Japanese 'え' sound - [ɛ] or [e]?

Question

OK, this has been driving me nuts. My friend and I have been learning Japanese off and on for many years, though neither of us are fluent. We got into an argument about the pronunciation of 'え'.

My textbooks said it was an English "short e" as in "bet" or "step." In my opinion, that's the IPA [ɛ]. That's how I've always pronounced it (unless it's a long ええ　and then I use [e:]).

My friend was pronouncing it like the Spanish "e", closer to the English "long a" sound (although not a diphthong) and like the IPA [e].

The wikipedia page on this is inconsistent. It says it's the IPA [e] sound as in American English "bet," but that is definitely not the way I or anyone I know pronounces bet! Here's "bet" in wikitionary, agreeing with me.

When I play clips of Japanese audio and listen closely, I hear [e] in some words and [ɛ] in others. What the heck is going on here? Is there any kind of rule, does it vary by speaker, does no one notice or care?

This difference is very noticeable to me when my friend is pronouncing it. Possibly she's Englishizing the vowel a little bit. When listening to native speakers, I have to focus slightly to hear the difference, but it's still very obvious. What gives?

Answer 1

Japanese, a language which has 3-level vowel height system, does not have the distinction of //e// and //ɛ//. Or speaking more correctly, Japanese え and お are (true) mid vowels, that their sweet spots fall just midway of theoretical [[e]] and [[ɛ]]. (We write them [[e̞]] and [[o̞]] in IPA if necessary.)

(chart from Wikipedia)

And as far as the Standard Japanese concerned, there is no conditional allophone (that you must pronounce exactly //e// in some cases and //ɛ// other), so you may hear the sound //e// or //ɛ// totally randomly, because it is in the range of ordinary fluctuation.

IPA symbols were created for Western European languages in mind, where 4-level height system is prevailing. Compare the Italian (4-leveled) and Spanish (3-leveled like Japanese) vowel diagrams.

Answer 2

As long as typical え is concerned, English e as in bed is a bit too low-toungue'd for え, though it's still included in allophone. In this regard, spanish e is much closer, or the same.

Answer 3

This has kind of already been answered but I wanted some practice plotting vowels according to their first two formant frequencies, so here's an attempt to confirm the sort of mid-ness of え using data. Note that it's all a sort of generalization since vowels vary quite a bit even just accounting for the speaker's head-shape and whatnot.

First, getting data from Japanese speakers on the Japanese え

I will be taking 4 samples per sound per speaker because that seems like a reasonable amount of work to subject myself to. I'll be looking for samples which have a clear, long え vowel because they'll be easier to accurately find the vowels in.

List of Speakers:
StrawberryBrown (took 映画, 英語, 撮影, ええ)
sorechaude (took 英語, 遠征, 工芸, 中米)
chiharu (took 個性が豊かだ, 規定値, 優越性, 統計図)
poyotan (took 書影, CADデータ から2Dモデルを作成, 四カ年計画, 低刺激性) skent (took 先制, 属性, 聖ヨハネ, 繁栄)

I analyzed each of them in PRAAT and the formant frequency data can be found here.

This chart shows how they all look in a vowel space. Ignore the numbers Excel puts on the axes. They're just the minimum values

I'm (as far as I can tell) using the same set up as this except in Excel which doesn't make quite as nice of a graph. (x axis 1300->150, y axis 2900->375, logarithmic scale)

I then initially decided to gather data on English words with e and English words with ɛ without realizing that the two symbols refer to the same phoneme in English. I got them both from searching Forvo for let, mess, neck, deck, pet, and tell.

Here are the two overlaid.

I did try in the end to find at least one example of an actual ɛ and used the Wikipedia recording to get this yellow dot above the Japanese data. (Ignore that one of the other dot types reverted colors please.) The ɛ is ostensibly supposed to be above the e and the e below the ɛ, so I'm not quite sure how to interpret this, but it probably seems at least that the Japanese え is separate from e and also from ɛ and maybe somewhere in-between

Edit: Actually the e that corresponds to ɛ is in fact supposed to be below the Japanese え, so this makes perfect sense. (Since they're actually all ɛ, probably) The "e" that would be probably above the Japanese え is the one in bait. I have no idea why there's such conflicting information on what's an e and what's a ɛ (are dialects quite that varied?) but I'm certainly not the person to sort it out. Questionable now is why the ɛ from Wikipedia is above the Japanese samples though

Answer 4

To answer this question, I checked the sounds of /ɛ/ and /e/ here.

And then I compared it to the pronunciation of 絵 which is the closest word you have to the sound of え. For good measure, I checked the pronunciation of エネルギー.

I conclude that it is closer to /ɛ/ than to /e/.