Taiwanese and Old Norse words for "homestead, village"

« previous post | next post »

[This is a guest post by Chau Wu]

Tai Po District 大埔區 is one of the 18 districts of Hong Kong whereas 大埔县 (Dabu xian) in Guangdong is a Hakka culture center bordering on Southern Fujian. In Taiwan the term 大埔 (Tōa-po·) is found in about 40 place names such as 大埔鄉 Tōa-po·-hiong, 大埔村 Tōa-po·-chhun, 大埔里 Tōa-po·-lí, etc.

In fact, Tw 埔 (po·) ‘homestead, village’ is the most popular Taiwanese word in place names (Sino-Platonic Papers, No. 262, p. 123). The lexicographer 陳修 (Tân Siu) states in his 台灣話大詞典 (The Great Dictionary of Taiwanese, page 1379) that, “我們台灣以埔po· 為地名者特別多 (In Taiwan we use 埔po· in place names especially plentifully).”

Its corresponding word in Old Norse, bær ‘homestead, village’, is also the most popular word for naming places by the Vikings. Examples are: Sjöbo in Sweden, Maribo and Rødby in Denmark, Valebø in Norway, and Fellabær in Iceland. Its loan to English becomes -by as in Hornby, Gatsby, and the “by” in “bylaw”.           Pointing to its popularity, Cleasby and Vigfusson state that, "wherever the Scandinavian tribes settled, the name by or bö went along with them." (An Icelandic-English Dictionary, page 92). It appears that this unique Nordic custom of using bær/bo/by in place names is carried on in Taiwan.

In a recent sporting event, the Ironman 7.3 World Championship race, held in Nice, France, the winner was Gustav Iden from Norway.  He wore a cap from a temple 順澤宮 (Sūn-tek-kiong) in 埔鹽 (Po·-iâm) of Changhua 彰化, Taiwan (see this article).

A descendant of Vikings whose word bær went along wherever they migrated was wearing a cap with the word 埔 (po·) stamped on it and won the championship title. He said he liked to wear it in competition because it looks cool.  The temple is now swamped with orders for the cap, as described in this article.

It is interesting to note that, if Tw 埔 po· is indeed related to the ON word bær, then it has come full circle — with a hidden linguistic secret advertised on the champion’s cap.


"Norwegian athlete’s Taiwan 'lucky hat' goes viral:  Ironman Gustav Iden won his event in France but received international attention for his hat, which name checked a Taiwan temple", by George Liao, Taiwan News (9/9/19)

…[T]he hat he wore bore the legend: "Puyan Shunze Temple" (埔鹽順澤宮).

…The 300-year-old temple in Puyan Township (埔鹽鄉), Changhua County (彰化縣), honors the god Hsuan Tien Shang Ti. It has a mud statue of the god from when the temple was rebuilt after being destroyed by fire in 1948, the Puyan Township Office website confirms.

According to the CNA report, Iden noticed the hat lying on the ground when he took part in an Ironman competition in Tokyo. He picked it up, and took good care of it in the hope it would bring him luck – which it did, as he won the event in Nice, France….


Philological and phonological notes by Diana S. Zhang

Indeed 埔 is a late character and dialectal word. It is not even included in 韻鏡 that was attributed to the Five Dynasties Period. It should be indigenous to the Southern Min as a variant of 埠 "flat open land," but the Wu dialect also uses it as a variant of 浦 "river bank, estuary." The earliest written attestation in a Classical work should be two articles by 歸有光 (1507-1571), who was native to 昆山 (in modern Jiangsu), in which he recounted his maternal grandmother, Ms. Zhou's hometown:

"去縣城東南三十裏,由千墩埔而南直橋,並小港以東,居人環聚,盡周氏也。" ——《先妣事略》

“余外家世居吳淞江南千墩埔上。” —— 《悠然亭記》

When I looked at how 埔 was used as a 通假 for 浦 in Gui Youguang's time, I looked up the gazetteer of 昆山 during the 嘉靖 reign (1522-1566), right in the middle of Gui's lifetime. I found this: "相傳其北三十里,地名木瓜,有墩九百九十有九,與此合為千墩,因名。其下為千墩浦。澱山湖水由此入吳淞江。" (昆山縣志 vol. 3)

So this may be a borrowing from the Southern Min to the Wu. In any case, 埔 is a topolect word and no attestations are found in the Northern and Southern Dynasties written records to my own knowledge so far. But its entrance to Sinitic seems cleared up with this borrowing process: 埠——埔 (Min) and 埔——浦 (Wu). Then, in the Ming, with Gui's work, it entered the 漢文言 realm (I suppose).

Since neither 埠 or 埔 is contained in 韻鏡, and 1) the MC for 阜 buH and 甫 puX only contrast by initial voicing and tones; 2) both 埠 and 埔 are /pu:/ in the Min, I believe that it was natural for them to be variants for the Min native speakers.


"Green box deep male shrine" (8/31/19) — for a discussion of tourist sites around in Baihou Town 百侯镇, Tai Po 大埔, Guangdong Province

"An early fourth century AD historical puzzle involving a Caucasian people in North China" (1/25/19)


  1. Jonathon Owen said,

    September 10, 2019 @ 12:54 pm

    I know the Vikings really got around, but is there any evidence that they made it all the way to Taiwan?

  2. Suzanne Valkemirer said,

    September 10, 2019 @ 1:17 pm

    Given the fact that the number of phonemes in each language is relatively small (Taa,a language of Botswana and Namibia, is said to possibly have the largest number),

    Given the fact that the phonotactics of every language limits the number of possible sequences of phonemes,

    Given the fact that not all permissible sequences occur (see "Jabberwocky" for examples), and

    Given the fact that words tend to be short rather than long (no word consists of, say, 100 phonemes),

    Each of the world's languages is likely to have at least a few words that — by sheer coincidence — are similar in form and meaning to words in another language.

    The words given in the post are an example of the many such coincidences.

    [(myl) Indeed. See Don Ringe "Probabilistic evidence for Indo-Uralic." in Nostratic: sifting the evidence (1998), and works therein cited:

    I don't see anything in this post that would suggest even taking the trouble to make a systematic evaluation of the probability of chance similarity.]


  3. Andrew Usher said,

    September 10, 2019 @ 7:46 pm

    Exactly. If that is the point of this point, it is quite misguided. Not only is it almost sure that 'baer' and 'po' are unrelated, their use as very common elements in place-names is likewise.

    This is not to say – given the title of the work referred to above – that I deny the Nostratic hypothesis. There we have more than random lexical resemblances, and further good independent reasons to believe in the relation.

    k_over_hbarc at yahoo.com

  4. Suzanne Valkemirer said,

    September 10, 2019 @ 7:49 pm

    @ MYL. I agree, especially since the Norwegian word has only three phonemes and the Taiwanese one, only two.

  5. John Swindle said,

    September 10, 2019 @ 8:38 pm

    It depends on which character we use to write the "by" in "Oh, baby!"

  6. Jerry Friedman said,

    September 10, 2019 @ 9:20 pm

    Is there any attempt at doing something like the Ringe paper in regard to claims of borrowing? I guess all you could do would be to say that the number of candidate borrowings was suspiciously high—or that the number was so low that there was no reason to think similar words were borrowed without lots of specific evidence for a particular word.

  7. Chris Button said,

    September 10, 2019 @ 10:25 pm

    Outside of "the probability of chance similarity", it's worth looking at other factors that can influence the likelihood. A phonologically curious form (e.g. 犬), a bound form (e.g. 珊瑚), an isolated form (e.g. 熊), a peculiar usage (e.g. 來/麥), a particular graphic form (e.g. 巫), an outlier from attested word-families (e.g. 桂), external evidence of being a wanderwort (e.g. 馬)… Key to all of these is an adequate reconstruction of earlier historical stages of Chinese, and it's an unfortunate reality that Old Chinese reconstructions (at least those presented in any accessible and somewhat complete form to the public) are woefully inadequate in that regard and, furthermore, often obscure etymological relationships between words as much as they elucidate them.

  8. John said,

    September 11, 2019 @ 1:02 am

    I've never heard of 埔 being glossed as "homestead" or "village." My understanding has always been that it means "plain, flatland" and Taiwan's Ministry of Education online dictionary supports this too:


  9. maidhc said,

    September 11, 2019 @ 5:02 am

    Maybe instead of the Vikings going to Taiwan, it was the Taiwanese who made it to Denmark. And set themselves up as language teachers. They might have told the locals they were teaching them Latin or whatever was the prestige language at that time.

    Perhaps it's time to bring back Yu Ming Is Ainm Dom.

  10. maidhc said,

    September 11, 2019 @ 5:05 am

    I seem to have botched the link, it is at https://www.youtube.com/watch?v=JqYtG9BNhfM

  11. Sally Thomason said,

    September 11, 2019 @ 6:19 am

    It's unfortunate that a post like this has appeared on Language Log, which is meant to be a forum for reliable information and promising ideas about language.

  12. Philip Taylor said,

    September 11, 2019 @ 6:21 am

    "Yu Ming Is Ainm Dom" — Wonderful. Thank you so much for posting the link. For reasons that are irrelevant here, I was feeling pretty depressed when it arrived, and the opening of the film seemed to match my mood, but when the Irish barman asked "Hey, did you know that Paddy speaks Chinese ?" I burst out laughing, and the end of the film was just the icing on the cake. A very, very, sincere "thank you".

  13. Victor Mair said,

    September 11, 2019 @ 7:02 am

    Thank you, Chris Button. These are aspects of culture and history that linguists need to take into consideration.

    And how about the issue of Nostratic raised by Andrew Usher, since it is the context in which Don Ringe carried out his evaluation of probabilistic evidence? As Usher says regarding the Nostratic hypothesis, "There we have more than random lexical resemblances, and further good independent reasons to believe in the relation."

  14. Philip Anderson said,

    September 11, 2019 @ 7:25 am

    Is baer/bo/by really as universal in the Scandinavian world as claimed? In England it is common in the Danelaw, where the Danes settled, but not in names of Norwegian origin – I can’t think of any examples in Ireland either.

  15. Chau said,

    September 11, 2019 @ 8:27 am

    I would like to thank Professor Liberman and Suzanne Valkemirer for suggestions to look out for chance similarities. I am always mindful of the pitfalls. Allow me to introduce an approach I have taken to circumvent this problem. Because it is impossible to insert a figure in the comment section, I shall verbally describe this approach as best as I can and I do hope you could visualize it. This Website does not allow subscripts, therefore, I will spell out (subA), (subB).

    Let’s say Language A has a set (vocabulary) of words distributed on X,Y coordinates. The two axes are: X for sound and Y for meaning. A specific word in Language A can then be defined as [X(subA),Y(subA)]. As mentioned by Valkemirer, by sheer coincidence one is likely to find a word in Language B defined as [X(subB),Y(subB)] to be similar in sound and meaning. Let’s say the probability of such a similarity by coincidence is P(X,Y).

    Now let’s introduce another axis Z which is perpendicular to the X,Y plane. This axis is for cultural specificities (i.e., customs). The scale of the axis is divided into several groups (categories), each group is then further subdivided, which in turn is further subdivided, and so on. To assign a number to a specific custom, we can use a system analogous to the library catalog numbers for books (such as the Library of Congress catalog number). Now Language A has its vocabulary defined by a 3-D space where words are distributed along X,Y,Z axes. Similarly, Language B has its own 3-D distribution.

    Let’s use the ON word bær as our example. In the X,Y two-dimensional system, this word is spelled b-æ-r and has the meaning of ‘homestead, village’. On the Z-axis, it is located on a point defined as “Geography” > subcategory “landmass” > subcategory “human habitation” > subcategory “place names” > subcategory “favorite place name” > unique feature “the most favorite place name”.

    Now, because cultural features are not limited by the number of phonemes, the phonotactics, limitation of permissible sequences, or word lengths, whereas the variety of cultural features is rather open-ended, hence the denominator N in calculating chance probability is extremely large. Thus, the probability P(Z) of similarity due to chance, which is equal to (1/N), would be extremely small.

    ON bær is the most popular lexical element for place names in Nordic customs, as testified by Cleasby and Vifugsson. Tw po· 埔 is also the most popular word for place names in Taiwan as affirmed by Tân Siu 陳修. The probability due to coincidence of their similarities in sound, meaning, and cultural specificity is P(X,Y) × P(Z). Because of the extremely small value for P(Z), I would say the product of the two P’s is infinitesimally small.

  16. Jen in Edinburgh said,

    September 11, 2019 @ 10:47 am

    Philip Anderson: the Scottish 'equivalent' seems to be bolstadr, producing -bister in the north, and -bost in Lewis and Skye. I'm not sure if that's a linguistic difference (NO v. DK) or a land use difference or something else.

  17. Philip Anderson said,

    September 11, 2019 @ 4:19 pm

    Although Cleasby and Vigfusson (https://www.ling.upenn.edu/~kurisuto/germanic/oi_cleasbyvigfusson_about.html), 1874, may be the authoritative Old Icelandic dictionary, is it an authority on Old Norse place names in general?

  18. Jen in Edinburgh said,

    September 13, 2019 @ 6:17 am

    Sorry, -bolstaðr. I do know what an eth looks like, I was just writing from memory and getting confused with sted/stadt.

    On the Z-axis, it is located on a point defined as "Geography" > subcategory "landmass" > subcategory "human habitation" > subcategory "place names" > subcategory "favorite place name" > unique feature "the most favorite place name".

    I'm not at all sure I'm understanding this correctly. If one landscape is made up mostly of high hills and deep valleys, and another of flatlands and shallow pools, the common words used in describing them are likely to be very different – and you wouldn't expect the word which means 'peak' in the first landscape to turn up meaning 'pool' in the second, just because they're similarily common.

    I suppose it's possible that settlement names (or settlement patterns themselves) tend to follow a common distribution which isn't much affected by differences in culture or landscape, and so the common meanings are common everywhere, although I have no idea if this is actually the case – it seems unlikely to me, but many unlikely things are true).

  19. Gunnar H said,

    September 14, 2019 @ 2:16 am

    This theory seems to ignore that Old Norse bær (shouldn't that be bœr?) is not some isolated, inexplicable word. It's part of a group of Norse words with related meanings (such as the ból in -bolstaðr, and in modern Scandinavian bo "to live/reside", bod or bu "shed/booth", bygg "building", bol "nest/lair/hive", bur "cage", by "village/town/city", etc., etc.), and cognate words in languages that are much more obviously related than Taiwanese (such as German bauen and Bauer, or English bower).

    All of which allows the reconstruction of a proto-Germanic root, būaną and its most likely meaning. If we insist on seeking some link to Taiwanese, it must be with this earlier form.

  20. Martyn Cornell said,

    September 15, 2019 @ 2:40 pm

    Taipo in Hong Kong is where the South China Morning Post is printed. I always found the similarity to the English word “typo” amusing.

  21. James Kabala said,

    September 19, 2019 @ 8:30 pm

    Isn't this kind of thing how people used to try to prove Welsh Indians and the like?

  22. Chris Button said,

    September 19, 2019 @ 9:47 pm

    I think you're referring there to prosodic similarities between Welsh English and Indian English, from whence came the term "Bombay Welsh"

  23. Philip Anderson said,

    September 20, 2019 @ 6:57 am

    @Chris Button
    No, he is talking about the peoples now usually known as Native Americans, in particular the Mandan.

    It was claimed that their language included Welsh words, and that they were descended from the Welsh prince Madog and his companions.

  24. James Kabala said,

    September 20, 2019 @ 3:00 pm

    Sorry – I guess I was in the spirit of those who theorized in that way. I doubt if anyone who believed in Prince Madog also used the term Native Americans! But I am kind of intrigued by the Bombay Welsh (Mumbai Welsh?) now.

  25. Chris Button said,

    September 20, 2019 @ 5:45 pm

    Ah I see! I hadn't heard about that before. But I am equally intrigued.

    To be clear, "Bombay Welsh" only refers to the coincidentally similar prosodic characteristics of Indian English accents with those from south Wales.

RSS feed for comments on this post