Proposal for a mass edit of street names in Republic of Cyprus

[PeachyOne]

Greetings! Although I endeavoured to create a Cypriot Matrix chat to discuss this stuff and some discussion was had, the contributors were mostly from the north or non-Greek (Cypriot) speakers from the south, so I wanted to bring it up here to perhaps get it to a wider audience. The Cypriot community in OSM is quite small after all.
Full disclosure, I myself am Greek (from Greece).

Basically the vast majority of street names in the Republic of Cyprus are all romanised, or written in English.

Basically, this goes against the on the ground principle in general, and status of names in Cyprus in particular. In general, the name key in the south is reserved for Greek, and in the north for Turkish.

Romanised street names are still useful, and are recorded when the original name is in Cyrillic and Greek script under the int_name tag. Street signs in Cyprus (at least in Nicosia where I have checked) also record this romanised name under the Greek one:

Any name tags that are in English are also in error: this is a common mistake made by English speakers in Greece also, and I presume other places as well. Local sign, local name. Doesn’t matter if its named after John Kennedy, that can go in name:en if you like, but name is still for local language, and int_name is still for romanisation:

Even with Avenue in English, John Kennedy is still written in romanised Greek (Tzon Kennenty) so I think it is undisputed that it should not be written in English. After all, in English he would be called John F Kennedy to begin with Whether these English names are valid or invalid is up for interpretation, but I generally keep them when the street is named after an English speaking person as I have seen actual usage of such names.

Basically, my proposition is a mass edit, where all romanised names would be moved to int_name, and all English names to name:en and equivalent Greek names written from scratch by me and/or other Greek-speaking community members.
I was hoping to recruit Greek Cypriots into this effort mainly to figure out names particular to Cyprus more easily, as reversing ISO 843 is not 1 to 1 and involves some guess work/googling if you are unfamiliar with the name. But names of Greek political/historical figures are quite widespread in the Republic of Cyprus and can be reconstructed with 0 controversy.
I don’t yet have a plan for this laid out in detail, but basically the only automated step would be the mass moving of the name tag into the int_name tag. In order to avoid spillover into the Buffer Zone, where English is standard and to reduce the size of the changeset, I proposed to do this edit in chunks (perhaps region by region).
Obviously no names in the north would be affected, and these have long (rightly) been written in Turkish, as is spoken there.

Really what I am seeking by writing this thread is a reflection of consensus and that I am not out of line. In the Matrix chat, mostly the concerns were around lack of Cypriot Greek speakers, but as the peculiarities of the dialect generally do not manifest in official government use (e.g. street names) I believe this to be a non issue.

[dieterdreist]

From the pictures you present it appears that the street signs are all bilingual, in other areas (admittedly where several languages are commonly spoken by the people), we add both names to „name“ and add the individual names also to language specific name tags, e.g. name:en, etc.

I agree that it seems odd to have Kennedy in English not written as Kennedy.

[Logan_Mc]

As a former corporate editor, I used to contribute edits specifically to this country for a time. From what I remember, my former boss said there was one guy who added the Romanized names en-masse. If I remember correctly, it was a dude from the UK that liked to vacation there (but don’t quote me on that. That may not be correct). I remember myself and another colleague reaching out to discuss adding Greek names as the default name value (with help from the local community wherever possible) where they could be verified (using Mapillary mostly), but the effort fizzled out. I was bummed out, but it was a good decision, because such efforts are best lead by the local community, not a corporate entity many thousands of kilometers away.

There was also a guy who added fictional roads (like fictional trunk roads) all the time, which caused a major headache for obvious reasons, and that was considered more of an issue/priority to remedy/work on than the romanized greek names.

[PeachyOne]

It is not in fact bilingual, but a transliteration of the Greek text under ISO 843. This is the name that goes into the int_name tag. If it were billingual, Vasilissis Amalias would be rendered as Queen Amalia Street, as it refers to Amalia of Oldenburg, 1st Queen consort of Greece.

In an actual bilingual place like Hong Kong, the street/place has an English name and a different, but usually related, Chinese name. Not simply a transliteration of one into the other.

I think we can all agree for example, that the English name of Λεμεσός is Limassol.

Large image

Yet here it clearly says Lemesos. This is transliterated Greek signage, for the convenience of foreigners, not English text. Even if it was, this would suggest similar signage in Bulgaria, North Macedonia and Greece makes these bilingual nations also.

Large images

Who in the world is Viktoros Ougo? Victor Hugo, famed French writer. His name rendered neither in French nor in English, but in transliterated, romanised Greek.

[PeachyOne]

In general this is pretty common. Visiting mappers add names that they can write in their own script rather than leave it blank for local mappers to fill in, or use the name:en tag instead of int_name. It happens a lot in the Greek islands for example. I don’t blame them, but it does cause occasional headaches.

In the Republic of Cyprus the vast majority of streets have been added by these sorts of people. In the North where there are more active mappers who speak Turkish, (and fewer tourists for that matter) this has been amended where the Turkish and Latin script differ. In the South, the most active mappers are not speakers of Greek themselves, so the status quo remains.

[Willem1]

Disclaimer: I’m not from Cyprus

But I remember that when I did a bunch of mapping around Athens a long time ago (2009?) at that point a lot of street names were written in the Latin script there too. In the next few years, this seems to have been cleaned up nicely. It would make sense to me if the same cleanup is done for the Republic of Cyprus.

[rhhs (Richard)]

Note that in Bulgaria, the official rule is that names of streets named after foreigners should use the original foreign spelling. Hence ul. Marie Curie and not ul. Mariya Kyuri. I understand the rules are different in the Republic of Cyprus, so those rules should be followed so that they correspond to “on the ground” truth.

[PeachyOne]

In general though, the primary name tag remains the bulgarian name, no? This is mainly what i am arguing for, replacing the primary tag, that is now romanised Greek, with the Greek name.

P.S. I only happened to use ul. Marie Curie because it was an image I had in my personal archive. The street in question seems to have an int_name transliteration, and name:en. If anything it should have name:fr

[Logan_Mc]

The proposal we submitted was a little more complex than I articulated previously. Essentially, the company I worked for at the time offered to reapply the names as could be confirmed on signage from ground imagery. If there was a greek name, in greek script on the sign, I believe we would have put that in the primary name tag, if there was a transliteration only, that would also have been put as the primary name. If there was both greek script and a transliteration, greek would be the primary name amd the transliteration the internationational name, I think.

[Logan_Mc]

I hope the cypriot community (and I refer specifically to the republic of cyprus, not the northern portion of the island), maybe with some help from the greek osm community, is able to resolve this name situation to their liking.

[rhhs (Richard)]

@PeachyOne

In general though, the primary name tag remains the bulgarian name, no? This is mainly what i am arguing for, replacing the primary tag, that is now romanised Greek, with the Greek name.

Yes. The guidelines are documented here (scroll down further for the Greek guidelines). I suggest that after guidelines for Cyprus are decided on, they are documented on that page too.

@PeachyOne

P.S. I only happened to use ul. Marie Curie because it was an image I had in my personal archive. The street in question seems to have an int_name transliteration, and name:en. If anything it should have name:fr

Thanks for pointing that out: I corrected it now. Adding name:fr is not really necessary as users with French as their interface language will get to see the int_name anyway. Now there’s a risk that Polish users will want to see name:pl=Marie Skłodowska-Curie added…

[PeachyOne]

It’s honestly even worse than I thought, because the romanisation carried out in the past is both internally inconsistent, and inconsistent with on the ground truth/international standards.

Here and here, we have two different romanised spellings for the same name.

The actual street sing says this:

Its hard to read, but this says (Leof) Kon/nou Palaiologou. A slash is a common way to abbreviate the name Konstantinos in writing. Ergo, we have a third, entirely different romanisation that neither non-Greek speaker chose to follow. This would be the standard romanisation under ISO 843 and the one that should be in OSM.

But most importantly, since there is only one standard way used by governments (both national and local) to romanise Standard Greek, there would be zero confusion/room for interpretation if Greek names were the main name tag. As it stands now, there are plenty of examples throughout Cyprus were the data can be interpreted by a machine as two different streets when it is simply inconsistent spelling through haphazard romanisation.

Obviously I am not saying that the hard work of other mappers was not without merit, but if an organised effort is not taken, eventually native Cypriot mappers will begin piecemeal modification of their own, and this will in turn add an entire new layer of inconsistency.

[PeachyOne]

Although it wasn’t automated as it involved lots of Turkish transliterations I wanted to manually review, this changeset in the (formerly) Turkish Cypriot quarter of Moutallos reflects the type of tagging that would result from the mass edit if it were to go through. My kind regards to Afrandez, who made a changeset just before that added the Turkish names, assisting me in reconstructing who was meant from the inconsistent signage/transliteration.

[pnorman (Paul Norman)]

I support a cleanup like this. Romanized Greek is not English. Neither are the versions of Eastern Asia CJK languages that use the latin script. But,

@PeachyOne

Romanised street names are still useful, and are recorded when the original name is in Cyrillic and Greek script under the int_name tag.

I don’t think int_name is the right tag for this. We don’t use that for romanized Kanji or other similar situations. name:el-Latn seems to be the most suitable. If you do this, be sure to set name:el as well so a user who wants Greek can get it in their language choice.

[PeachyOne]

While I don’t disagree with the usage of a tag like name:el-Latn in theory, in practice it has near zero use, while int_name is widely used in countries that use Cyrillic and Greek script to indicate a transliteration, and documented as such on the wiki.

In Cyprus itself (if i’m not mistaken) this was also seen as pretty standard for this use before my proposal, wherever Greek names are employed already (probably in no small part to its usage in Greece however).

Is it necessary after the fact to break it into separate tags for el-Latn, bg-Latn etc.? They all serve the same objective purpose of transliterating the name tag, whatever language that name tag may be in. In any case, that would be a whole separate cleanup of its own.

[alan_gr (Alan)]

The Serbian equivalent name:sr-Latn is used on nearly 250,000 objects. Perhaps that has evolved differently because parallel use of the two alphabets is more common in Serbia than in Bulgaria or Greece.

[Duja]

Exactly, since Serbian Latin is a fully-fledged, standardized, and widely used alphabet (even more than Serbian Cyrillic). We even have a full map rendered in it.

[alan_gr (Alan)]

Interesting, I see there is also a lot of usage of int_name in Serbia. Is that basically used for the name in Latin characters without diacritics / special characters?

(A little off-topic I know, but I think the theme of how int_name is used is relevant. I had only been aware of int_name previously for things like seas that are bordered by multiple languages).

[Duja]

@alan_gr

Interesting, I see there is also a lot of usage of int_name in Serbia. Is that basically used for the name in Latin characters without diacritics / special characters?

It is indeed (except in a few cases where a genuine int_name exists for the feature).

I’m not terribly happy with our approach of manually populating everything (we have sets of four names for pretty much everything: name=, int_name=, name:sr=, name:sr-Latn=), since the basic Cyrillic name can be transliterated 1:1 to other forms. However, they provide a good workaround in cases when data consumers that don’t implement country-specific transliterations, so they could fall back to int_name to search for or render.

[PeachyOne]

In Serbocroatian the Latin script is a tool used by local users since the 19th century. In Greek or Bulgarian, the transliterations on signage are for the convenience of foreigners only. Greek speaking people never use this transliteration to actually write out the language, and often adopt conflicting ones to conform to the norms of local languages when immigrating: see the many forms of Giannis.

@Duja , would you mind explaining in a bit more nuance how int_name differs to name:sr_Ltn in your country’s case?

In any case, if there is issue with the usage of int_name, it should be dealt with separately throughout the project with input from each local community, I cannot claim to speak for the entire Greek community, which is much more active than the Cypriot community at hand here. I assume the Bulgarian and Belarusian communities would have their own input on the matter.

On the topic at hand, i don’t mind using name:el-Latn myself for the Cypriot edit. This would however create a break in policy between the two main Greek speaking nations on the planet that it would be useful to keep in mind.

It does solve a Cypriot issue I encountered, of the implication int_name raises that it is the Official name recognised internationally of any given street. For example in Paphos there is a street named after Kemal Ataturk. It has been tagged with
name/name:el=Κεμάλ Ατατούρκ
name:tr=Kemal Atatürk
name:en=Kemal Ataturk
int_name=Kemal Atatourk

The “international” version of Ataturk’s name is clearly the english one and I added that tag so that it can act as a fallback, but the wording of the int_name tag implies the greek transliteration is international. When adressing a letter from say the UK, i suppose that is indeed the case, as the Turkish names of more obscure Turkish lords may not be legible to a Greek speaking Cypriot postman, but still.