Tap water water

« previous post | next post »

From a friend who is travelling in Japan:



inryousui to shite mo goriyou itadakemasu 飲料水としてもご利用頂けま す
("Can also be used as drinking water")


Tap water can be used for drinking water


Sudonmureul siksuro iyonghasyeodo doemnida // Sudosmul-eul sigsulo iyonghasyeodo doebnida*
수돗물을 식수로 이용하셔도 됩니다
("It is all right even if you use tap water for drinking" –> "It is all right to use tap water for drinking // You may use tap water for drinking." –> "Please feel free to use tap water for drinking")


Nǐ néng hē zhège zìláishuǐ shuǐ 你能喝这个自来水水
("You can drink this tap water water")

Think about it: you can drink the water of this tap water.

*A note on Korean Romanization from an anonymous colleague:

As for the romanization, it depends on who you're trying to talk to. Both spellings you have are versions of RR ("Revised Romanization"), the dogmatically official system of the Republic of Korea. But before going into detail about the difference between the two, I can't resist venting a bit.

I personally have a dislike for RR, because its creation was a nationalistic strong-arming of non-Korean opinions. Most Westerners hated it from the beginning; many still do. But another, more specific reason for my distaste is that the transcriptions of two of the vowels are an embarrassment. What I mean specifically is that the representation of the vowel ㅓas [eo] comes from the spelling "Seoul", which was contrived by French missionaries, who used [se] to represent the syllable 서 and [oul] to represent 울. Thus, [seo] for 서 is a mistaken segmentation made presumably by someone who didn't know French or the history of the spelling and probably thought [Seoul] was an English-based spelling. Then, to add to the misery, someone decided to create the transcription [eu] for the vowel ㅡ by analogy with the first mistake, [eo]. I'm told that around 1999 the gathered worthies who made the romanization decisions in 2000 knew about these problems (a friend of mine who was part of the discussions told me he objected strongly), but the group went ahead anyway, almost, it seems, to spite Western objections. They saw themselves disrespected by Westerners who had laughed and joked about the crazy spellings of an earlier version of RR that had been posted on signs around Seoul. (The most famous of those unfortunate spellings was "Dog Rib Gate"–sometimes rendered "Dog Lib Gate"–for 독립문 "Independence Gate" [Chinese 独立門]. There were so many jokes about that, you can't believe!)

Of course, Koreans had a point about Westerners being imperious and obnoxious and culturally biased, and, to a certain extent, even racist. But this romanization was not the way to assert Korean pride and independence, at least in my opinion. In some cases, the pushing of RR became militant over the next decade. Government publications still will not accept any other spelling systems, nor will degree-granting institutions when considering applications. I have friends who, on their own websites, had used the older and much more dignified (not to say staid) McCune-Reischauer system (still used by LC and, heretofore, by most Western publications), and yet they were harassed and battered by Korean hackers and emailers insisting that those spelling "mistakes" be corrected. Similarly, corrections and battles continue on Wikipedia to this day. So unpleasant.

Okay, having gotten that off my chest, let me get to the matter you actually asked about, your two versions of RR. The official document promulgating the "Revised Romanization" specified that there could be two versions of the new system, and that's what you have. The first was for more general, non-technical uses, such as the spellings of names and organizations in Western writings. The spellings in this more general system took into account the sound changes that notoriously take place morphophonemically in Korean (assimilation of the nasality in your last word, for example, or Dongnimmun instead of Dogribmun for 독립문 'Independence Gate'). The second was said to be a more "precise" transcription of Hangul to be used by professional linguists (so, Dogribmun), transcribing each Hangul letter by matching it in a one-to-one way with a Roman letter or digraph. (As a professional linguist myself, though, I can tell you this more "professional" system is not really practical for most linguistic purposes, especially ones tracking historical changes. For that purpose I use the expanded version of Yale Romanization Martin devised in his matchless Reference Grammar of Korean.)

You might also be aware that the North Koreans don't use RR anyway; they still use their own romanization system based on MR. That means that if and when the two Koreas ever get together, they'll have to work out those differences–as well as what they want to call the country!

And one more thing: One very important factor that led to the creation of RR was the creation of Pinyin. The fact that the Chinese had what Koreans perceived as total independence in that creation process caused Koreans to regard Pinyin as a model for asserting national self-respect.

[h.t. to Jing Wen; thanks to Bill Hannas, Minkyung Ji, Bob Ramsey, Haewon Cho, Daniel Sou, and Nathan Hopson]

Update, 6/16/13: J. Marshall Unger of Ohio State University sends the following:

Defects in the Ministry of Culture Romanization for Korean

    1. Yale romanization is better designed.
      1. YR uses <.> and <q> to indicate reinforced obstruents not notated in hankul in a principled manner.
      2. The YR distinction <u> ≠ <wu>  can be extended to handle Middle Korean /o/ (so-called alay a)≠ /wo/ consistently. (In fact, there is morphophonemic evidence that the /w/ in these MK vowels was originally segmental.)
      3. YR has long been the de facto standard among linguists.
      4. YR is more in harmony with the principles that are said to make the Korean alphabet a model of linguistic accuracy and efficiency.
    2. Representing aspirated obstruents with p, t, ch, k but unaspirated obstruents as b, d, j, g or p, t, ch, k depending on position, is misleading.
      1. The phonetics of Korean are different from those of Chinese.
        1. The voice-onset lag for aspirates is much longer.
        2. Plain Korean obstruents, unlike those of Chinese, are phonetically about as aspirated as automatically aspirated English initial voiceless obstruents.
      2. Phonemically, aspiration in Korean is segmental, as shown by regular rules of metathesis (/h.k/ > [kh] etc.). 
      3. The differences in initial and non-initial obstruent representation perpetuate a feature of McCune-Reischauer romanization that many felt was objectionable.
      4. The new notation leads to a radical change in the initial consonant of many words, thus complicating sorting, indexing, and other practical applications.
    3. The choice of the digraphs <eo> and <eu> is inapproriate.
      1. The motivations for these choices are mistaken.
        1. Use of <eo> is based on a misegmentation of the old French romanization, in which <Seoul> has <e> for /e/ and <ou> for /wu/. 
        2. The use of <eu> for /u/ is strictly by analogy.
      2. Phonemically, consonantal /w/ combines with /u/ to produce /wu/, a fact captured by the YR notation <u> ≠ <wu> and sacrificed in the new romanization.
      3. Although McCune-Reischauer <ŏ> and <ŭ> are avoided, no system notates phonemically long vowels in the standard language, when necessary, except with macrons, so diacritics have not been entirely eliminated.
    4. Political considerations ought not contaminate policy decisions concerning orthography.
      1. The representation of aspirates is modeled on the Pinyin romanization of Chinese.
      2. Instead of creating an international standard, the MOC romanization has merely introduced a new system that muddies the waters.
      3. The rejection of YR is motivated in part by its similarity to North Korean romanization.
      4. It is also rejected in part because of “not invented here” nationalism.


  1. cM said,

    June 15, 2013 @ 6:23 am

    I was slightly confused by the "venting" part at first, for example: "…who used to represent the syllable 서 and to represent 울."

    Looking at the source code, there are double blanks in the text where I would have expected transcriptions.

    It seems some software somewhere ate strings, presumably because they were enclosed in < >.

  2. Victor Mair said,

    June 15, 2013 @ 7:02 am


    Thanks for pointing those things out. We had a system change at around 5 p.m. yesterday, just when I was attempting to make this post, so that may account for some of the problems to which you are referring. I'll try to get them sorted out within the next hour or so.

  3. Nic said,

    June 15, 2013 @ 7:29 am

    I think "for" is used as "as", rendering drinking water into an NP and being much closer to the original meaning

  4. Victor Mair said,

    June 15, 2013 @ 8:54 am

    I think I know what happened. In our system, anything within angular brackets gets stripped away, so I've replaced all of that material with romanizations inside of square brackets, thus [eo], etc. I hope everything is all right now.

    [(bgz) You can get around the garbling by using the &lt; and &gt; codes, like so: < >]

  5. flow said,

    June 15, 2013 @ 11:23 am

    this reduplication thing is a japanese thing. it's not uncommon to hear and see 'Fujisan Mountain' or 'Kamogawa River'.

    in kyoto—an unusual city where some streets do have names—you'll find one major road named 東大路通り, Higashi-Ōji-Dōri. now 通りdōri of course is just 'street', but 大路 ōji is '(broad) street', too. presumably at some point in history, the name was just 東大路 'Eastern Broadway', but as it became more common to call streets this-dōri, that-dōri, whatever-dōri (like japanese ships which are all named whatever-no-maru), and add to this that ōji is probably not terribly transparent to many japanese anyway, it became 東大路通り 'Eastern Broadway Street'.

    to my surprise and delight, there's still that hilarious street sign standing in kyoto that i happened to find on a bicycle tour in 1995. you may have a glimpse here: http://goo.gl/maps/ahhZQ and here: http://goo.gl/maps/Lxs0N —it's not terribly easy to read, but yes, it says 東大路通り and in english: Higashi-Ōji-Dōri Ave.! East Grand Street Street Street.

  6. HP said,

    June 15, 2013 @ 2:46 pm

    @ flow: The Midwestern US town I grew up in has a major thoroughfare named "Boulevard Road," FWIW. Maybe not a specifically Japanese thing?

  7. Jorge said,

    June 15, 2013 @ 3:04 pm

    That's interesting about the French. Well, c, q, r, x, z, zh (especially when some of these are followed by i) are hardly the best way to romanize their respective mandarin sounds too, so at least the koreans got that right about copying pinyin

  8. George Grady said,

    June 15, 2013 @ 5:41 pm

    Hickory, North Carolina, has a lot of street names like this: 19th Avenue Place NW, 10th Street Boulevard NW, 4th Street Place NW, 3rd Street Lane NE, 23rd Street Drive SE, 8th Avenue Drive SW, etc., etc., etc. It's really quite bizarre.

  9. Sjiveru said,

    June 15, 2013 @ 6:13 pm

    So RR is built off of a misunderstanding, but if we happen to find MR aesthetically unsatisfying, what should we use, then? :P

  10. David Morris said,

    June 15, 2013 @ 9:24 pm

    I have an irrational fear of diacritics, even when they are an integral part of whatever-it-is (language, moderately advanced maths etc).
    In 2006, as I was gaining my first TESOL certificate, I investigated countries I could possibly live and teach in. I decided on Korea partly because of the spelling system – I had a reasonable chance being able to read Korean, compared to Chinese or Japanese. The phrasebooks/ textbooks I looked at used a number of transliterations, but the one I bought used RR, so I just went with that. (If RR hadn't been introduced, and I'd had to use MR or Yale, I probably still would have gone to Korea. I have an even greater fear of Chinese and Japanese scripts.)
    For *generalist* purposes, I would unhesitatingly use RR, and recommend that anyone else does.
    Final thought – why does anyone need to use romanisation of Korean, given how easy it is to read hangeul? (I was about half way between the wise man who 'can acquaint himself with them before the morning is over' and 'a stupid man [who] can learn them in the space of ten days.')

  11. Akito said,

    June 15, 2013 @ 10:23 pm

    Re reduplication in "Kamogawa River", etc.

    I don't know if this is particular to Japanese or not (I think not). There is a problem with "Todaiji", "Toshogu", and the like. It would be odd to say "(the) Todai temple", "(the) Tosho shrine", because "Todai" and "Tosho" (in these cases) are not independent words. Better say the whole words, and yet, for people new to the language, short descriptive words are necessary, especially on first mention. There is no easy way out.

  12. Akito said,

    June 15, 2013 @ 10:33 pm

    I think "for" was used by analogy with (say) "The cup/glass provided can be used for drinking water," where "drinking water" has a different grammar.

  13. Colin Fine said,

    June 16, 2013 @ 3:09 am

    Since "drinking water" (water for drinking) is a familiar phrase to me, I didn't find anything odd in the English.

  14. Daniel Trambaiolo said,

    June 16, 2013 @ 4:45 am

    The awkwardness of the Chinese translation is striking. Even Google does a better job: "飲料水としてもご利用頂けます" -> "它可作為飲用水". (My own attempt at translation: 此水可飲用—although this might be a little too terse. Perhaps something like 自來水請放心飲用 might be more appropriate?)

    The only scenario that I could imagine leading to this sort of thing is that somebody in the company claimed at their job interview that they could speak Chinese, the boss remembered this when preparing the label, and they were too embarrassed to confess that they weren't up to the job of doing the translation. This is the sort of thing that happens all the time with English, but it is a sign of the times that it is now also happening with Chinese.

    In addition to the peculiar reduplication in 自來水水, one of the most interesting things about the Chinese translation is the use of 你能 ("you can"). This is the sort of thing that one never sees on labels or signs, which generally use something more like a passive construction. None of the other languages include a second-person pronoun (although it is implied by the honorifics in the Japanese and Korean), and its inclusion is more characteristic of spoken language than writing on labels. The fact that the translator felt the need to include it is typical of the sorts of strategies that semi-competent speakers use to ensure that their message gets across, increasing the level of redundant information beyond what a native speaker would use.

    Incidentally, can anyone here pass judgement the naturalness of the Korean translation? To me it seems at most very slightly awkward; about the same level as the English, and certainly much better than the Chinese. (I would have expected a label to use 수돗물을 rather than 수돗물을 or to rely on context and omit the phrase entirely, as in the Japanese. But my Korean isn't great, and I could be wrong about this.)

  15. Victor Mair said,

    June 16, 2013 @ 6:09 am

    @David Morris

    I hear what you're saying about diacritics. It's no wonder that you didn't even consider Vietnamese at all!

    As for why we need Romanization for Korean, it's because there will never be more than a tiny fraction of the total number of people on earth who know Hangul / Chosongul (Han-geul / Chosǒn'gǔl). If we want to explain things about Korea — its language and literature, people, history, and culture — to those vast numbers of people who don't know Hangul / Chosongul (Han-geul / Chosǒn'gǔl), we need to have a functional Romanization for that purpose.

  16. Victor Mair said,

    June 16, 2013 @ 8:13 am

    @Daniel Trambaiolo

    Excellent analysis on the Chinese! I hope that someone will take you up on the Korean.

  17. Gene Buckley said,

    June 16, 2013 @ 8:57 am

    The second was said to be a more "precise" transcription of Hangul to be used by professional linguists (so, Dogribmun), transcribing each Hangul letter by matching it in a one-to-one way with a Roman letter or digraph.

    The usual term for this is transliteration rather than transcription. The relationship between the two is an important means of illustrating how the writing system works; it may in turn reflect the phonology or history of the language, depending on the nature of the writing system.

  18. Coby Lubliner said,

    June 16, 2013 @ 9:16 am

    Not knowing anything about Korean, I have never understood why a name pronounced [no] is spelled Roh. Can anyone explain?

  19. Daniel Trambaiolo said,

    June 16, 2013 @ 12:28 pm

    @Coby Lubliner
    My understanding is that the transcription of [no:] as Roh for the surname relates to the etymology and historical pronunciation. This surname is derived from the Chinese surname Lu 盧, for which the expected Korean equivalent would be /ro:/ 로, but which has become pronounced as /no:/ 노 in the modern language. This phonetic change is specific to the surname; other words pronounced as /ro:/ 로 have not undergone a similar change to /no:/ 노.

    There is a similar change in the pronunciation of the Chinese surname Li 李, for which the expected Korean pronunciation would be /ri:/ 리, but which is usually pronounced in modern S Korea as /i:/ 이. For historical figures, this surname is sometimes written as Yi (e.g. Yi T'oegye, Yi Yulguk), in accordance with the usual modern pronunciation; however, contemporary S Koreans often transcribe it as Lee (e.g. Lee Myung-bak) or Rhee (e.g. Syngman Rhee), in accordance with the etymology.

  20. Daniel Trambaiolo said,

    June 16, 2013 @ 12:31 pm

    Looking back over my earlier comment, I realize that the penultimate sentence is nonsensical—what I meant to write was: "I would have expected a label to use 수돗물은 rather than 수돗물을 or to rely on context and omit the phrase entirely."

  21. Coby Lubliner said,

    June 16, 2013 @ 12:55 pm

    I can understand why historical-etymological considerations might play a role when romanization is adopted for a language's own use. (I remember being flabbergasted when the leader of the Malagasy musical group Tarika, the singer Hanitra, introduced the band as [tak] and herself as [hanʧ], but I figured there must be a reason.) But do Koreans use romanization for themselves, the way the Chinese use pinyin? If not, I don't understand the reason for non-phonetic transcription.

  22. Chau Wu said,

    June 16, 2013 @ 9:29 pm

    Reduplication seems to happen to Taiwanese river names too. A river that flows by the city of Taipei is named Tām-chúi-hô 淡水河 (Tamshui River). 河 hó [Tw. hô] is ‘river’ as in Huáng-hó 黃河 (the Yellow River) whereas 水 shŭi [Tw. súi (lit.) /chúi (vern.)] also means ‘river’, formerly used in river names such as 渭水 Wèishŭi and 洛水 Luòshŭi in Old Chinese times. The word shŭi 水 ‘river’ is also seen in the title of the classic Shŭi-jing-zhù 水經注 ‘The Annotated Book of Rivers’.

    An intriguing historical fact prompted me to dig deeper into the river name. When colonists started to migrate en masse from Fujian to Taiwan in the 17th century, they first settled at the mouth of a large river near the northern and southern ends of the island. They named the respective river Tām-chúi-hô 淡水河 independently. In late Ching Dynasty when communications between the northern and southern parts of Taiwan flourished, the river in the south was renamed Ē-Tām-chúi-hô 下淡水河 (Lower Tamshui River) to avoid confusion. The new name continued to be used during the Japanese era.

    I was wondering why the early settlers named large rivers Tām-chúi-hô consistently. Then I found precedents in the names of four major rivers that empty into the Black Sea. Enumerating from the east, River Don empties first into the Sea of Azov at Rostov and then the Black Sea. The Dnieper does so at Kherson and the Dniester near Odessa. Finally the Danube empties into a large delta in the Romanian coast. All these river names have their roots in Proto-Indo-European *dānu- ‘river’ (Pokorny, 1959, Indogermanisches eytmologisches Wörterbuch, I, 175; Mallory and Mair, 2000, The Tarim Mummies, p. 106). The Avestan dānu- is the base for Ossetic Don ‘river’, the modern Dnieper can be derived from *Dānu apara ‘the river to the rear’, and Dniester from *Dānu nazdya ‘the river to the front’, with ‘rear’ and ‘front’ being from the perspective of the Scythian homeland. The Celtic Dānuvius, derived from dānuv-, gave rise to one of the most romantic river names Danube.

    So it is my hypothesis that following the sound development of u to v, dānu- ‘river’ became *dānv-, (alternatively, with elision of -u-, *dānuv- > *dānv-) which then in turn gave rise to tām 淡, the first element in the name Tām-chúi-hô 淡水河.

    The late Professor Tsung-tung Chang of Frankfurt, Germany, derived liú 流 (Tw. liû) ‘to flow’ from PIE *sreu- ‘to flow, river’ (Sino-Platonic Papers, No. 7, 1988, p. 31). I see that PIE *sreu- can also give rise to súi 水 (Tw. vern. chúi) ‘river’ with simplification of the initial cluster, the common word for river names in Old Chinese, as mentioned above. There is a pattern of sound correspondence between European -eu/-ue and Taiwanese -ui. Examples: Gk. pleúmōn ‘the lungs’ > pleu- > Tw. hùi 肺 ‘the lungs’; Germanic *þeuχam ‘thigh’ > *þeu- > Tw. thúi 腿 ‘thigh’ (Tw. th = aspirate t); L. duo ‘two’ > It. due > Tw. tùi 對 ‘a pair’; Eng. sue ‘follow (archaic)’ (French suivre ‘to follow’) > sue- > Tw. sûi 隨 ‘to follow’ (Sir Thomas Malory’s Le Morte D’Arthur, Book III, Chap. 10, has, “and then he sued fast upon him…”, here sued = followed.) Thus, súi 水, the second element in Tām-chúi-hô 淡水河 also connotes ‘river’.

    Finally, I also see that Tw. hô 河 ‘river’ can be derived from Gk potamós ‘river’ (potamós > po- > Tw. hô 河). There is a common exchange between p and h as evidenced in Tw. pak 北 ‘north’ vs. Japanese hoku. In a previous posting in Language Log, we discussed about Nippon vis-à-vis Nihon for 日本.
    Therefore, we can see that all three elements in Tām-chúi-hô 淡水河 can trace their etymology to Indo-European roots, and surprisingly, the name Tām-chúi-hô 淡水河 can be interpreted as a triple play of ‘river’ – River-River-River.

  23. Daniel Trambaiolo said,

    June 17, 2013 @ 12:18 am

    A further correction: the changes in pronunciation are indeed phonetically regular in accordance with the "initial sound rule" 두음법칙 頭音法則, which I believe is observed in the language of the Republic of Korea but not in the People's Republic of Korea. (I am not sure why I had convinced myself that this sound change was restricted to the surnames.)

    @Coby Lubliner
    Concerning the use of etymological romanization when only foreigners write in Roman characters, this sort of thing can happen if the phonetic changes have taken place fairly recently—the old spellings can persist in foreign languages even after the pronunciations have changed for native speakers. An obvious example is the spelling of the Chinese capital as "Peking," which continued to be used in English well after regular phonetic changes altered the pronunciation of the second syllable to something more like "jing." (The difference between p and b in Peking vs. Beijing is a matter of transcription conventions, but the difference between k and j reflects historical phonetic change.)

    I do not know exactly when the relevant phonetic shifts took place, but since they are mostly restricted to the present-day ROK and not seen in the PRK, I would suspect that they occurred sufficiently recently that spelling in foreign languages has not had time to catch up.

  24. ahkow said,

    June 17, 2013 @ 7:21 am

    @Chau Wu:

    Your arguments seem a little bewildering to me.

    I think most people would parse 淡水河 as [淡水][河] "freshwater river" not [淡][水][河] "bland river river".

    Second, the southern Chinese convention for naming rivers is 江 "jiang" (as in 长江 Yangtze or 珠江 Pearl River) not 河 and 水 which are for the Yellow River and its tributaries.

    Lastly, that's not how the comparative method works…

  25. JMU said,

    June 17, 2013 @ 8:15 am

    Though English sue may indeed be a borrowing from the Norman French reflected in French suivre, the latter is from Vulgar Latin *sequere, (Classical has the deponent sequi) ‘follow’. If it is unlikely that pIE roots appear in Chinese, it is, I should think, even more unlikely that the sound changes that took place in Romance and in early Chinese should happen to have been the same.

  26. KeithB said,

    June 17, 2013 @ 9:06 am

    In Los Angeles, it is common to say "The La Brea Tar Pits" which comes out to "The The Tar Tar Pits".

  27. Daniel Trambaiolo said,

    June 17, 2013 @ 3:30 pm

    Your explanation of the phrase 淡水河 might seem plausible, but Chau Wu's hypothesis gains credibility when we consider the presence of Dutch colonists in Taiwan during the seventeenth century. Since the Dutch were more familiar with Indo-European than with Sinitic languages, in communicating with Chinese settlers they might well have strung together roots from different origins in the attempt to make themselves understood. We could reasonably speculate that Dutch colonists might have gesticulated towards a large river and said something like "dānu-sue-potamos," which was later simplified to Tām-chúi-hô by Chinese settlers who were not themselves familiar with Indo-European languages. The phonetic similarity to the phrase "freshwater river" would then have given rise to the transcription as 淡水河 on the basis of a false etymology. The truth of this hypothesis is far from certain, of course, but it certainly deserves further investigation.

    Incidentally, a similar series of misunderstandings was responsible for the English word "kangaroo." As is well known, this word derives from the sentence "I don't understand what you are saying," spoken in the language of an unidentified Aboriginal tribe in response to a European who asked "What is that animal?". (The similarity of "kangaroo" to "gangurru," the Guugu Yimidhirr word for the species, is probably no more than a coincidence.) However, I have noticed that "kangaroo" sounds remarkably like the Chinese phrase "kan gu rou" 看古肉 ("look at the old meat"), which might have been spoken by Chinese traders along the Northern coast of Australia who were unimpressed with the quality of sea cucumbers being offered to them for purchase. My hypothesis is that the local Aborigines did not understand what these Chinese traders were saying, and the phrase later became used in their own language as a conventional means of expressing lack of understanding. I will leave it to Chau Wu and others to trace the phrase back beyond Chinese to its earlier Proto-Indo-European roots, since this lies beyond the scope of my own expertise.

  28. Chau Wu said,

    June 17, 2013 @ 3:52 pm


    The naming of rivers in Taiwan does not follow the southern Chinese convention using the word 江 jiang. No river in Taiwan uses that word. All use the word 溪 xi, except the two large rivers 淡水河 (Tām-chúi-hô) discussed in my previous comment. Please see the map on the Website of Water Resources Agency of the Ministry of Economic Affairs of Taiwan:


    溪 xi generally connotes ‘creek’ or ‘brook’ (small stream), but in Taiwan, the word is used for rivers. The largest and longest river is 濁水溪 Lô-chúi Khe. Furthermore, the 下淡水河 Ē-Tām-chúi-hô was renamed 高屏溪 Kao-Peng Xi after the Nationalist Chinese Government took over Taiwan. It goes to show the almost-exclusive use of 溪 for river names in Taiwan.

    I agree with you that the normal way of parsing Tām-chúi-hô would be like [淡水][河], reading like ‘Freshwater river’. That was how I parsed it before I started to ponder about the unusual name of Tām-chúi-hô. In fact, the largest river in Taiwan, 濁水溪 Lô-chúi khe, is parsed in this fashion, [濁水][溪], indicating that its water is quite muddy (because of the fast flow of water). However, the river was given that name, not by early settlers, but (much later) by the Japanese Government during the Japanese era, and, it is safe to say, the Japanese officials were implicitly following the conventional parsing when naming it such.

    On the map on the Website above, you will find a river named 鹽水溪 Iâm-chúi Khe in southern Taiwan. We would take it that the river water is a bit salty if we follow the normal way of parsing. However, the parsing does not work in this case. The river used to be called 新港溪 Sin-káng Khe, a downstream section of which is called 鹽水溪 Iâm-chúi Khe because it passes by a village called 鹽水 Kiâm-chúi. The Nationalist Chinese Government renamed the whole river based on this downstream section.

    With this explanation, I hope you will understand why the unusual name Tām-chúi-hô got me thinking beyond the conventional southern Chinese name 江 for rivers and beyond the normal way of parsing.

  29. Daniel Trambaiolo said,

    June 17, 2013 @ 5:14 pm

    @Chau Wu
    Setting aside my (obviously facetious) remarks above, are you able to comment on the relationship between Danshui 淡水 as a name for a region in northern Taiwan and Danshuihe 淡水河 as the name of a river passing through this region?

    Are you claiming that Danshui was the original name of the river, and that "he" 河 was later added because local Chinese speakers did not interpret "shui" as meaning "river"? And that "Danshui" subsequently came to refer to the region rather than to the river itself? Do you believe that the difference between 鹽水 and 鹽水溪 resulted from a similar process? If this is what you are claiming, then when do you believe these changes took place?

    These sorts of questions should be readily answerable using Qing gazetteers, but I don't have easy access to these at the moment.

  30. Jason Stokes said,

    June 17, 2013 @ 11:33 pm

    @Daniel Trambaiolo

    Incidentally, a similar series of misunderstandings was responsible for the English word "kangaroo." As is well known, this word derives from the sentence "I don't understand what you are saying," spoken in the language of an unidentified Aboriginal tribe in response to a European who asked "What is that animal"

    I'm afraid it's a bad idea to repeat the Kangaroo etymology story on a language blog! There is not the slightest evidence for this story. On the other hand, the Guugu Yimidhirr etymology is well documented. Captain James Cook recorded the word as "Kangaroo" from the Guugu Yimidhirr in his journal when beached in the Cooktown region for repairs in 1770 and from there it entered English. "Kangaroo" (or "Ganguru") actually refers to only one species of kangaroo, and is therefore an overgeneralisation like Neo-Melanesian "Pisin" (Tok Pisin dialect)/Pijin (Bislama dialect), from English Pigeon, which in Neomelanesian is an overarching category for all birds.

  31. Daniel Trambaiolo said,

    June 18, 2013 @ 3:22 am

    @Jason Stokes
    I certainly hope that nobody will choose to cite me as an authority for the Kangaroo etymology story, although it would be interesting to see whether the internet picks up on my proposed Chinese etymology. Perhaps some enterprising Chinese news organizations will be willing to go beyond their usual sources of news reports (e.g. The Onion) and
    make an official announcement of this groundbreaking discovery.

  32. Jason said,

    June 18, 2013 @ 6:14 am


    Sorry I was skimming today and my irony meter failed. Understood.

  33. Daniel Trambaiolo said,

    June 18, 2013 @ 8:20 am

    No apology necessary. I knew about the true etymology of the word in general terms, but was unaware of some of the details mentioned in your comment and am happy that an irony meter's failure gave me the opportunity to learn them.

  34. Chau Wu said,

    June 18, 2013 @ 9:07 pm

    @Daniel Trambaiolo

    Let me address your questions in the second paragraph first. The 鹽水溪 Iâm-chúi Khe, the downstream segment of 新港溪 Hsin-káng Khe, used to be called 許縣溪 Kho·-koān khe. It was renamed 鹽水溪 because it passes by the town of 鹽水 Kiâm-chúi. The Nationalist Chinese Government then applied this name to the whole river. All these were done by whimsical government officials; the naming was so arbitrary that we need not attach too much meaning to it.

    Now to answer your first question, my understanding is that the river was named 淡水河 first, and then the town’s name 淡水 followed. I have seen an ancient map of Taiwan, which shows the southern Tamsui River (in southern Taiwan). The river is labeled 淡水河. At the time, I was concentrating on that river, not paying attention to the northern one. My guess is that it probably bears the same label. I tried to locate the map or a digital image of it on the Internet, but to no avail. Based on the labeling of the southern river, it is safe to say the northern one would bear the same name.

    What is now the tourist and university town of 淡水 Tamsui was originally a settlement of a native Austronesian Formosan tribe. The early immigrants from Fujian learned the place-name as 滬尾 Hó·-bé from the native tribe. Later as the port town developed next to 滬尾, the name 淡水 Tamsui appeared, most likely based on the name of the river. (Up to my parents’ generation, the two names of the town, 滬尾 and 淡水, were used interchangeably.)

    There is an historical precedent for this. In the 3rd century B.C., Greek colonists from the Bosporan Kingdom founded a colony at the mouth of Don River (Tanais River in Classical Greek). The Greeks named the town Tanais after the river (reminding us of the PIE *danu-). Please see the map at the Website below:


    The following are links to two ancient maps available on the Internet, both of which have been digitized and rendered interactive – you can move the map around and zoom in or out. The first one was by a Dutchman in ca. 1652. The second was commissioned by Emperor Kang-xi around 1700. In both maps the town 淡水 is shown, but the river is, unfortunately, not labeled.



    Late in Ching Dynasty, the name 淡水 was extended to the entire drainage region of the river, and an office, 淡水廳 (Prefecture of Tamsui), was established later as an administrative office to cover the entire northern Taiwan. After Japan took over, Taipei was set up as the capital and 淡水 retreated to where it began. So at present, 淡水 refers only to the town at the mouth of the river.

  35. Chau Wu said,

    June 18, 2013 @ 9:24 pm

    @Daniel Trambaiolo

    Somehow the link to the map of Greek colonies of North Black Sea does not automatically get to the site. Please copy and paste the Webpage to the search bar. The two subsequent links work fine.

  36. KWillets said,

    June 20, 2013 @ 4:31 pm

    Kim-Renaud actually gives two reasons for the Roh/Noh/Li/Yi etc. confusion: differences in North/South dialect handling of the initial , and recent changes in the ROK towards writing family names as "Hangulizations" of the original Chinese, counter to the law of initials. In the second case 이 is written as 리 but pronounced as the former.

    The law of initials is that initial ㄹ [l/r] strengthens to ㄴ [n] which is dropped for ni or ny, so li -> ni -> i, or lo -> no.

    Northern dialect does not change the spelling to reflect the phonetic changes, but Southern does, so we see things like "Rodong Missile" which is pronounced Nodong and written phonetically in the South.

    To add to the fun, recent borrowings don't follow the rule, eg "리뷰" (ri-byu, review).

  37. KWillets said,

    June 20, 2013 @ 5:37 pm

    Correction: the NK pronunciation is different in tandem with the spelling. "Ro" is pronounced and spelled that way there.

  38. JK said,

    June 25, 2013 @ 3:15 am

    I am late to your discussion of Korean, animal names and spurious etymologies, but I happen to be an expert on these subjects and would like to share some interesting facts:

    If you know even a tiny bit of Korean, you will have noticed the gaping semantic gap between the Korean noun kolay¹ “whale” and its derivative verb kolu-² “choose; select” – you must be thinking, “What were they thinking?”
    To understand this peculiarity, some people may find it helpful to know that kolu- is in fact a calque formed analagous to the German umlaut verb wählen “(s)elect” < Wal “whale”:
    Pre-migratory period Germanic peoples turned to a whale oracle for important choices including the election of their leaders; when whale populations dropped, their role was replaced by nuts, which were used in much the same way.³ The tradition has been lost but lives on in a few NHG words such as Wahl­beobachtung “election monitoring” (literally “whale watching”) and Walnuss/​WalnußJuglans regia (a species of nut)”.

    Illustrating that borrowing is not a one-way street, we can now convincingly retrace the German noun Pinguin “any bird in the subfamily Spheniscinae” of previously unknown etymology to its Korean form pingkwuin⁴ “sled” + “-person”, presumably coined in recognition of the birds’ aptitude for sliding on their bellies.
    Sea level rise and continental drift have since increased the distance separating populations of Spheniscinae from the area in which Korean is spoken, to the point where speakers no longer felt the need to have a word for these magnificent creatures, so that the word pingkwuin has all but disappeared from the language. We expect the German word Pinguin to undergo a similar decline for the same reason.⁵

    A more complete treatment of these phenomena and what they suggest about the origins of Burushaski is soon to appear in a peer-reviewed publication.

    ¹ 고래; MR: korae; KMOCT-TS: gorae; KMOCT-TL: golae
    ² 고르-; MR: korŭ-; KMOCT: goreu-/goleu-
    ³ http://japanese.joins.com/upload/images/2006/01/20060124203748-1.jpg
    ⁴ 빙구인; MR: pingguin; KMOCT: bingguin/binggu-in
    ⁵ Inuit languages lexically distinguish by properties such as hue, age, speed, density or the shape of individual's flukes. The number of Spheniscinae-related terms and expressions in use among speakers correlates with proximity to the Pole; the amount of such vocabulary in Inuit languages is exceeded only by that in the Polish language itself.

  39. Paul Zhou said,

    June 25, 2013 @ 5:44 am

    I read the Chau Wu’s paper with interest. But according to the view and the method of Comparative Historical Linguistics, (Tw)‘淡水河’Tām-chúi-ho could not be correspondent with (IE)‘river-river-river’. The reasons are following: (Only the word ‘淡’ )
    [1] The Old meanings are not correspondent.《说文解字–水部》“淡,薄味也。”In Old Chinese, ‘淡’ means ‘blank taste’, does not ’river’.
    [2] The Old Sounds are not correspondent. OC.淡 *dams, ’blank taste’;IE danu-, ‘river’.
    [3] The times of the two words are not correspondent. ‘淡水河’ was named in Taiwan, as what the author said, in about 17th century. But the Indo-European roots ‘*-’ were existed in about 2000BC. In 17th century, ’淡水河’Tām-chúi-hô means modern’s meanning in Taiwan Language or Mi-Nan Language闽南语, it is opposite to ‘salt water’, the river offered the migrants the fresh river in the necessities of life.
    In a word, if we want to compare the words in different languages in order to find if they are cognate, we should use the old forms, but not modern forms.

RSS feed for comments on this post