Hokkien at UCLA

« previous post | next post »

Article in Taiwan News:

"UCLA students learn about Taiwanese Hokkien in MOE*-supported course:

Course examines Taiwan’s widely-spoken dialect ‘in different forms of cultural production’", By Stephanie Chiang (4/19/22)

*Ministry of Education

UCLA began offering its first Taiwanese Hokkien course in January 2020:

The description of the course entitled “Taiwanese Language and Culture” reads, “Taiyu, or Taiwanese (also known as Minnan, Hoklo, or Hokkien, depending on context or region), is the language that most Taiwanese people use in daily lives, including everyday interaction and communication, entertainment, social and cultural events, etc.” The four-unit course offered to upper-division students requires students to have taken at least a year of Chinese courses or a Chinese placement test showing equivalent knowledge.

I wish they didn't have the prerequisite mentioned in the last sentence and don't understand the reason for such a requirement.

Note from AntC:

Good to see Taiwanese / Taiyu / Hoklo getting recognised in UCLA's course.
But hmm? Most Taiwanese use Mandarin most of the time, is what I've observed. Older people use Hoklo, and listeners of any age will generally 'get the gist'. (Or claim to — is that an 'Illusion of Understanding', per myl's post?)

When I lived in Taiwan from 1970-72, Taiwanese was still widely and actively spoken, but when I've gone back to Taiwan during the last two decades, I don't hear much Taiwanese being spoken, especially by people under forty or so.  If the Taiwanese people don't pay attention to the dwindling numbers of Hoklo speakers, they will soon lose their Mother Tongue.


Selected readings


  1. Michael Watts said,

    April 20, 2022 @ 11:38 pm

    I wish they didn't have the prerequisite mentioned in the last sentence and don't understand the reason for such a requirement.

    The reason seems straightforward enough: teaching a language to someone who is already familiar with a closely related language is a very different project from teaching the same language to someone who is totally unfamiliar with it. That's why my sister took a class named "Portuguese for Spanish Speakers" instead of "Portuguese for People Who Have Never Encountered a Foreign Language in Their Life". The content is different. The approach is different.

    It would appear that UCLA doesn't anticipate enough demand for Hokkien to run two courses, and they expect most of the demand that does exist to be people who will have knowledge of Mandarin. So they're running one course that caters to the majority of the audience.

  2. Victor Mair said,

    April 21, 2022 @ 12:10 am

    Quite the contrary.

    My experience at Penn and elsewhere is that a Mandarin requirement for people who want to learn Taiwanese or Cantonese or Shanghainese dissuades them from enrolling in such courses because it privileges Mandarin speakers, when the majority of individuals I know who want to study these other Sinitic languages specifically do not want to learn Mandarin. In order to take Taiwanese, Cantonese, Shanghainese, such students feel that they are being forced to take 2 languages (Taiwanese, Cantonese, Shanghainese, etc. AND Mandarin), when what they really want is just T, C, or S).

    The logic is similar to forcing students who want to learn Literary Sinitic / Classical Chinese to first become proficient in Mandarin (we've discussed this often on Language Log), or forcing people who want to learn Latin to first learn Italian, Sanskrit to first learn Hindi, etc.

  3. Jenny Chu said,

    April 21, 2022 @ 12:26 am

    I bet it's nothing more than a matter of convenience for the instructor. In my first Czech class, I recall there were two distinct groups: those who had studied Russian (or, in one case, Polish) and those who had not. The lecturer kept having to decide how much time to spend explaining things like perfective prefixes, or case endings, knowing that the Russian speakers would instantly get it while the others (e.g. those who had studied French or Spanish as their only foreign language) needed a lot more time to get their heads around such concepts.

  4. CrawdadTom said,

    April 21, 2022 @ 2:56 am

    I've been in Tainan for three weeks, and most people I hear talking, of all ages, are speaking Taiwanese. They may switch to Mandarin to talk to me, but they seem to be going about their business primarily in Taiwanese.

  5. Victor Mair said,

    April 21, 2022 @ 5:45 am


    Yes, Tainan and elsewhere in the south are a special case. Try Taipei if you get a chance to get up there.

  6. A'ióng said,

    April 21, 2022 @ 6:20 am

    Absolutely agree with you on that one, @Victor Mair.

    In my own experience, Mandarin was a actually a severe handicap in learning Taiwanese, as it made it all too easy to fall back on what learners would consider to be “obvious” cognates and structures, but which turn out in the end to be entirely different animals.

    Not to mention that, undoubtedly, the course focuses on Kanji whereas I am a very firm believer that Taiwanese should be taught with full romanization until some level of fluency is attained.

    I have seen similar patterns in other learners, and my own students (who I always encourage to approach Taiwanese without Mandarin or Kanji) seem to do much better, in a much shorter period of time.

  7. Victor Mair said,

    April 21, 2022 @ 6:30 am

    Bless you, A'ióng!

  8. Jonathan Smith said,

    April 21, 2022 @ 7:42 am

    This is a very cool program, though I am not a fan of this prereq either…

    The perilous position of Taiwanese seems to be well-understood by all who are into language / consider the problem seriously — of course there is also the naive view that all is well because there are TV shows in it / older folks use it / some listening ability + clunky speech = fluency, etc.

    There are of course north-south differences, but lots of studies / news reports on the issue from Taiwan over recent years confirm that the downward trend is by no means confined to the north.

    Re: whether knowledge of Mandarin or some other Chinese language is an aid or an obstacle to learning a new one: if one produces Language B by direct reference to Language A, relying on analogies which turn out to be far from perfect, there are of course problems; nonetheless, massive structural parallels across Sintic mean that knowledge of one language of course gives the learner a tremendous boost in terms of general intuitions, as Michael Watts / Jenny Chu suggest.

    And teaching / learning Taiwanese via Chinese characters is a downright crime… such an approach greatly encourages the reliance on the flawed analogies noted above; i.e., it masks rather than flags differences.

  9. Victor Mair said,

    April 21, 2022 @ 8:20 am

    From Grace Wu (long term, revered instructor of Taiwanese at UPenn; her father was an important teacher and author of textbooks for Taiwanese during the second half of the 20th century):

    It is very true that leaning Taiwanese in romanization (without Mandarin characters) is much faster.

  10. Kingfisher said,

    April 21, 2022 @ 10:44 am

    Would the same principle apply to teaching/learning Literary Sinitic? Starting out by using modern written pronunciations (of whatever language)?

  11. Coby Lubliner said,

    April 21, 2022 @ 10:50 am

    While the language situation in Taiwan is not quite the same as in Catalonia (where, for example, Catalan is the primary school language), there remains the common feature that, just as in Taiwan nearly all Hokkien-speakers also know Mandarin, so do nearly all Catalan-speakers know Spanish.
    It was my experience while living in Catalonia that outsiders who learned Catalan without knowing Spanish were at a distinct social disadvantage. For one thing, they were cut off from the large Spanish-speaking community. But even primary Catalan-speakers, especially of the older generation, often use Spanish words in place of the prescribed Catalan ones, and a conversation among Catalan-speakers might suddenly veer into Spanish.
    I would also surmise that a Hokkien-only speaker in Taiwan would be cut off from most public media.

  12. Neil Kubler said,

    April 21, 2022 @ 1:57 pm

    At Williams we offer courses in Basic Taiwanese and Basic Cantonese in alternate years and regularly achieve enrollments in the 6-12 range. For either, the prerequisite is 2 years of Mandarin or else permission of the instructor. 90% of the course is on the oral skills, but we do spend a bit of time on reading. The objective fact is that if one has basic (or even better intermediate or advanced) proficiency in one Sinitic language, then learning another goes much faster (in 1 semester of Basic Cantonese, we reach the level students attain after 3 semesters of Mandarin; and in 1 semester of Basic Taiwanese, we reach the level students attain after 2 semesters of Mandarin — which makes sense since Min split off from the common ancestor of the Sinitic languages many centuries before Yue and Mandarin did). If there are students who haven't learned any Mandarin but are strongly motivated to take Taiwanese or Cantonese, I usually will accept them but tell them they will have to spend twice as long preparing as the other students. This is not at all about the instructor's convenience but rather about efficient pedagogy, i.e., helping the majority of students learn better faster. Where I used to work (FSI), I would sometimes organize M>C and M>T "conversion courses" and colleagues in other language programs similarly from time to time offered Spanish>Portuguese and Russian>Ukrainian conversion programs.

  13. DS Zhang said,

    April 21, 2022 @ 5:04 pm

    As for learning any language the script of which is not Roman or alphabetic, in romanization, I have two things to mention. They may be much relevant, or less relevant, or irrelevant to the Taiwanese-learning case per described in this post. Yet I nevertheless would like to bring them up and welcome discussion on to what extent can these two examples be applied to the problems in the post.

    1. I started learning Sanskrit nine years ago, the second year after I came to the States from China (my home country). I have to say that one factor almost ruined my Sanskrit learning experience that over the years after which I must re-take elementary Sanskrit again and again and repeat this "unlearning" process in order to pave flat the abyss of bad habits that my first-year Sanskrit has dragged me into. That was: to have learned Sanskrit in Romanization!!! My teacher told me that language was mostly to communicate, and that even for a classical language like Sanskrit, one could always first quickly transcribe a text into its romanized form before reading, not to mention that some texts were already transcribed (and I believe that there are websites which can automatically transcribe for you nowadays). Therefore, at the very beginning stage of my Sanskrit learning when I had to struggle with both acquiring a new script (Devanagari) and a new language and when my memorization and learning process were accordingly slow, my teacher prevented me from feeling at home with the Nagari script and almost forced me to complete the homework in romanization. I did. This is perhaps one of the things that I regret the most throughout my life. Things were indeed much quicker after I gave up on the script. I did learn much vocabulary and could read chunks of romanized text in "Sanskrit". However, over the years, as I wanted to proceed learning Sanskrit and going deeper into it — especially now that Sanskrit is my primary research language — I found it EXTREMELY HARD to read real Sanskrit rapidly. For volumes and volumes of primary sources, there's no way that I spend a couple of hours romanizing all of them before I start reading or quoting; when I want to read commentaries, real literature, or even simply compare two texts or parse the Sandhi — it is so so so so much more difficult for me to break down a large compound in Devanagari and tell where to stop for each element within it, had the text not been in Romanization. We learn a language in order to use it within its real context. Some "shortcut" may be quicker for the learner in the first place, yet the cost of not having strived oneself to be acquainted with the real writing system makes every step more difficult as one attempts to explore in depth. Had there been time machine, I would have never ever listened to that teacher and learned Sanskrit in Romanization. I would have paid effort to overcome the first stage of facing a distinctive writing system from the Roman alphabet, and train my brain to find a way out with new mnemonic skills.

    For the past nine years as I kept learning more languages — classical and modern — again and again I prove the above problems true. I began learning Urdu two years ago, but my curricular burden was too heavy to afford time learning Nastaliq, thus I quickly switched to the pattern of "sneakily" using romanization again in order to submit every homework in time, instead of sitting down and conquering Nastaliq. My Urdu quickly declined only weeks after I stopped practicing it. In contrast, in 2021 as I began my elementary Telugu, having learned hardcore lessons from Sanskrit and Urdu, I swore to stick loyally to using Telugu script only — however time-consuming it was — for every word that I learn, every sentence that I memorize, and every essay that I write. Now it's been a year. My Telugu does not only strengthen, but becomes hard to forget even if I put it aside for a couple of months.

    The above story is about the jeopardy of making seemingly "rapid" progress in the short run, without getting oneself acquainted with the new script only via which one might step further in the real-life context of a language.

    2. "What is your name?" should be among the first sentences that every language learner acquires, despite which specific language they are learning. However, in the Chinese context, following an answer of "My Name is…. (if the name is in Chinese)", the next question would always, always be (almost without exception from either Chinese people around me or non-Chinese people who visited China): "怎么写? / 是哪几个字?" ("How do you write it? / Which corresponding characters?)" — Because of the large quantity of homophones in Chinese, it is more of the corresponding character that distinguishes a name and its unique meaning from the other, rather than the sound. For example, 张晓明 (given-name lit. "morning-light") and 章筱铭 (given-name lit. "bamboo-epigraph") are both Zhāng Xiǎomíng, yet they do not only carry different meanings and wishes from their name-granting parents, but also denotes two different lineages / surnames with distinct etymology / ancestral history!

    For my own experience as a native speaker of Mandarin, names are not only memorized by sounds, but also by corresponding characters. I (personally) find it quite difficult if someone only tells me the pronunciation of their names without giving me corresponding characters. This does not only apply to Mandarin names; same for Cantonese, Taiwanese, Japanese, Korean, even Vietnamese names. Vietnam already abandoned the hanzi system, yet when I tried to remember a group of 7 lovely Vietnamese friends as I met them in the college orientation activities, the first week after landing in the States — I asked each of the owner the meaning of his/her name and looked up their corresponding characters. I still have not forgotten any of those 7 names until now. Same for Japanese and Korean names, of historical figures or real-life friends. Those names are stored in my memory warehouse as characters, not sounds, at all.

    I'm not providing justification for anything; just to provide a personal instance of how a Chinese person's minds operates when it comes to the simplest and most widely used social scenario, "My name is". Perhaps it's just me, or perhaps I'm not alone. After all, learning Hokkien is for visiting Taiwan, learning Mandarin is for visiting China, and learning Hindi is for visiting Delhi, etc. It will be the local people and their minds that you'll interact with, especially if efficacy of communication is what you pursue to achieve.

    PS: For names that originally do not belong to the (historically) Sinographic sphere, one has to remember them by sounds because there are no "characters". ;) Yet, why say no to an upāya if there could be one at some point?

  14. V said,

    April 21, 2022 @ 9:55 pm

    Learning Sanskrit in (Bulgarian) transcription, initially, is easier than learning it with Devanagari directly, but that might be because there's an established Sanskrit Devanagari transcription into Bulgarian already. Unlike Mandarin, where it went though Russian, and it's strange.

  15. CrawdadTom said,

    April 21, 2022 @ 10:19 pm

    Yes, I know about Taipei. I've lived there since 1983. Sure, Mandarin predominates, but there are plenty of people speaking Taiwanese.

  16. Victor Mair said,

    April 21, 2022 @ 10:32 pm


    Yes, some old folks, and a few young'uns.

    See the remarks by Jonathan Smith above, also the article by Deborah Beaser (written in 2006) cited in the "Selected readings".

  17. Chas Belov said,

    April 22, 2022 @ 12:20 am

    When I was devouring Taiwanese popular music in the 90s and oughts, I was pleased to discover several Hokkien albums by Wu Bai & China Blue (who also have recorded in Mandarin) as well as Lim Giong and the Dust of Angels soundtrack. All the music I hear from Taiwan nowadays is in Mandarin; that said, my discovery is quite random so there could very well be Hokkien music still being released and I would miss it. There were also a couple Hakka albums released by the Labor Exchange Band.

    And I write my Chinese name:
    白力漢 (traditional)
    白力汉 (simplified)
    (It's a translation, not a transcription, of my "English" name.)

  18. Terpomo said,

    April 23, 2022 @ 3:39 pm

    Apropos of your comment about Mandarin prerequisites for Classical Chinese, Victor, this little satirical piece is relevant (it's very likely been posted here before but may be of interest to those who haven't seen it.)

    [VHM update: See now "On how (not) to learn Latin via French" (4/24/22).]

  19. A'ióng said,

    April 23, 2022 @ 10:41 pm

    The vast majority of Taiwanese literature written in the past 150 years is written in a standardized form of romanization. Some large percentage of words either don't have Sinitic roots, or the Sinitic roots are unknown.

    The most active teachers and writers today continue to use romanization as the primary orthography.

    The best contemporary Taiwanese dictionary is fully in romanization.

    It is simply incorrect to conclude or imagine that the "correct" script for Taiwanese is kanji. Both scripts are used, and in my opinion and experience (shared with the vast majority of literate Taiwanese speakers), they are either on more-or-less equal footing, or romanization is preferred for a number of reasons.

    In fact, I would argue that the only reason kanji are still used today is because the majority of learners come from a native Mandarin background, and these learners often make very general (and incorrect) assumptions about how the Mandarin script might carry over to Taiwanese. It does not, in fact, carry over that well. Writing in kanji often significantly hampers the writer's thought process as s/he finds it difficult or impossible to find clear and unambiguous kanji for many words, even simple words like "yes" ("hèⁿ" in romanization; there is no obvious kanji alternative).

    Thus the written word is modified so that it no longer matches the spoken nearly as well, and sounds stilted and awkward when read aloud as written. You won't find "hèⁿ" written much if at all in kanji texts, and you probably won't even recognize it if you do. This kanji-centric writing easily becomes a kind of “Mandarin-lite”, and is frequently admonished by skilled native speakers who immediately recognize that it has many grammatical and word-choice issues.

    All of those problems are resolved by switching to romanization, which, again in my opinion, is the main reason romanization has been the preferred orthography since it arrived to Formosa in the mid-late 1800s.

  20. Jonathan Smith said,

    April 25, 2022 @ 8:39 am

    Interesting comments; some random thoughts…

    Agree re A-iong's point about "the" Taiwanese script (among others)

    re: DS Zhang's points, partly this depends on one's focus; if simply texts, than no two ways about it you must confront those texts on their own terms; if simply spoken language, text employed (when employed at all) will relate to capacity to capture features of spoken language… in reality we of course all want to do some of both of these.

    agree with DS Zhang re: the place of the written character in remembering Chinese words including names — looked at another way, this is exactly the problem: when learning new languages which are / can be / have been written with Chinese characters, looking at characters builds counterproductive and often downright erroneous neural connections.

    Some Taiwanese speakers take great pleasure in writing Taiwanese-language essays / journals in Hanzi — excellent, but when these fall into the hands of children/grandchildren, how much of the meat of Taiwanese will they transmit? Not "none," but…

    So e.g. Fangyan is great but in comparative terms frankly horrible at telling us anything substantive about speech varieties of Han…

    Re: names, if as a Mandarin speaker one knows the Chinese characters for a Vietnamese/Taiwanese/Korean… person's name, does one *really* know their name? One might know say "morning" + "bright" but can't *say* the name to members of that speech community… depending on the language involved, one often can't even, say, find information on that person on the internets, as I know from experience :D… bilingual Mandarin/Taiwanese speakers seem often to code-switch into Mandarin for prominent personage's (and others') names when speaking Taiwanese rather than "reading" the characters in Taiwanese… indeed given proper nouns/terms widely known in Mandarin, "reading" characters in Taiwanese in this way seems to strike speakers as a downright laughable enterprise… interesting that "reading out" characters for say Japanese names is so normalized in the Mandarin[speaking community…

  21. Terpomo said,

    April 25, 2022 @ 12:41 pm

    A'ióng, even if there's no single standard for writing Taiwanese in Sinograms, isn't it manageable enough as long as you have some standard or another, that has characters for all the words? Doesn't the MOE have one they promote? Or is it missing characters for some words?

  22. A'ióng said,

    April 26, 2022 @ 1:02 am

    @Terpomo As we all know, presumably, any language could be written with any set of arbitrary symbols, well defined and agreed upon.

    The issue is not that it *can't* be done, but rather that (A) there is no well defined set, and (B) there is no agreed upon set. The efforts by the ROC government have proposed around 700 characters for specific words. Taiwanese has many tens of thousands of words, and a huge number are left unaddressed in the recommendations. But perhaps even more importantly, there is a very large undercurrent of resistance to using even the 700 recommended characters, since many of them do not find roots in Taiwanese but are rather ad hoc and inappropriately borrowed from other languages (e.g. Mandarin, or Shanghainese among others) where perfectly good options with "Taiwanese roots" are available.

    Some examples for the latter point being:

    ROC 著 vs. Taiwanese 着, with 著 being traditionally reserved for the verb "to author", with a different pronunciation.

    ROC 囡 vs. Taiwanese 囝, where 囡 is borrowed from Shanghainese with no real argument given as to why that should be done when 囝 is already widely used and recognized.

    ROC 的 vs. Taiwanese 个, where 的 is traditionally reserved for the word TEK.

    The list goes on. And as long as it is, it (and both lists, in fact), are severely lacking. Just for one example, the extremely common word for YES (HÈⁿ) has no common character, in either script.

    All of the above issues are fully resolved by using romanization, which has been more or less standardized through widespread use for more than a century at least.

  23. Jonathan Smith said,

    April 26, 2022 @ 7:38 am

    There are a number of different perspectives to consider re: Taiwanese…

    For a native user, ANY (more-or-less reasonable) set of conventions will work — so the outsider winds up frustrated about debates over Romanization systems (anything works; just choose!) and Hanji conventions (same!).

    So the idea with "囡" vs. "囝" is to distinguish the words gín(-á) 'child i.e. little kid' and kiánn 'child i.e. son/daughter' in writing — which seems pretty reasonable, but again lots of things are reasonable enough to work.

    For learners, though, or for those interested in preserving Taiwanese into the future, Chinese characters in any of the ways they're currently employed are a poor choice, full stop.

  24. Terpomo said,

    April 26, 2022 @ 3:51 pm

    What would your opinion be on simply making up characters ad-hoc for words the author doesn't know any character for?

  25. Jonathan Smith said,

    April 27, 2022 @ 3:10 pm

    @Terpormo me? :D if so, I guess I would say… the desideratum seems to be orthographic conventions that unite a Taiwanese-using community across (1) space and (2) time. A set of conventions achieving (1) could in theory incorporate tools from any source (old or [originally] ad hoc characters in traditional or novel applications, phonetic writing of whatever kind, etc.); it seems now that the MOE proposals are best positioned to gather steam, but who knows… as for (2) though, given a pseudo-unified Chinese cultural sphere, logographic Hanzi tend naturally to squash dialect diversity over time, most obviously but not exclusively with respect to phonology… it feels like a serious attempt here should turn to Romanization or some parallel.

  26. Jonathan Smith said,

    April 27, 2022 @ 3:10 pm

    sorry, sp., Terpomo

RSS feed for comments on this post