In a native speaker's mental model, are long vowel sounds - for example the おう in 「教室」 the ああ in 「唐揚げ」, the えい in 「先生」 or just 「いい」 (I chose these examples quite deliberately, since they represent fairly different "categories" of long vowels; please consider them all) - a long vowel sound or two vowel sounds following each other? In other words, when they say a long vowel, are they deliberately saying one long vowel sound or two of them directly following each other?

Does it perhaps depend on which vowel it is? Does it depend on the word? Perhaps neither, or something else entirely? This is something I've been wondering for a while now, and I feel knowing the answer would help my latent understanding of the language and pronunciation.

I realize there's probably no definite answer, at least not without conducting some scientific research, but having the input of some native speakers or someone who's talked to native speakers about this would be invaluable.

  • In the game しりとり I believe you can treat long vowels either way, but I'm not sure if that's helpful to answering the question.
    – Blavius
    Commented Dec 24, 2015 at 17:41

4 Answers 4


I'm a native speaker. When you tell a native Japanese speaker to say these words veeeeery slowly, they would say:

  • きょ、う、し、つ。 (or きょ、お、し、つ。)
  • せ、ん、せ、い。 (or せ、ん、せ、え。)
  • か、ら、あ、げ。
  • こ、ん、ぴゅ、う、た、あ。 (コンピューター)

And if you ask "How many 'sounds' are there in those words?", they would count using their fingers, and say 4, 4, 4 and 6, respectively. So this means so-called "long vowels" are fundamentally two separate sounds in native speakers' minds.

Of course it doesn't mean we deliberately say two vowels in succession in everyday conversations. We do it simply unconsciously.

Reference: Japanese On (or morae)

Japanese morae system and English syllables are so much different that it took me years to understand why English 'strike' is one "sound" (syllable) to English speakers, while it was clearly five sounds (す、と、ら、い、く) to me.

  • 1
    This is a fantastic answer! Thank you. One question, though, regarding the "we do it simply unconsciously" part: does it "feel" like you're saying two sounds after one another that bleed into each other, or just prolonging one, inherently? And is that mental model the guideline for timing how long "long vowels" should be in regular speech? Perhaps these aren't questions that can be answered, seeing as they're not things you think about...
    – obskyr
    Commented Dec 25, 2015 at 19:08
  • 3
    @obskyr The first one is a tough question for me, but we do have a word 長音 (aka 伸ばす音), so basically we think "a long vowel has two morae". See this glico/pineapple/chocolate game, which is described here. Regarding the second question, this model is definitely important in Japanese poetry like haiku/waka (see this). In regular speech, we don't care anyway, but we can somehow distinguish short and long vowels.
    – naruto
    Commented Dec 25, 2015 at 21:10

[W]hen they say a long vowel, are they deliberately saying one long vowel sound or two of them directly following each other?

If this is about , as the tag indicates, the answer will be: two, or neither (at least in Standard Japanese).

It's merely two same vowels adjacent by chance when in between two words, or between word stems and inflections. In your examples, 唐揚げ apparently consists of two words (morphemes) kara + age, thus the two cannot be merged into one in a speaker's mind. いい, although being short, is made of i (stem < yo) + i (adj. ending) and follows it too.

Meanwhile, it's a short vowel followed by a lengthener, when seen inside an indivisible word. 教室 and 先生 in your examples fall under this (have nothing to do with orthographical spellings, just to be sure). By "lengthener" I mean, we recognize a prolonging phoneme //ʀ// as one of moraic phonemes of Japanese (the rest is, moraic nasal //ɴ// a.k.a. ん and geminator //ꞯ//* a.k.a っ, as far as widely accepted among researchers). //ʀ// has no sound value by itself, but can make previous vowel a mora longer. Phonemic representations of 教室 and 先生 are respectively //kyoʀsitu// and //seɴseʀ//.

So what makes you happier if we assume a //ʀ// instead of a long vowel? I think this page I found makes a fairly neat summary on the advantages of //ʀ// analysis. It says (with adapted terms):

  1. When you transpose morae in wordplay, 貧乏【びんぼう】 "poor" with first and third mora interchanged should be ボンビー, which is only accountable if the word was //biɴboʀ//; if it were //biɴboo//, the result would be //boɴbio// ボンビオ.
  2. A series in demonstrative paradigm こう ("in this way"), そう ("in your/its way"), ああ ("in that way"), どう ("how") would be more consistent represented in //koʀ//, //soʀ//, //aʀ//, //doʀ//, rather than simply a long version of each vowel.
  3. (omitted; it's about distinguishing two short vowels and one long vowel.)

cf. the Wiktionary entry of 里親 ("foster parent"):

Compound of sato ‎(“village”) and oya ‎(“parent”). Often cited in contrast to satōya ‎(“sugar dealer”) in discussion of the phonological distinction between long vowels and geminate vowels.

*It's a small capital Q, which has only recently been included in Unicode.

  • Well, to be completely honest the question wasn't really about the laws of Japanese phonology - that was just the closest I got. There's no tag that specifically means "mental model of sounds".
    – obskyr
    Commented Dec 25, 2015 at 18:53
  • 1
    @obskyr But "mental model of sounds" you said is just the meaning of "phonology"! :D Commented Dec 26, 2015 at 8:08

I had a specific experience with 聞く, when it's in て形. So, 聞いて. I used to say "kii-te" emphasizing the length of the vowel. But then, a native told me to separate it, so I now pronounce it "ki-i-te".

It seems that it depends on the word for whether you separate or not. I've noticed with verbs, however, that you separate since, I guess, it's to distinguish the fact it's being conjugated. With nouns, it seems you just put emphasis on the long vowel part when speaking at normal pace. So, in other words, using 先生, you would say, "sen-SEI" rather than "SEN-sei."

...if that makes any sense. Anyway, hopefully my experience helped you, and if you have any further questions, feel free to, well, 聞いてください, haha.

  • So, when you changed how you said this, did the way you thought about it also change? In this case, did you initially think of it as two parts - "kii" and "te" - and afterwards think of it as the three parts "ki", "i", and "te"?
    – obskyr
    Commented Dec 25, 2015 at 18:58
  • 1
    I did, yes. Before I had thought the きい part was one long sound, but now I see it as two separate sounds. As I've learned more over the years though, I have come to the conclusion that even if you pronounce something like it's one long vowel, Japanese (as @naruto pointed out) still think of it as two. Learning, yo. haha Commented Dec 26, 2015 at 19:13

When they say a long vowel, are they deliberately saying one long vowel sound or two of them directly following each other?

I think there is no convention, as long as there is no risk of misunderstanding.

委員長{いいんちょう} may be pronounced with two slightly separated いいsounds or a long いー.

大山椒魚{オオサンショウウオ} (giant salamander) may be pronounced like おーさんしょーうお, おおさんしょーうお, etc (nothing that varies too much from this will be understood as anything else than 'giant salamander').

Does it perhaps depend on which vowel it is? Does it depend on the word? Perhaps neither, or something else entirely?

I think most on-yomi readings, in the same kanji, do not use two vowel sounds following each other. 栄養 would sound just like えーよー, 修正 would be しゅーせー. The sound often slightly goes to the sound of the second vowel but it is continuous (no perceivable 'pause' between vowel sounds). There is no obvious need to emphasize the vowel sounds like え or しゅ because saying like that is enough.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .