 13 features, added in 2020, clustered near Yefren and Zuwara. Small, specific, plausible.

I’m deliberately not naming the account in the title, and I’d rather this be about the data than the mapper. I’m happy to share the full query output, sample sets and methodology with anyone who wants to check my numbers — I would genuinely rather be corrected than be right.

Why I think it’s worth a look

  1. name is meant to be the name in local use. A transliteration of the Arabic name into Tifinagh is not a name in local use; it’s a rendering of one script in another.
  2. Coverage doesn’t match the language. 162 Tripoli places is the clearest example — that isn’t a disputed boundary case, it’s a city where nobody uses these names.
  3. It’s already contested. Several other mappers have reverted parts of it. There’s no forum thread I could find and no DWG ruling, so it’s currently being resolved one revert at a time.
  4. Downstream, it’s costing real users. Every renderer that shows name shows Tifinagh first to Arabic-speaking Libyans. We’re one app; the ones who don’t write in just stop using it.

What I’m asking for

Not a mass revert by me. I don’t map Libya at the object level and I’m not going to take unilateral action on 1,255 objects in a live dispute. I’d like the community’s view on:

  • whether these should be split — name carrying the Arabic, name:ber carrying the Amazigh — which would satisfy both readerships and is what multilingual name tagging is for;
  • whether transliterations belong in name:ber at all, as opposed to genuine endonyms;
  • whether this warrants DWG involvement, given the scale and the existing reverts.

If the consensus is that I’m wrong about any of this, I’ll say so publicly and to our users.

Thank you.

2

(answering just this part of the question)

It is of course your choice to do that, but if you want control over which language(s) to show that probably isn’t the best option. Given that I suspect you want a just small area, generating and hosting a vector or even a raster map should be pretty trivial, and might even fit on your existing hardware.

For other services, have a look here.

3

Thank you — that’s useful, and I’d been assuming self-hosting was heavier than it is. We’re already costing a Protomaps extract; “might fit on your existing hardware” is encouraging and I’ll test it.

To be clear though, I raised this as a data question rather than a rendering one. We can fix our own map either way — but every renderer showing name will show Tifinagh first to Arabic-speaking Libyans, so switching our tiles solves it for our users and nobody else’s.

So I’d still value a view on the tagging: should name carry the Arabic with name:ber carrying the Amazigh, and do letter-for-letter transliterations belong in name:ber at all? I’m happy to be told I’ve misread the convention.

1 Like