Wikipedia talk:Naming conventions (Korean)

People name section

Moved from Wikipedia talk:Manual of Style/Korea (2024 Rewrite & Proposal)

 – seefooddiet (talk) 08:27, 3 October 2024 (UTC)

My proposed version:

1. WP:COMMONNAME
2. Personal preference
3. Split based on pre-1945+NK, post-1945 SK, and diaspora.
   • If pre-1945 or North Korea, use MR with no hyphenation or spaces between syllables in given name, assimilate spelling of personal name (한복남 -> Han Pongnam, not Han Poknam), do not assimilate between surname and given name (백락준 -> Paek Nakchun, not Paeng Nakchun), and do not convert surname to modern common modified transliteration. Recommend (but not mandate) that 이 -> "Yi" and not "I" for surnames.
   • If SK, use RR. Hyphenate given name, do not assimilate spelling of given name (e.g. 김복남 -> "Kim Bok-nam", not "Kim Bong-nam"), and also convert surname to South Korean common spelling (currently given in the table; I may prune the table to only include the names with unambiguous common spellings).
   • For diaspora, determine which language name is most appropriate (Russian, English, Chinese, Japanese, Korean, etc) based on primary nationality/where most notable. If non-Korean language name is most appropriate, romanize per those language guidelines. If their notability is strongly tied to Korea, determine which of the above two options they are most tied to, and follow the option's guidance.

For explanations, see this WIP essay. seefooddiet (talk) 07:52, 24 August 2024 (UTC)

As an update, this section is pending a decision on NK romanization. seefooddiet (talk) 06:53, 5 September 2024 (UTC)

@Seefooddiet, for MR romanization, I think there might be a good case for hyphenation for personal names. The no hyphenation rule for MR seems to come from the 1961 guide, however, there are more modern revisions of McCune–Reischauer (2009 Library of Congress version) that do use hyphenation. From what I've seem most Western Korea Studies programs and academic libraries also use the ALA/LC revision of MR. Examples: [1][2][3][4][5]. I would also point out that romanization of North Korean names tend to either have a hyphen or a space, having neither is pretty rare. For example, most media romanized 장성택 as either Jang Song-thaek or Jang Song Thaek, but not Jang Songthaek. ⁂CountHacker (talk) 09:38, 1 October 2024 (UTC)

We may need to more thoroughly research what version of MR is most commonly applied in practice. Hard to do, given that I've spotted papers with romanization mistakes in them and people almost never specify what version of MR they follow. Anecdotally I think the books and papers I've read that used MR didn't tend to use hyphens in names.

For NK names, while that is true, my main concern was the consistent application of some MR version. If we decide that 1961 is most common, I would be skeptical of (but would not completely rule out) ad-hoc modifications to 1961 to resemble more common NK practices.

You're welcome to research the topic; I'll try to work on it too. seefooddiet (talk) 09:55, 1 October 2024 (UTC)

The ALA-LC (Library of Congress) system is just yet another separate romanization system. It is not appropriate to treat that as MR. (In fact, the ALA-LC system does things that the original MR explicitly prohibits/discourages.)

Anyone can come up with a new romanization system by modifying an existing system, but that should not be regarded as a newer version of that existing system.

For North Korean names, following North Korea's official romanization system (NKR) might be an option, but this idea is already discarded. 172.56.232.137 (talk) 05:25, 2 October 2024 (UTC)

Whether it's a version of MR or not has little impact on what we do. We just need something to use. If that version/system ends up being the most commonly used, we should consider following it. Either way, the Library of Congress itself considers it a version of MR ("The Library of Congress will continue to follow the McCune-Reischauer system to romanize Korean with the exceptions noted in this document."), and other sources seem to call it a version. seefooddiet (talk) 05:47, 2 October 2024 (UTC)

I think it would be best to use the most recent revised version of MR. The idea that the ALA-LC revision is a different romanization system is simply untrue. Most people acknowledge that it is just a revision of MR, for example, Chris Doll, a supporter of Revised Romanization, states that "the Library of Congress (LC) promotes the phonetically based McCune-Reischauer (MR) to Romanize Korean words".[6]. This academic paper here [7] calls it the "ALA-LC rules of McCune-Reischauer". Most Western academic libraries utilize the 2009 ALA-LC revision of MR. I don't see the point of using an older version of MR from the 60s over a more modern one from 2009. The 2009 version also has additional rules that would have been irrelevant in the 30s or 60s such as rules for romanizing foreign loan-words starting with ㄹ.

No hyphen or space in the given names is also not consistent with how North Korean names are presented. Most literature will either use a hyphen or a space, such as the AP stylebook. I do admit on the historical front, for pre-1945 figures, there is a decent mixture of hyphens vs no hyphens. I would also note that officially Revised Romanization recommends using no hyphens over hyphens, yet we've decided to use hyphens for RR. Hyphens are a quick way to show a reader which part of a Korean name is the given name, and that's why it was recommended in the original WP:NCKO over spaces or no hyphens. ⁂CountHacker (talk) 11:01, 3 October 2024 (UTC)

I'm open to the possibility of ALA-LC but at present I'm skeptical of adopting it.

Most important is determining what is most commonly done. Libraries are a pretty niche field. You've conceded that history writings have mixed practice; that's arguably the area we should be giving the most attention to. Again, I think this situation needs more thorough research.

This is my own analysis, but I'm loathe to adopt the ALA-LC system because of how complicated it is. My primary interest is making Wikipedia usable for the average person. I think it's safe to say that there's near 0 regular Wikipedia editors that know how ALA-LC works in detail. On the other hand, the 60s version of MR probably has been the most familiar version.

I wouldn't say the 60s version is outdated; it's perfectly serviceable. Hepburn romanization and pinyin are also old but still work great. 60s MR is the same way; I can't think of any cases, especially on Wikipedia, where there are such significant problems with MR that the ALA-LC version feels needed to me. Certain small features may feel better, but does that merit switching to a more complicated and possibly more niche system? I'm not sure, but I'm skeptical.

We may not even need to adopt the entirety of ALA-LC; if it's hyphenation in names that you want, similar to how we modify RR and ask for hyphens in names, we may be able to just modify 60s MR and add it. You'd have to prove hyphens in names for MR is more common though.

Ultimately, this all boils down to needing to do more research. seefooddiet (talk) 22:26, 3 October 2024 (UTC)

You mention pinyin but the current version of pinyin is actually younger than the 2009 MR, having had its last revision in 2012. However, I would be willing to compromise and go for 60s MR but with hyphens. Having the hyphen modification would make it consistent with how we treat RR names. Didn't know about the automatic romanization code, 60s MR would be a lot easier to code compared to the 2009 MR with its additional rules. I would probably say in 99% of cases, 1961 and 2009 MR would be the same, so going with 1961 can be okay. ⁂CountHacker (talk) 00:42, 4 October 2024 (UTC)

What will matter is research about what is commonly done; that is what grounded the RR hyphenation decision, and it should be what grounds our decision with hyphenating MR. seefooddiet (talk) 01:09, 4 October 2024 (UTC)

Also, the IP user and I have been working on an automatic romanization module for the 60s version of MR. The logic is in place, but needs to be integrated into Wikipedia.

The 60s version of MR is close to deterministic and much simpler to code than the ALA-LC version. The ALA-LC version has an issue where hyphenation of names changes depending on whether the name is "Sino-Korean" or not; determining what names are Sino-Korean is complicated and subjective, making the module even harder to code. We could implement ALA-LC by making compromises or assumptions on issues like these, but it'd still be hard to code. Neither of us are really willing to dive into that coding project.

In short, the 60s version is just much easier to work with, both for regular Wikipedia editors and for our upcoming module, and I don't feel a strong need to adopt ALA-LC. seefooddiet (talk) 22:32, 3 October 2024 (UTC)

Ok, now that we've decided not to use NKR, this proposal is ready.

The broad strokes of it are very similar to what is currently done; I'm hoping this won't be surprising. seefooddiet (talk) 06:43, 8 September 2024 (UTC)

@Dantus21 @Freedom4U @CountHacker @00101984hjw Sorry for tags; looking for feedback on the proposal so we can keep this moving. Nearing the finish line. seefooddiet (talk) 20:30, 18 September 2024 (UTC)

Overall the proposal looks good. Although I did fix up the common surname spelling to be SK only, I'm a little skeptical about it now since I'm not sure how often it is used in reliable sources. If other users like it though I'm okay with it.

Another note: do Wikipedia essays typically use first person plural? I noticed a lot of use of "we" and it seemed a little jarring to me; granted if it is used in other essays I can let it be. Dantus21 (talk) 14:57, 19 September 2024 (UTC)

I'm going to be almost entirely rewriting that section btw. The common surnames table needs to be trimmed to just names for which there's overwhelming consensus on the common spelling. These consensuses are shared in nearly all RS and even in passports; some evidence can be found here: Korean name#Romanization and pronunciation.

E.g. "Kim" easily should almost always be romanized that way. On the other hand, more ambiguous cases like 정/Jung/Jeong/Chung shouldn't.

I'll look into revising the use of "we"; was just a passive decision that I'm not attached to. Is the skepticism on sounding like it's speaking for the community? seefooddiet (talk) 19:47, 19 September 2024 (UTC)

By the way, this is what I wrote about that surname list:

this surname list may not be sufficient. What about surnames like 문 and 신, which are commonly written as "Moon" and "Shin" (instead of "Mun" and "Sin") in English-language text? 172.56.232.246 (talk) 00:54, 20 September 2024 (UTC)

I think we should be conservative about which names to add to the surname list. Before a spelling is included, evidence should be provided of a widespread acceptance of that spelling.

Examples (made-up numbers), if you can prove that 95% of people spell their surname "Kim", then we recommend that spelling. However, if the spelling is 60% "Kim" and 40% "Gim", we shouldn't recommend any spelling; too divided.

So far, I only have evidence for Kim, Lee, Park, and Choi, so that's all I'll include in the table for now. Do you have any evidence for "Moon"? seefooddiet (talk) 01:13, 20 September 2024 (UTC)

Should we use this study as a reference? (the stats are at the end) It’s from 2007 but I don’t know how much it would’ve changed since then. The data from that has "Moon" at 73.5% Dantus21 (talk) 01:20, 20 September 2024 (UTC)

More recent studies are preferred; the ratios do indeed change. seefooddiet (talk) 02:00, 20 September 2024 (UTC)

I think I wrote that after seeing the 2011 South Korean passport statistics (see page 172 (207th page in PDF) of this document): MOON(14815) 70.28%, MUN(6158) 29.21%, ... 172.56.232.246 (talk) 01:29, 20 September 2024 (UTC)

Arhg complicated... Is 70% enough? I'm not sure, but I think it is.

• If we accept 70% as enough, we can expect to be correct 70% of the time and incorrect 30%.
• If we don't accept, we get 70% incorrect. That's objectively worse.

A counterargument to the above is that defaulting to consistent romanization systems when there's uncertainty yields more recognizability. But if we want consistency, shouldn't we use pure RR, with no hyphens and surname modifications?

But if we went pure RR, I think "Bak" and "Gim" would be more confusing and obscure to the average person than "Park" and "Kim". It'd also be clearly more wrong: for "Bak" we'd be getting 99% of cases wrong for a small gain in recognizability for the few who actually know RR.

Summary: I think 70% is enough, and that we should keep modifying RR names using the hyphen and surname conversion. It feels the least confusing to the most amount of people. I don't know about 60% though. seefooddiet (talk) 02:28, 20 September 2024 (UTC)

I’m a little skeptical of 70% being the bar. While it is true that it would have less damage than 30%, by that logic anything that has above 50% (like Jung) would be the ideal choice, which we’ve all (or at least you) agreed would be too divided. I’d say that an 80%—90% (honestly 90% in my personal opinion, but I can compromise) should be the bar. These modified spelling should only be used when they are nearly unanimous. Dantus21 (talk) 12:37, 20 September 2024 (UTC)

What are your thoughts on the tradeoff between precision and recognizability of RR? I initially had the same opinion as you, but then I thought about it and realized few people even recognize strict RR in the first place, so recognizability is hardly there anyway. So then I weighed precision (probability of being correct with a surname) higher.

In other words, you could argue a 50.1% name is not enough to merit the sacrifice in recognizability. I'd argue a 70% name gets closer to meriting that sacrifice because of the high precision. I'm still on the fence though. seefooddiet (talk) 17:46, 20 September 2024 (UTC)

Ok, I finished the surname table and the people names section. The surname table is a headache; there's too many possible names. A huge lookup table would too much bureaucracy for little gain, so I decided to limit the table to the top 12ish most common surnames and only those with a >80% common spelling. Also, I added "Oh" and "Woo"; otherwise these are single-char names that are hard to read. This covers around 70% of the 2015 population of South Korea. Evidence is provided at the romanization essay. seefooddiet (talk) 02:36, 24 September 2024 (UTC)

Addresses

Should we use "-gu", "-si", etc for addresses? I believe this is what addresses actually use in South Korea; this feels more technically correct to me. Currently in WP:NCKO, we broadly recommend the use of " District" instead of "-gu" and remove "-si" altogether. May be good to add an exception to when formal addresses are being asked for. seefooddiet (talk) 18:46, 12 October 2024 (UTC)

Place names

Xposting and expanding upon Talk:Han River (Korea)#Hangang River. This is a discussion on how the South Korean govt recommends spellings like "Xgang River", "Xsan Mountain", etc, while the press continues to resist adopting this.

Our practices are weirdly inconsistent; we recommend "Xsan" type patterns for most things, but rivers are "X River" and provinces "X Province". I think this seems to match what Korea JoongAng Daily does though; if you search for various patterns along these terms you get more results that align with what we're doing.

Tl;dr I think we're doing the right thing? seefooddiet (talk) 19:53, 20 November 2024 (UTC)

It seems to be normal practice in English-language texts about Korea to say Mount X, X Mountains (or Mountain Range), X River and X province. Kanguole 23:14, 20 November 2024 (UTC)

Thinking about this; difficult to prove broad patterns like this. seefooddiet (talk) 10:41, 21 November 2024 (UTC)

North Korean romanization...

NK romanization is still a mess. I'm conflicted on whether we should continue to exclude North Korean romanization from use on Wikipedia. See Talk:Kim Ju-ae#Requested move 5 January 2025.

Following WP:KOREANNAME leads to "Kim Chuae" being the recommended romanization for her, but hardly anyone would recognize that spelling. "Ju-ae" or "Ju Ae" would be preferrable to "Chuae".

Need to think about this. seefooddiet (talk) 07:41, 6 January 2025 (UTC)

Fortresses naming consistency

Thinking about establishing a naming convention for fortresses/castles.

Currently there's mixed practice on Wikipedia: in Category:Castles in South Korea there's Doksan Fortress, Namhansanseong, Namhae Castle. But the majority of articles use the "fooseong" format. Not really sure what to do, but think it's better to have a standard than not for this case.

Various weak thoughts:

• Using "fooseong" would more closely reflect current practice.
  • We also name temples similarly. Also similar are some of our geographic features.
• UNESCO uses "Namhansanseong", not "Namhan Mountain Fortress", "Namhansan Fortress", "Namhansanseong Fortress", etc. Namhansanseong is arguably the most famous Korean mountain fortress.
  • Although counterargument: the South Korean govt will use "Namhansanseong Fortress"; they use patterns like this in general.
• It's hard to determine a general naming convention otherwise. A lot of these fortresses don't seem to have great coverage in English.

seefooddiet (talk) 22:51, 4 February 2025 (UTC)