The many varieties of Japanese regional speech

« previous post | next post »

Anyone who learns Standard Japanese and then travels around outside of the Tokyo area will quickly come to realize how distinctive and numerous are the local forms of language once one leaves the metropolitan region of the capital.

Some interesting aspects of this phenomenon are presented in a new article in, "Linguistic Treasures: The Value of Dialects", by Kobayashi Takashi, professor at the Center for the Study of Dialectology, Tōhoku University, who specializes in dialects and the history of Japanese.

Kobayashi begins:

Japan has a great number of dialects. One scholar divides the archipelago into 24 areas with distinct regional linguistic forms. Yet, this is a very broad classification, and if one pays attention to variations in grammar and specific words, it is no exaggeration to say that there is a dialect for every city, town, village, and hamlet.

These were once looked down on for their association with uncultured, provincial speakers. In recent years, however, dialects have been increasingly appreciated for the pleasure they can bring to verbal interaction and their ability to draw people from a particular area closer together. This has led increasingly to their use in the names of local products, and their incorporation into plays and TV dramas.

Notably, the 2013 NHK morning drama Ama-chan (Little Diver) brought the expression of surprise jejeje to national attention, which was one element in its success. The phrase je comes from just one part of Kuji in Iwate Prefecture, although ja is in use over a much wider area centered on Iwate. Other unique ways to voice one’s amazement that are found in dialects but not standard Japanese include waiha, sāsa, ūu, and chopped off forms of da and ba. The wide range seen even in this category of utterance demonstrates how rich in dialects the country is.

The author then embarks on a discussion concerning the origins of the dialects, and he finds their "Roots in the Old Capitals":

The main source of dialect forms comes through transmission of language from the center to the regions. New words in former capitals like Nara and Kyoto gradually spread through neighboring areas to the wider nation. These successive waves created regional linguistic differences, or dialects.

This means that many dialect words derive from standard terms in more ancient forms of Japanese. For example, the word menkoi in Tōhoku, equivalent to kawaii (cute, adorable), comes from megushi, a word seen in the Man’yōshū poetry collection from the Nara period (710–94). Churasan, an Okinawan word for “beautiful,” is related to kiyora nari, seen in works from the Heian period (794–1185) like The Tale of Genji.

As well as words in literary texts, historical terms that have vanished completely from the mainstream live on in contemporary dialects. They are like time capsules we can use to encounter the Japanese of centuries ago. For their role in preserving traditional culture, it is not overstating it to say that dialects are a national treasure that we must protect.

Another aspect of dialect usage examined by the author are the social and emotional bonds they evoke:

I work at Tōhoku University’s Center for the Study of Dialectology, which has researched threatened dialects in the area devastated by the Great East Japan Earthquake and tsunami of March 2011, and worked to support their revival. At a meeting we held with residents in the first year of its operation, a woman from Miyagi Prefecture rose to offer an opinion.

She described how seeing the slogan of encouragement Ganbarō Miyagi (roughly “Keep at it, Miyagi”) in standard Japanese would sometimes make her angry, having lost her family and many friends. “What more can I do?” she would think to herself. But when the phrase was in dialect like Ganbappe Miyagi or Magenē zo Miyagi (Don’t give up, Miyagi), she would find herself agreeing. “Dialect words really are filled with the spirit of the people who live in an area,” she said. “They have a soul in them. And I think that’s how they have the power to support victims of the disaster.”

Finally, Kobayashi looks into threatened dialects, notably those of the Tōhoku (Northeast) region:

The 2011 disaster destroyed many communities. Dialects are shared in the lives of community members, and they are lost with the loss of those people and the weakening of local ties. When there is no place to speak a dialect with neighbors or relatives, it becomes buried under waves of standardization. The calamity that hit Tōhoku spurred on an already ongoing decline.

Yet there are strong ongoing efforts to preserve dialect in tsunami-hit areas, such as play performances and retellings of legends, as well as teaching of dialect in schools. The Center for the Study of Dialectology is working to record conversations showing how dialect is used in everyday life in situations like asking favors and accepting or refusing, as well as expressing gratitude, apologies, joy, sorrow, or surprise. These common exchanges are where the beating heart of dialect can be found.

There are thousands of local and regional variants of spoken language in China, and — following Chinese custom — I refer to them as fāngyán 方言 ("topolects"), upon which innumerable Language Log posts touch and which has entries in Wiktionary, American Heritage Dictionary of the English Language, and elsewhere.

The term hōgen 方言 ("dialect") is also employed in Japanese, but local varieties of speech are more distinctively referred to as ben 弁 — a simplified Japanese character that collapses three traditional characters, for which see this Wiktionary article, the one in question being 辯.  In Modern Standard Mandarin, the latter is pronounced biàn and means ("debate; argue; discuss").


Selected readings

"Dialectology of Japanese reflexive exclamations" (1/8/17)

"An Eighteenth-Century Japanese Language Reformer" (4/23/15)

"Intelligibility and the language / dialect problem" (10/11/14)

"Japanese dialects" (Wikipedia)

[h.t. June Teufel Dreyer, Don Keyser]


  1. Antonio L. Banderas said,

    March 24, 2020 @ 4:21 pm

    I'd love to know if spontaneous romanizations arise in local 弁 which don't fit the standard syllabary.

  2. Chris Button said,

    March 24, 2020 @ 7:25 pm

    But when the phrase was in dialect like Ganbappe Miyagi or Magenē zo Miyagi (Don't give up, Miyagi), she would find herself agreeing.

    I lived in Iwate prefecture for almost a year in my late teens. I remember that "ē" sound for "ai" well–"magenē" here being standard 負けない "makenai", although I don't recall that the "k" would have become a "g" where I was staying in Ōfunato. The -ppe in "ganbappe" for "ganbarō" is presumably of the same origin as the "-be" sound in common phrases like "ikube" for 行こう "ikō".

  3. Chris Button said,

    March 24, 2020 @ 9:53 pm

    I remember that "ē" sound for "ai" well

    I recall it also being used for "-oi" too. So a word like "omoshiroi" 面白い would be "omoshirē"

  4. Peter Grubtal said,

    March 25, 2020 @ 2:15 am

    Variation with location seems to be inherent in language, the most natural thing in the world.
    Rather than a theory of "origins of dialect", more to the point is "origin and imposition of uniform national languages".
    No doubt there are many studies on this. It's a process which could be followed fairly closely in European states such as France, Italy and Germany. In the 19th C., dialects were still rampant in these countries (probably in the spoken langugage, by population, predominant), and have still not been completely eliminated.

  5. KeithB said,

    March 25, 2020 @ 8:58 am

    What a coincidence, isn't that the "Covid-19" hashtag dash?

  6. Michael Watts said,

    March 25, 2020 @ 4:43 pm

    Other unique ways to voice one's amazement that are found in dialects but not standard Japanese include waiha, sāsa, ūu, and chopped off forms of da and ba.

    This gives me some misgivings about the idea of marking "long vowels" with macrons in Japanese. Do long vowels exist? What's the difference between ūu and uū?

  7. Philip Taylor said,

    March 25, 2020 @ 5:06 pm

    I have very little familiarity with Japanese, Michael, but I have for a long time believed (rightly or wrongly) that spoken Japanese differentiates between /ɒ/ (o) and /əʊ/ (ō). If that were true, would it not confirm that long vowels do indeed exist ?

  8. Michael Watts said,

    March 25, 2020 @ 6:15 pm

    If that were true, would it not confirm that long vowels do indeed exist ?

    Well, it wouldn't demonstrate that two vowels can be phonemically different while varying only in terms of their length — those are different vowels, not shorter and longer expressions of the same vowel.

    In the pronunciation scheme used by American dictionaries and taught in American schools, the vowels of "hat" and "hate" are called "short a" and "long a", and the dictionaries indicate them as "a" and "ā". Does that confirm that American English contains long vowels? Of course not. TRAP and FACE are two different vowels — two different, unrelated, vowels, much like DRESS ("short e") and GOAT ("long o").

  9. Michael Watts said,

    March 25, 2020 @ 6:20 pm

    I should note that you appear to have given British English phonemic transcriptions for the English LOT and GOAT vowels, not transcriptions for Japanese /o/ and /ō/. It would be difficult for English vowels to establish much of anything about Japanese.

  10. Philip Taylor said,

    March 25, 2020 @ 6:30 pm

    Thank you for your answers, Michael. I do not seek to argue with your analysis, and would only observe in passing that my choice of British English phonemic transcriptions was entirely the result of my almost total lack of familiarity with Japanese, to which I confessed at the very beginning of my question to you. Had I even tried to guess at the correct Japanese phonemic transcriptions, I would have been way out of my depth.

  11. Chris Button said,

    March 25, 2020 @ 6:41 pm

    @ Michael Watts

    This gives me some misgivings about the idea of marking "long vowels" with macrons in Japanese. Do long vowels exist? What's the difference between ūu and uū?

    With that question, you've just answered what I was trying to get an answer to over on this thread:

    The Japanese version of the article writes "ūu" as ウーウ in katakana. The ー clearly identifies the long vowel with the first ウ and justifies the existence of ー as a necessary part of the script versus just writing the vowel sound again.

    As for the actual pronunciation, prosody should take care of the distinction between the ウー and the ウ.

  12. Michael Watts said,

    March 26, 2020 @ 8:46 am

    I agree that the Japanese orthography described makes it explicit that there's a long ū followed by a short u. I don't see why we should trust that orthography more than we trust English orthography when it tells us that "peddle" and "petal" are distinct. (This is an especially apt example, because at least Merriam-Webster explicitly indicates different pronunciation for those two words, with /d/ or /t/ following the spelling. The Cambridge Dictionary correctly notes that "petal" is flapped in American English, but does not so note for "peddle".)

    Is there, empirically, a pronunciation distinction between ūu where there is a word boundary between ū/u, and uū where there is a word boundary between u/ū? Is there a distinction between ō as written with a length marker and oo as written with two full お characters? (I was under the impression the answer to this second question was "no".)

  13. Michael Watts said,

    March 26, 2020 @ 2:08 pm

    (Or for a strictly within-Japanese example, consider the orthographic distinction between oo and ou, which definitely doesn't exist in reality.)

  14. Chris Button said,

    March 26, 2020 @ 3:31 pm

    @ Michael Watts

    It comes down to prosody. In something a short as ウーウ I'd imagine it is nothing more than a pitch distinction.

    As for "ō" versus "ou", the latter is just a way of more strictly representing spelling in kana. So something like がっこう "school" is gakkō if you want to reflect the fact that the "ō" is a long vowel, or gakkou if you want to reflect the kana spelling of こ "ko" followed by う "u".

  15. Jonathan Smith said,

    March 26, 2020 @ 11:29 pm

    To the extent this discussion is about whether there are "real" long vowels in the world's languages, then clearly yes… if the question is do they exist in Japanese, I suppose the answer is "obviously" since, e.g., to 'with' vs. tо̄ 'east' and countless others are distinguished only by length, but at the same time "maybe not" since you could divide words into mora-bbles and call it all hiatus. I recommend Jim Breen's Jdic recording of say ほうおう 鳳凰 over somewhat mysterious "ウーウ"…

  16. Jonathan Smith said,

    March 26, 2020 @ 11:36 pm

    and no it is not true that orthographic "ou" always = "oo"; I don't know in detail but think that verbs for instance really end in [oɯ̟] or however you care to represent it…

  17. Chris Button said,

    March 27, 2020 @ 7:14 am

    @ Jonathan Smith

    and no it is not true that orthographic "ou" always = "oo"

    おう and おお are both "ō". The former is sometimes spelled as "ou" to reflect the kana spelling

  18. Jonathan Smith said,

    March 27, 2020 @ 10:10 am

    Sorry, I don't know in detail… probably our statements above are not clear enough. I meant that verbs like おもう, おおう, おそう really end in [oɯ̟]. If you are talking about spelling per se, then in the middle case orthographic "おう" is not /ō/. But this could be careful speech or there could other factors I'm totally ignorant of…

  19. Chris Button said,

    March 27, 2020 @ 11:21 am

    Gotcha. Yes you're right in what you're saying. I hadn't quite figured out what you meant by verbs above. When it's not forming part of a long "ou" vowel with "o", then "u" is distinct. So "omou" has a root "omo-" (as in its formal form "omoimasu") and the "u" at then end there does not form a long vowel but is a separate "u" sound.

  20. Chris Button said,

    March 27, 2020 @ 11:26 am

    So to compare the above words, we have we have gak.kou "school" (long vowel) but "to think" (no long vowel)

  21. Jonathan Smith said,

    March 27, 2020 @ 12:01 pm

    Yeah but there seems to be no (or very little) basis for parsing such words differently, so ko.o.ko.o 'high school', 'think'… so I say whether or not "long vowels" proper exist depends on the prosodic analysis

  22. Chris Button said,

    March 27, 2020 @ 1:35 pm

    I think it's because you don't get a long vowel when the "u" plays an inflectional role. That's also the case with two "u" vowels in something like "suu" (to smoke).

    Would be great if someone well versed in Japanese historical phonology could comment here.

  23. Chris Button said,

    March 27, 2020 @ 1:49 pm

    But yes, that in itself probably ultimately comes down to a prosodic analysis.

  24. Chris Button said,

    March 27, 2020 @ 9:44 pm

    Ok here goes:

    There are five long vowel phonemes in Japanese:
    /aː/, /iː/, /uː/, /eː/, /oː/

    Excluding loanwords in katakana where the macron is used across the board (as "ā", "ī", "ū", "ē", "ō"), the modern Hepburn treatments are:

    "ā", "ii", "ū/uu", "ei/ē", "ō"

    "ā" is used for /aː/

    "ii" is used for /iː/. Presumably this is because "ii" often occurs as part of an inflection (e.g. やさしい "yasashii" versus やさしく "yasashiku"). However, this is confusing when things can't inflect (e.g. おじいさん "ojiisan") where "ī" would make a lot more sense and is consistent with the "ū/uu" alternation below. Interestingly, Okada's description in the IPA handbook uses /iː/ for "ojiisan" but /ii/ for "yasashii".

    "ū" is used for /uː/ unless it can be inflected in which case "uu" is used. So 空 (くう) is "kū" but 食う (くう) is "kuu" since it can inflect as 食います "kuimasu".

    "ei" and "ē" are used for /eː/ based on hiragana spellings. So えい is "ei" and ええ is "ē" in spite of both being /eː/. Presumably they were once distinct?

    "ō" is used for /oː/ regardless of whether the hiragana spelling is おう or おお. This is discrepant with the "ei/ē" alternation above. However, "ou" is used when the "u" can be inflected, in which cases the pronunciation is also distinct. So 王 (おう) is "ō" and pronounced /oː/ while 追う (おう) is "ou" and pronounced /ou/

    I've been happily plugging modified Hepburn into the "Derivational Dictionary of Chinese and Japanese Characters" that I'm compiling without really thinking about any of the above. I think at the very least I'm going to add "ī" to use alongside "ii" in a manner consistent with "ū" and "uu".

  25. Jonathan Smith said,

    March 29, 2020 @ 2:10 pm

    Ah for purposes of your dictionary? Just something standard surely rather than tweaking? I thought you might be a JSL fan…

  26. Chris Button said,

    March 29, 2020 @ 9:09 pm

    I am a JSL fan, but modified Hepburn is the one most commonly found nowadays, and I certainly won't be marking pitch accents. I'm using pinyin for the same reason.

    It seems that adding "ī" to use alongside "ii" to parallel "ū" and "uu" is a standard tweak so I might make that one. I think it's probably best to leave everything else alone. That "ei" vs. "ee" one is going to bug me (JSL only has "ee"), but I think it would be unwise to touch it.

  27. Chris Button said,

    March 29, 2020 @ 9:35 pm

    Actually, far harder than the romanization is figuring out which pronunciations to include with the non-Jōyō characters so as to be functional.

    The dictionary itself will include the combined 2,769 Jōyō (常用) and Jinmeiyō (人名用) characters designated by the Japanese Ministry of Education to which I'm adding a further 737 characters to cover the 3,500 most frequently used characters according to the Japanese Agency for Cultural Affairs.

    There are also roughly 3,000 more entries, for which I'm not providing Japanese pronunciations, that form the remainder of the 6,500 level 1 & 2 Tōngyòng Guīfàn (通用规范) characters, designated by the Chinese Ministry of Education, that (almost) completely cover the 4,808 Chángyòng (常用) characters designated by the Taiwanese Ministry of Education.

  28. Chris Button said,

    March 29, 2020 @ 9:38 pm

    Or should that be Jinmēyō :)

  29. Jonathan Smith said,

    March 30, 2020 @ 10:34 am

    Sounds good. Quite a project you've bitten off…

    Re: the standard statement "there are five long vowel phonemes in Japanese," another way of looking at it is that if one really wants to talk about contrastive "long vowels" in Japanese then one also has to accept "diphthongs" of all kinds, which seems pretty silly…

  30. wren romano said,

    March 31, 2020 @ 3:16 am

    As someone with knowledge of historical phonology and orthography of Japanese,…

    In Modern Standard Japanese there is no phonetic distinction between that which is spelled 「おう」 and that which is spelled 「おお」, both are spellings of /oː/ (or /oo/ if you take a moraic stance on how to IPA-ize Japanese). Ditto for 「えい」 and 「ええ」 as alternate spellings of the same /eː/.

    The orthographic distinction is capturing etymological differences which were phonologically neutralized over the ages. For example, 大きい /ookii/ "big" was historically written/pronounced 「おほきい」/oɸokii ~ owokii/. The old 「おほきい」 spelling stuck around long after the pronunciation had changed to /ookii/, but in 1946 there was a major spelling reform which eliminated these non-phonetic spellings. Originally the 「おお」 vs 「おう」 reformed spellings denoted different vowels/vocalic-sequences, but those have merged since then. The 「ええ」 vs 「えい」 story is exactly parallel.

    In addition to eliminating silent intervocalic H/W, the 1946 spelling reform also cleaned up verb endings, since the etymological morphology no longer matched modern pronunciation. This is most marked in the volitional forms since they had V+V combinations which merged into diphthongs across the morpheme boundary, but also shows up in other forms as well. This class of historical sound changes and the accompanying spelling reform are known as 音便 /onbin/ "euphony", and you can read more about this on wikipedia.

    Of course, once you step outside Modern Standard Japanese, there's always the possibility that the dialect of interest does distinguish them. However, while dialects may pronounce things differently, for my part I've never encountered dialects making specific use of this orthographic distinction to try and express their phonological distinctions.

  31. Chris Button said,

    March 31, 2020 @ 7:29 pm

    @ wren romano

    Thanks for that detailed post. I'd figured it would be something like that.

    I've always meant to read Marc Miyake's book on Old Japanese, but have never found the time. I've never needed to really verse myself in such things since my dictionary includes Japanese (along with Mandarin) pronunciations solely for the purposes of functionality. The "etymological" component is premised on Old Chinese reconstructions and early inscriptional evidence. Old Japanese, of course, comes rather later than that, and I have no intention of addressing Pulleyblank's work in that area of Middle Chinese.

  32. Jonathan Smith said,

    April 2, 2020 @ 1:16 pm

    ^ Ah thanks wren romano. I am interested in these changes insofar as they relate to the emergence of long vowels cross-linguistically… in my relatively ignorant POV, representations "oː" / "ō" and parallels in Japanese seem OK pedagogically but dubious as phonological representations.

  33. Chris Button said,

    April 3, 2020 @ 6:07 pm

    @ Jonathan Smith

    if one really wants to talk about contrastive "long vowels" in Japanese then one also has to accept "diphthongs" of all kinds then one also has to accept "diphthongs" of all kinds

    So returning to the "magenē" for 負けない "makenai" example above, it's interesting that the ない component is often clearly broken up as "na.i" in singing but sounds more like English "nigh" when spoken.

  34. John J Chew said,

    April 7, 2020 @ 10:00 am

    Japanese dialectologists can write Miyakoan in modified kana, so I would imagine more normative dialects wouldn't cause much trouble. :)

RSS feed for comments on this post