Talk:Imjonseong Fortress

Requested move 9 February 2025

Imjonseong Fortress → Imjonsŏng – I recently WP:BOLDly moved to the proposed target, but was reverted with rationale "the only source in the article uses Imjonseong Fortress".

This is a perennial problem for romanization of Korean terms. In general, MOS:KO and WP:NCKO assume that a single attestation to a spelling is not enough to adequately establish a romanization, because there are so many conflicting romanization practices in use.

Either way, the practice for this is clearly lined out in WP:KO-BUILDING: we establish COMMONNAME (I'd argue one doesn't exist; attestations to this fortress in English are sparse), and if one doesn't exist, we romanize following MOS:KO-ROMAN. KO-ROMAN says it's a pre-1945 concept, so we follow McCune–Reischauer, which is Imjonsŏng. seefooddiet (talk) 22:06, 9 February 2025 (UTC)

This article has one source, which originates from the Korean government and uses Imjonseong Fortress. I appreciate all romanization difficulties, I worked with zillions of Russian and Ukrainian articles where one has the same issue, but with all respect, unless we have more sources, I would oppose. Ymblanter (talk) 15:55, 10 February 2025 (UTC)

Weak oppose. I'm familiar with the difficulties here, including the ongoing debate about the utility of the current MOS. Where the sole source used in the article uses RR, which is also accessible to a general English-speaking audience (because it lacks the ⟨ŏ⟩), it seems we should give some weight to that. Google Scholar returns only one result for Imjonsŏng Fortress and a handful for Imjonseong Fortress (most of these come for the same two sites). WP:TITLECHANGE favors maintaining longstanding, stable article titles absent a compelling rationale. For these reasons, I oppose, but would defer to editors with more expertise and better arguments. --MYCETEAE 🍄‍🟫—talk 18:10, 10 February 2025 (UTC)

I've been thinking about above few points for past few days. The reason the MOS and NCKO are built on not trusting single attestations is because there are like 7-8 different groups that each have different standards. If you take a single sample you risk sampling from any of those groups. Ymblanter mentions "the Korean government", but remember that there are two of them, and that other groups outside of the Korean govts disagree with both of them. Even if we choose to obey one of those standards, people have complained in the past that we haven't listened to some other group, and their complaints aren't without merit.

This is the reason why we have these defaults built into MOS:KO-ROMAN. When we rely on single attestations, we risk swinging between titles each time new attestations are added. It also hurts WP:TITLECON. Per the above comments, "unless we have more sources" is the issue. Each new source would likely swing the title.

It's been well-established that in general, writings on Korean history still majority use McCune–Reischauer, per WP:ROMANKO. This is the reason we default to MR. But if you decide to value single attestations over the default practice, the default practice is effectively worthless; you can find single attestations to basically any spelling, including ad-hoc, for most Korean terms. So then we shouldn't have defaults at all.

In short, I wavered for a bit then reaffirmed my belief that relying on single attestations is likely to be distracting and costly for us because of how messy Korean romanization is. We have well-researched and argued default assumptions that we should rely on. seefooddiet (talk) 15:50, 13 February 2025 (UTC)