What Percentage of a Country's Surnames Are New — and When They Appeared

South Korea's 2000 census counted 286 surnames. The 2015 census counted 5,582 — a twentyfold jump in fifteen years. Read quickly, that looks like an explosion of new names. It is nothing of the sort. No new Korean surname was invented in those fifteen years; the counting changed, and most of the "growth" is a definition quietly swapping itself out underneath the number. That gap — between what a rising surname count looks like and what actually moved — is the subject of this article.

"What percentage of a country's surnames are new" is one of those questions that sounds precise and turns out to have no clean answer almost anywhere. Registries count new names differently, or don't count them at all, or count something adjacent and label it "new." So rather than manufacture a false precision, this piece does two things: it shows the handful of places where the data genuinely exists, and it is honest about the much larger space where it does not.

The Korean twentyfold jump is mostly a definition change

Statistics Korea (KOSTAT) is the source of the 286 → 5,582 figures, and KOSTAT itself is clear that the increase is driven by naturalised citizens and multicultural families rather than by any change in native naming. Unpacking the two endpoints shows why the ×20 is an artefact of comparison, not a discovery:

Of the 2015 total, 1,507 surnames have a Hanja (Chinese-character) reading behind them and roughly 4,075 do not — overwhelmingly foreign-origin names written phonetically into Hangul by naturalised citizens and their families, plus the effect of switching to administrative-record methods that capture more distinct strings. The 286 figure from 2000 was, in effect, counting only the traditional core. Comparing it to the 2015 all-strings total is comparing two different objects.

The tell is what happened to concentration. If Korea's surname stock had genuinely fractured, the grip of the biggest names would have loosened sharply. It barely moved: the ten most common surnames covered 64.1% of the population in 2000 and 63.9% in 2015 — essentially unchanged. Kim alone is still about a fifth of the country. A twentyfold rise in the number of surnames alongside a top-ten share that holds steady to a tenth of a percentage point is the signature of a methodological expansion, not a demographic one. KOSTAT's own phrasing — that the count grew because of naturalisation and better methods — says as much.

So the honest reading of Korea is: the number of surname strings on file rose about twentyfold; the number of traditional Korean surnames did not change; and the newly counted names are real people's names, but "the surname count grew ×20" and "×20 more kinds of surname now exist in Korea" are two different sentences, and only the first is true.

France: the tail is swelling, and you can date it

France offers the rare thing — a way to see when new names entered, because INSEE's Fichier des noms tabulates surnames by the birth decade of their carriers, from 1891 to 2000. Alongside the named surnames, the file carries an aggregate row for rare names that are too infrequent to list individually — an "other names" bucket. That bucket is where newly arriving, still-rare surnames land before any single one of them is common enough to earn its own line.

Measured as a share of each birth cohort, that aggregate grows across the twentieth century — on the order of 6% of the 1941–1950 cohort rising to roughly 11% of the 1991–2000 cohort. In other words, among people born in the 1940s, about one in sixteen carried a surname rare enough to be swept into the residual category; among those born in the 1990s, closer to one in nine. The tail is thickening, and — consistent with the individually named cases like Traoré and Diallo, which climb steeply across the same cohorts — much of the thickening is the accumulated arrival of many individually rare names.

Two limits keep this honest. First, the same file excludes people born abroad, so first-generation arrivals are absent and the residual share is a floor, not a ceiling — the true widening is larger. Second, "rare name" is not the same as "new name": the bucket also holds genuinely old, genuinely rare French surnames. The growth of the bucket is strong evidence that the base of names is broadening over time, but it is not a clean "% new per decade," and it should not be presented as one.

Names that simply were not there before

The most concrete version of "how many surnames are new" is a direct before-and-after: take a surname list from one date, take it from a much later date, and read off the names that appear in the second but not the first.

The Netherlands allows exactly this. Dutch genealogical analysis of the surname registry finds that the most common surnames in 2007 that were absent from the 1947 list are, in order of frequency: Yilmaz, Nguyen, Ali, Mohamed, Ahmed. Every name in the mid-century top ranks sounded traditionally Dutch; sixty years later, the most common newcomers are Turkish, Vietnamese and Arabic. Yilmaz — Turkey's single most common surname — plus Kaya and Öztürk now register where, in 1947, there was no entry at all. This is the cleanest kind of evidence for "new": not a name climbing a ranking, but a name for which the earlier ranking had no row.

Sweden shows the same thing without a paired historical file. Its current surname statistics contain layers — Arabic (Ahmed, Hassan, Ibrahim), Vietnamese (Nguyen) — that a Swedish census of the 1950s would not have recorded in any measurable quantity. The layer is new in the plain sense that it was previously not there; what Sweden lacks, unlike the Netherlands, is a tidy 1947-versus-2007 pairing to put a single percentage on it.

Why "% of surnames that are new" has no clean answer

Set the cases side by side and the reason the question resists a number becomes obvious. Each country answers a subtly different question, with a different threshold, and the threshold is almost never stated in the headline.

Three structural problems recur. Thresholds: every surname count is really "how many surnames have at least N bearers," and changing N moves the total by an order of magnitude — a name can cross from "unlisted" to "listed" without a single new person being born. Naturalisation and registration: a surname that existed abroad for centuries is, from the registry's point of view, "new" on the day its first carrier is entered — so "new to the registry" and "new in the world" are different facts wearing the same word. Methodology: as Korea shows, switching how the census tabulates names can multiply the count without any real change on the ground.

There is also no cross-national convention. Some registries suppress counts below three, others below five, others below twenty; some publish by birth cohort, most by living snapshot; some (Korea) count phonetic strings, others (France) count listed forms with a residual bucket. Stacking these into a league table of "% new surnames per year" would be comparing measurements that were never taken the same way. For most countries the figure simply does not exist, and the responsible move is to say so rather than invent one.

What we can and cannot say

What the data supports, stated conservatively:

• Where a country publishes a surname list at two dates, the newcomers are datable and nameable. The Netherlands is the clearest example: the top surnames absent in 1947 and present in 2007 are known by name.
• Where a country tabulates by birth cohort, the widening of the name base is visible over time. France's rare-names aggregate grows across cohorts, and individually named migration-origin surnames climb steeply across the same span.
• A rising surname count is not, by itself, evidence of rising surname diversity. Korea's count rose twentyfold while its top-ten concentration held at ~64%. Always ask what changed: the population, or the counting rule.

What the data does not support: a precise "X% of country Y's surnames are new as of year Z." That number is not published, not comparable across borders, and — because of thresholds, naturalisation, and methodology — not even well-defined for most countries. The most useful honesty here is to separate the two firm anchors (Korea's 286 → 5,582 with its caveat; the Netherlands' 1947-to-2007 newcomers) from the much larger territory where the only correct answer is "the base of names is broadening, but no one has counted the percentage — and here is why they can't."

Data as of 2026-07-18

Figures are drawn from the cited national-registry sources and their published commentary. Where a number is a comparison artefact rather than a real-world change, that is stated in the text above.

• South Korea — KOSTAT (Statistics Korea), Population and Housing Census. 286 surnames (2000) → 5,582 (2015); of the 2015 total, 1,507 Hanja-based and ~4,075 non-Hanja; top-ten share 64.1% (2000) → 63.9% (2015). The increase is attributed by KOSTAT to naturalised citizens, multicultural families, and methodology. <kostat.go.kr/&gt; · corroboration: <en.wikipedia.org/wiki/List_of_Korean_su…;
• France — INSEE, Fichier des noms de famille (surnames by decade of birth, 1891–2000), including the aggregate "other/rare names" row. The file excludes people born abroad, so the residual share is a lower bound. <insee.fr/fr/statistiques/353663…;
• Netherlands — surname registry, 1947 vs 2007 comparison as reported in Dutch genealogical analysis: most common 2007 surnames absent in 1947 are Yilmaz, Nguyen, Ali, Mohamed, Ahmed. <dutchgenealogy.nl/popular-dutch-surnames…;
• Sweden — SCB (Statistics Sweden), name statistics. Arabic and Vietnamese surname layers present in current statistics and absent from mid-century records; SCB discontinued name-statistics production in 2024. <scb.se/en/finding-statistics/…;

On sensitivity and scope. This article is about how registries count surnames, not about migration as a phenomenon or the people behind the names. "New" throughout means "newly present in a registry's records," a purely administrative fact. No claim is made about identity, ethnicity or origin, and the central caution of the piece is precisely that a growing count often reflects a changed definition rather than a changed population.