Joou-sama no Eshi

[Neushaartje]

@Lynkaster
How is the story from Huzimaki's perspective? There are very few chapters that are from any of the four's perspective? Most scenes feature at least two of the four main characters? We had countless scenes that just involved Yukimori and Miyazima, or Yukimori and Nagase, with Huzimaki nowhere præsent in them. All of the characters are given thought bubbles.

 

[Insignificant]

People's butthurt made me to give this manga a try.

 

[Insignificant]

Well, manga is average. And even good ending didn't save the day. I don't know if it's true, but some people say it got axed. That would explain the rapid development in last chapters. And if it's true, that some great irony you have here.
Well, with good ecchi at the start, great ending, and delicious people's pain in the comment section it's a firm 7/10.
https://----[! URL is only accessible to you. !]----/OCzmkaCQKWwPkEv1_2417QiZU2ogjCfYsF9tEGq7y98I0wo_m320zDo6IGyLA8_F0bHUEiGHT0-Hz_9OlfGMecji-yHomLm1kYJPDbDlUUs5H163Gy37oPwC8PLO_hobKAF35nShmbKoh8xEsXZYUSAIVT3eKYjmlyjs6WxhUmT-fl4g1r5k20ONfRyDzOSdVQI46C3TlRPwImPYqETPjPu1RuDH88JRgye28igRCMjPdvSyftXqm97Zr7mdXPFxTL5tQ_fSi0MHm7nrtDQmFIXs-2pOXg19p4M0wUkmJfO37UydKzjr4U9CWIyW548EQWNxz5nbVXrCDJturyQ6wSg3r-A0EtQRt09VPZuvuGPotrX3hbFate-MzYDGodJanjEumzq2OraZIUbS6xjU4AfzzAdw3BagLiTAU1D-JqKdqvF4IhEZjSu-FC2xrRtwPC2KU7eaFh2DVcaiJQF6dmgE1hNIGc50gt-sZc_m1qRQcEO6gotVZ09p3N-8OW0BStPRKhsi5SL_3c0YQJUYCq_LeBfgWZp40ulZb3rbsxi-hwoqAQNTR4KbzkET3X_jtVcgfXkU1LP4PwuLMbSx8pjx903GreWlwF-vs7x_2YKkGw3JKhZ5cQjOrxYivmT0UtctmiKTEVp-4ewLjQgfCQaTwc57F_wqT1blGVgSK8stzp7VBfRBgTznsepd=w1791-h1008-no

 

[sholum]

Sorry, I'm just here to get triggered over @trapsarebetter using kunreishiki for romanization.

 

[Neushaartje]

@sholum
I actually use the original, lossless scheme developed by Aikitu Tanakadate, which keeps everything as spelt, not the later cabinet direction thereof, which collides identically pronounced segments: so I would romanize it as "Noidi Itou", not "Noizi Itô".

 

[sholum]

@trapsarebetter
Well, that's a bit better, since you're at least going for a complete function (grid representation, like the 五十音). But it defeats the purpose of romanization in this context: transcribing Japanese to be more intelligible to other (Western) languages. If someone has to go through the learning effort to know that di should be pronounced ぢ, why not just learn the kana?
Not that Modified Hepburn is any better (おお and おう convert identically).

Ideally, we'd use a system that was both unique and understandable to the intended audience. Personally, I lean towards making it more understandable, but try to maximize both using exceptions for dzu, dji, and the not-really-diphthongs (where I accept the loss of the second character and call the whole thing 'sha', for example).

If I were writing for someone that knows Japanese, I'd just use Japanese. Computers are much more convenient these days.

It's a pet peeve that I could expound on for paragraphs more, but I'll save it.

TL;DR
Just poking you a bit because romanization of Japanese is something I've put way too much thought into.

 

[Neushaartje]

@sholum
> But it defeats the purpose of romanization in this context: transcribing Japanese to be more intelligible to other (Western) languages. If someone has to go through the learning effort to know that di should be pronounced ぢ, why not just learn the kana?

I strongly disagree with the thesis that is often put forth that Hepburn-romanization is more intelligible to non-Japanese speakers; it is actually in many cases further away from it than Japanese-style romanization. In particular it's election to replace <sy> with <sh> and <ty> with <ch> only moves it further away from the actual pronunciation. English has the phonemic sequence /sj/ in words such as "assume" which is pronounced "assyume", not "asshume" which would be a mispronunciation. Likewise "shougun" is not pronounced "shōgun̄" but rather "syōgun̄" with "sy", not "sh", in Japanese, and hence spelt as such.

A system that actually guides pronunciation well would yield something like "Dzyo-ō-sama no Esyi", for し is pronounced neither "si" nor "shi" but "syi", in general, anyway, as it is pronounced more like "si" again when one raises one's voice, which makes it even more complicated. じょ is also pronounced like "zyo" when it follows a vowel and more like "dzyo" in any other context, never like "jo" — the idea of trying to capture the surface realization for English speakers is rather impossible task to begin with anyway, because as I said, what Japanese speakers non-contrastive is considered contrastive to English speakers, so a word like "いつか" can be heard as "ituka", "itsuka" or "itska" by English speakers depending on whatever the Japanese speaker feels like. Of course, a more famous problem is that though it's conventionally written as "r", half of the time it will sound more like "l" to the ears of English speakers.

 

[sholum]

@trapsarebetter
>In particular it's election to replace <sy> with <sh> and <ty> with <ch> only moves it further away from the actual pronunciation.
I am confused with this. English doesn't use the pattern of <sy> or <ty> to describe a sound anywhere near the pronunciation of those Japanese sounds; <sh> and <ch> are.
There is that argument to be had with regards to certain other Western European languages (namely French), but French is far from my specialty.
But since you mentioned Hepburn specifically and we're discussing this in English, we'll default to English.

Your example of "assume" is something I'd rather not mess with, since it's one of those words where the difference between British and American English really stand out. In American English, "assume" sounds like "uh-soom" not "assyume". I had to think about this one for a bit, since I'm an American.
Either way, though, 'assume' is still not spelled with 'sy'.

In standard Japanese (I won't speak to any other dialect, but for a bias reference, most of the time I've spent in Japan has been in Nagano and Okinawa), し does not sound like 'si' as a rule. Plenty of exceptions in context, as you mention, but if you asked for a Japanese person to read the character し it would sound closest to word 'she' (obviously without the vocal stress inherent to English). Of course, to keep things somewhat representative of the kana, it's preferable to write it as 'shi'. しゃ doesn't sound like 'sya', it sounds most similar to 'sha'. And so on.

> Of course, a more famous problem is that though it's conventionally written as "r", half of the time it will sound more like "l" to the ears of English speakers.

This famous problem... It's neither sound. It's a sound that isn't really used in English at all (the inverse problem with Japanese ESL students is for the same reason). If we're being strict with the way the sound is formed, it's closer to the English 'l', but it tends to sound more like 'r'.
The Spanish 'r', on the other hand, is very similar. And Spanish phonetics are fairly accessible to European and American speakers of English. So ultimately, I prefer representing it with 'r'.

> A system that actually guides pronunciation well would yield something like "Dzyo-ō-sama no Esyi"[...]

I think it's best to avoid putting consonants together when romanizing Japanese for English speakers (with exception for 'sh', since that represents a single sound); this goes along with my dislike of the use of 'sy'. This particular example of yours should make that really clear. It compels a native English speaker to put a lot of stress on the consonant sounds and to elongate them. You can see similar problems when English speakers try to pronounce Slavic or Nordic names.
Funny enough, when reading your example there, I read it like it was from Nou. So if we were back before the shogunate, then I might be sitting here agreeing with you instead, lol.
And, of course, I can't stand the use of macrons, because it drops information unnecessarily.
And the 'z' sound in English doesn't show up in じょ.


With regards to pronunciation similarity in general; people will still screw it up regardless of the system, unless they know what Japanese sounds like (for example, neither of our preferred systems would prevent people from pronouncing 東京 as 'toe-key-oh'; btw, I'd write it as Toukyou). But you can limit the range of the error by using familiar structures.

 

[Neushaartje]

@sholum
>I am confused with this. English doesn't use the pattern of <sy> or <ty> to describe a sound anywhere near the pronunciation of those Japanese sounds; <sh> and <ch> are.
It doesn't, because English spelling does not reflect pronunciation; <gi> in English is also very often not pronounced as in Japanese but as in the English word "ginger", but the actual sound occurs in English.

Wanting to introduce segmental orthographical conventions from English is silly when things like <hy> or <ky> already exist in Hepburn Romanization which also do not occur in English. If it can handle <hy> to indicate the pronunciation, it can also handle <sy> which indicates the actual pronunciatio; <sh> is simply a mispronunciation. "suitable", pronounced "syootable" and "shootable" mean two different things and the <sy> row in Japanese is pronounced like the former, not the latter.

>Either way, though, 'assume' is still not spelled with 'sy'.
Neither is "cute" spelt <kyoot> in English, but it is still pronounced that way, so why not the same objection against Hepburn's <ky>? the <ka> orthography is very rare in English, it is almost always spelt <ca>, why not do that too then if your goal is simply to mirror English orthography, not indicate pronunciation.

>In standard Japanese (I won't speak to any other dialect, but for a bias reference, most of the time I've spent in Japan has been in Nagano and Okinawa), し does not sound like 'si' as a rule.
Indeed, but it doesn't sound like "shi" either, as I said, it sounds like "syi" in general, but can also sound like "si", especially when raising one's voice, because Japanese does not contrast i/yi (or e/ye) phonemically. It is however never pronounced "shi", the sound of English "sh" simply does not occur in Japanese, at all. The vowel /i/ in Japanese is typically pronounced like "yi" when following a consonant, especially coronal ones, and "i" otherwise.

>This famous problem... It's neither sound. It's a sound that isn't really used in English at all
Actually it is a sound that occurs in English; it is in fact very often the same sound that many North American speakers make when pronouncing a "ẗ" or a "d" in between two vowels. Thence Japanese speakers tend to hear their version of /r/ when North-American speakers pronounce "pudding" giving "purin" but when other speakers do they her "pudingu"

>I think it's best to avoid putting consonants together when romanizing Japanese for English speakers (with exception for 'sh', since that represents a single sound);
And that is misleading, because it is in fact two sounds. か is to ikt as さ is to しゃ so why is <kya> acceptable, but not <sya>?

>And the 'z' sound in English doesn't show up in じょ
But it does, じょ is pronounced "zyo" after a vowel, and "dzyo" otherwise. As in is pronounced how in British English "presume" would be pronounced, as in "prezyoom", which is not pronounced "prejoom".

>With regards to pronunciation similarity in general; people will still screw it up regardless of the system
Which is exactly why I think a pronunciation-based system opposed to an orthography-based system is fruitless. But Hepburn in many cases isn't even a good pronunciation-based system that transcribes how it sound be pronounce — it is in fact further from that in many cases, because Japanese orthography is fairly phonemic.

>But you can limit the range of the error by using familiar structures.
Not if those structures repræsent different sounds. しゃ in Japanese is pronounce "sya", not "sha", "sh" is one sound, different from the two sounds "s" and "y" in succession, which is what the segment is in Japanese, and there are words in English that are contrasted by only this difference like as I said "suitable" vs "shootable", the former pronounced "syootable" and sounding different from "shootable".

 

[sholum]

@trapsarebetter
I'm not interested in derailing this thread, so we'll have to agree to disagree (but I'm going to try to get the last word, just because).
Not that I can know how you hear things, but none of your equivalences make any sense to me.

> But it does, じょ is pronounced "zyo" after a vowel, and "dzyo" otherwise. As in is pronounced how in British English "presume" would be pronounced, as in "prezyoom", which is not pronounced "prejoom".
But it doesn't sound like zyo... and じゅ doesn't sound like the 'su' in British pronunciation of presume, which is /zju:/
https://en.wikipedia.org/wiki/Voiced_alveolo-palatal_affricate
In English 'j' is /dʒ/; it is the closest that is intuitive for the isolated case. While your suggestion of 'dz' could be pronounced in a way that is closer to the above, I think it's more likely to be pronounced /dz/ and would conflict with the romanization of づ and ぢ (which would still have the same problem of being pronounced as /dz/, but their lower use frequency makes me prefer that 'j' be used for all of the じ variants).
To be clear, I think romanized representations should be based on the reading of the isolated structure.

><sy> which indicates the actual pronunciatio; <sh> is simply a mispronunciation
it doesn't sound like sy. It's closer to sh. The Wikipedia pages have audio samples, which is why I'm linking them.
the 'sh' sound in し:
https://en.wikipedia.org/wiki/Voiceless_alveolo-palatal_fricative
'sh' in English:
https://en.wikipedia.org/wiki/Voiceless_postalveolar_fricative#Voiceless_palato-alveolar_fricative
's' in English:
https://en.wikipedia.org/wiki/Voiceless_alveolar_fricative#Voiceless_alveolar_sibilant

And again, the example you give with assume:
in British English, the 'su' is pronounced 'sju:' (in American, it's pronounced 'su:'). Since you keep coming back to this, I decided to actually look it up.

>Actually it is a sound that occurs in English; [re: Japanese 'r']
Because the ラ行 sounds are formed with the tongue placed between the positions of English 'l' and 'd', but with only a momentary contact with the palate (which is why it gets some 'r' flavor). The Cambridge dictionary (the first one I found online that uses IPA for pronunciation) doesn't suggest that 'd' after a vowel is pronounced all that similarly to the 'r' in Japanese (they show the 'dd' part of 'pudding' as /d./), but the previous sentence explains why there is crossover.


>Thence Japanese speakers tend to hear their version of /r/ when North-American speakers pronounce "pudding" giving "purin" but when other speakers do they her "pudingu"
No. When the written word プリン was incorporated into Japanese, it was forced into the spelling it now has in order to fit with the writing system. The use of small kana to emulate foreign sounds (i.e. ディ) wasn't a thing until well after the word 'pudding' had been introduced. Mid 20th century (earliest date I could find with a quick search was 昭和２１ (1946)), vs pudding coming to Japan after foreign trade reopened at the end of the Edo period.
The writing reforms of post-war Japan make for some interesting times when researching Japanese WWII artifacts.

>And that is misleading, because it is in fact two sounds. か is to ikt as さ is to しゃ so why is <kya> acceptable, but not <sya>?
Compromising. Even if we don't use 'kya' (or kyo, as with my example of 東京) in English, it isn't such an unnatural structure for English as 'sya' is, IMO (and again, it's not even the correct sound). We're not going to get perfect pronunciation without teaching people the correct pronunciation anyway (and given the audience of this site, I think this reduces the amount of error even more), but this is both more intuitive than Nihon-shiki and more representative than Hepburn.

The point of representing し as 'si' (and the variants) in Nihon-shiki is to make to make things one-to-one with the 五十音 grid (サ行、イ段 = si). It is meant for use by Japanese speaking people. It's whole purpose was to replace the Japanese writing system. It preserves a lot of grammatical rules that other romanization styles do not, but does not mean that it makes sense to English speakers (or others).

 

[Neushaartje]

@Sholum

>But it doesn't sound like zyo... and じゅ doesn't sound like the 'su' in British pronunciation of presume, which is /zju:/

You know that /j/ is the i.p.a. symbol for the <y> in Japanese Romanization, right? This is pretty much saying that it's "zyu" — I can't claim to follow you any more.

>In English 'j' is /dʒ/; it is the closest that is intuitive for the isolated case.
No it isn't, /zj/ is, which is, as you said, the sound of "presume". "presume" is pronounced /pɹɪˈzjum/, with the /zj/ being pretty much exactly the sound of じゅ in Japanese which is also /zju/.

> it doesn't sound like sy. It's closer to sh. The Wikipedia pages have audio samples, which is why I'm linking them.
the 'sh' sound in し:
https://en.wikipedia.org/wiki/Voiceless_alveolo-palatal_fricative
'sh' in English:
https://en.wikipedia.org/wiki/Voiceless_postalveolar_fricative#Voiceless_palato-alveolar_fricative
's' in English:
https://en.wikipedia.org/wiki/Voiceless_alveolar_fricative#Voiceless_alveolar_sibilant

Because you're comparing it to English /s/, not to English /sj/ as in in /sjut/ for "suit".

https://commons.wikimedia.org/wiki/File:En-uk-suit.ogg Listen to the first pronunciation; this is pretty much completely identical to the rendition of Japanese /sj/. English contrasts /sj/ from /ʃ/ and /sj/ is definitely the closest to Japanese /sj/ or /si/ of both.

>No. When the written word プリン was incorporated into Japanese, it was forced into the spelling it now has in order to fit with the writing system. The use of small kana to emulate foreign sounds (i.e. ディ) wasn't a thing until well after the word 'pudding' had been introduced. Mid 20th century (earliest date I could find with a quick search was 昭和２１ (1946)), vs pudding coming to Japan after foreign trade reopened at the end of the Edo period.
The writing reforms of post-war Japan make for some interesting times when researching Japanese WWII artifacts.

Even then, it would have become プジング, had an Englishman rather than American said it — also note that it's not just how intervocalic /d/ became /r/, but that postvocalic /ŋ/ became /N/, a common feature of North American English is that the postvicalic velar nasal is realized so closely to /n/ that Japanese speakers no longer hear the difference.

Regardless: https://en.wikipedia.org/wiki/Voiced_dental_and_alveolar_taps_and_flaps, you can look up for yourself that this phone is indeed both the realization of the North American intervocalic /d/ at times, and the Japanese /r/.

>The point of representing し as 'si' (and the variants) in Nihon-shiki is to make to make things one-to-one with the 五十音 grid (サ行、イ段 = si). It is meant for use by Japanese speaking people. It's whole purpose was to replace the Japanese writing system. It preserves a lot of grammatical rules that other romanization styles do not, but does not mean that it makes sense to English speakers (or others).

It doesn't follow pronunciation but Japanese organization because Japanese does not contrast "i" from "yi" and considers it the same sound, yes. But to be most accessibly indicate the pronunciation to English speakers it should not be "shi" but "syi".

Again, listen to that fragment of how "suit" is pronounced; the initial segment is pretty much completely identical to what Hepburn choose to write as <sh>. English contrasts /sj/ from /ʃ/ with the former being almost completely identical to what Hepburn writes as <sh>, not the latter. <sy> as pronounced in "suit" which is pronounced /sjut/ is a much better approximation of the sound; it's in fact close to perfect.

 

[sholum]

@trapsarebetter
>You know that /j/ is the i.p.a. symbol for the <y> in Japanese Romanization, right? This is pretty much saying that it's "zyu" — I can't claim to follow you any more.
Yes. That's why I linked to /dʑ/ for a description of what the consonant of じゅ sounds like, so it would be /dʑu/ in total. English 'j' sound in 'jump' is /dʒ/.
English 'z' is just /z/. So romanizing じゅ as 'zyu' is suggesting a pronunciation of /zju/ (like in 'presume') to an English speaker (assuming they picked the right vowel sound). Whereas 'ju' suggests /dʒu/, which is much closer.

>Because you're comparing it to English /s/, not to English /sj/ as in in /sjut/ for "suit".
I isolated the /s/ to emphasize the difference.
The Japanese sound is /ɕ/, according to what I've found (I'm not that great at using IPA directly, tbh, so I have to look them up to compare).
The English 'sh' is /ʃ/.
These are both fricative and they are produced with the tongue at similar parts of the palate. Your substitution of /sj/ doesn't account for the fricative aspect, which is very noticeable compared to the minor difference in tongue placement. The only reason it comes close, if you ask me, is because in moving your tongue from the position of /s/ to /j/, you have to drag your tongue across the contact point for /ɕ/.
A comparison:
https://jisho.org/search/%E4%B8%BB%E7%BE%A9
https://dictionary.cambridge.org/dictionary/english/shoot
Compare these with your example of 'suit' and look up any others you'd like. The fricative aspect does not exist in 'suit' /sj/ and is obvious in both 'shoot' and 主義.
BTW, I checked with Google's TTS also: the fricative is a bit less noticeable in their rendition, but it also sounds crappy in general (the one at https://texttospeech.io/ sounds better). None of these sound like the 'su' in 'suit' to me.
Just as an interesting aside, while I was trying to find examples, I ran across this stackexchange question:
https://japanese.stackexchange.com/questions/11092/pronunciation-of-%E3%81%97%E3%82%83-%E3%81%97-%E3%81%97%E3%82%85-%E3%81%97%E3%81%87-and-%E3%81%97%E3%82%87

 

[cksn]

this is shit compared to kodomo no jikan
that actually showed realistic natural sexuality, it was a breath of fresh air
this is simply perverted

 

[Neushaartje]

@sholum

I actually forgot about this discussion and stumbled upon it again moons later, but incidentally I made some spectrograms a couple of months ago:

The fragments I used for this are obtained from https://forvo.com/word/%E9%80%B1/#ja, https://forvo.com/word/soon/#en, https://forvo.com/word/suit/#en, https://forvo.com/word/shoot/#en — I used the male voices for all of them.

It should be quite obvious that the “suit” matches the Japanese sound the most and that both are also a transition between three loci, indicating of that both are three phonemes. In English “soon” an “shoot” there are only transitions between two loci, whereas in both English and Japanese /sju/ there is clearly a transition from [ s] to [ j] to [ u] going on between three loci — the fricative part in /su/ and /šu/ in English is constant, whereas /sj/ in both English and Japanese has a moving fricative in a fairly similar manner.

 

[sholum]

@trapsarebetter
If you would, specify your choices more, please. There are at least two male voices for all of those words on Forvo, and they all sound different to me.

The female Japanese voices for 'shuu' support your argument better, btw. Two of them pronounce it with less of a fricative sound at the front and a distinct change from し to ゆ; the remaining female voice (from inuwanko), less so.
Neither of the male voices use the harder 's' sound, but skent retains the distinction between し and ゆ more than Jun_Japan does.

I agree that there is a relic of the fact that しゅ used to be しゆ in the pronunciation, such that it has more loci than the 'shoo' in 'shoot'. This still doesn't justify the use of 'syu' over 'shu' though (if anything, it's arguing for 'shyu'). If you look at the difference in the shape of the curve between 週 and 'suit', you can see that the transition from /s/ to /u/ in 'suit' is much more front heavy (with negative curvature on that third band from the top) than that of 週, where it very quickly settles to /u/ (and has positive curvature).

And while I may be misreading this next part: it seems to me that the consonant sound at the beginning of 週 matches to 'shoot' more than 'suit', which matches more to 'soon'.
Granted, the 'sh' in 'shoot' is much more dense (not knowing the technical term).
To be clear, I'm looking at the spread of the frequencies in each sound. The main feature I'm catching is the presence or lack of a band around 3kHz (assuming its linear scale).

While your spectrogram of 'shoot' (again, I'd like to know which voice you used, since the UK and US pronunciations on that page differ significantly) does show that it is a single vowel sound, because the Japanese sound settles to the final /u/ much more quickly than 'suit' (along with the other differences I noted), I think this shows the sound in 'shoot' to be a better analog than 'suit'. Of course, neither are perfect.

Just a reminder (for both of us), my argument is that, to a non-Japanese reading audience, specifically English speaking, certain nihon-shiki conventions do not convey a pronunciation as close to the original as other, foreigner-focused methods do.
Honestly, I think the only way we'd be able to actually prove one way or the other which one produces a closer sound would be to run a study where English speakers who don't know Japanese at all are primed with an audio track and then asked to pronounce words written in different styles and compare the results, preferably by having a native Japanese speaker assess whether they just sound like they have a thick foreign accent or whether the pronunciation is completely wrong. If you know of one, feel free to send it, I'd like to look at it. I doubt I could get my college to fund something like that (for one, my major wasn't even in the humanities, let alone linguistics).

 

[Neushaartje]

@sholum

If you would, specify your choices more, please. There are at least two male voices for all of those words on Forvo, and they all sound different to me.

The female Japanese voices for 'shuu' support your argument better, btw. Two of them pronounce it with less of a fricative sound at the front and a distinct change from し to ゆ; the remaining female voice (from inuwanko), less so.
Neither of the male voices use the harder 's' sound, but skent retains the distinction between し and ゆ more than Jun_Japan does.

As I remember I used “skent” as I remember that name but it was a while back — they all produce rather similar results though.

I would stress that it has nothing to do with fricativation, all four spectograms show a clear fricative that becomes a vowel; the vowel is indicated by the dark bands at the bottom indicating vowel formants, and the fricative is indicated by the dark bands at the to, indicating turbulence.

What stands out in both English and Japanese /sju/ is a dark band that slides from the top to the bottom, this indicates a change in location during the fricative of the place of articulation, in particular it indicates the tongue moving to the palate to create /j/, the palatal glide, and then during the vowel part the bands move again as the tongue moves back for [ u] — what stands out most is that both in English and Japanese /sju/, there is a clear transition between three loci, indicating that in both cases it is a realization of three phonemes, smoothly transitioning into one another. This is not the case in English /šu/, and /su/ which are two phonemes, and the fricative part is constant, and has no dark band that moves around — the fricative is constant.

I agree that there is a relic of the fact that しゅ used to be しゆ in the pronunciation

That <しゅ> was spelt as <しゆ> before 1940 was merely orthography, it was already pronounced the same way as it was today; the orthography was simply ambiguous and they resolved this by introducing smaller versions of many characters at that time.

This still doesn't justify the use of 'syu' over 'shu' though (if anything, it's arguing for 'shyu').

I do believe this has merit to be argued, yes. But whether shyu or syu is the closest to しゅ; shu is definitely the furthest with it's completely constant fricative.

If you look at the difference in the shape of the curve between 週 and 'suit', you can see that the transition from /s/ to /u/ in 'suit' is much more front heavy (with negative curvature on that third band from the top) than that of 週, where it very quickly settles to /u/ (and has positive curvature).

They are indeed not exactly the same, and you will find differences in these matters from speaker tos peaker.

But taken on the hole, it is undeniable that an English rendition of /sju/ is far closer to Japanese /sju/ than an English rendition of /šu/ is. It will always be an approximation, but <syu> comes closer, perhaps <shyu> is indeed even closer; it would be hard to tell because I know of no English word wherein /šju/ actually occurs in English.

While your spectrogram of 'shoot' (again, I'd like to know which voice you used, since the UK and US pronunciations on that page differ significantly) does show that it is a single vowel sound, because the Japanese sound settles to the final /u/ much more quickly than 'suit' (along with the other differences I noted), I think this shows the sound in 'shoot' to be a better analog than 'suit'. Of course, neither are perfect.

The vowel is not so much the interesting part as that the fricative part is almost completely constant. The quality of the fricative part does not change in any way in both “shoot” and “soon”.

I remember using the one from Earthcalling, hearing it again, the top one.

Just a reminder (for both of us), my argument is that, to a non-Japanese reading audience, specifically English speaking, certain nihon-shiki conventions do not convey a pronunciation as close to the original as other, foreigner-focused methods do.
Honestly, I think the only way we'd be able to actually prove one way or the other which one produces a closer sound would be to run a study where English speakers who don't know Japanese at all are primed with an audio track and then asked to pronounce words written in different styles and compare the results, preferably by having a native Japanese speaker assess whether they just sound like they have a thick foreign accent or whether the pronunciation is completely wrong. If you know of one, feel free to send it, I'd like to look at it. I doubt I could get my college to fund something like that (for one, my major wasn't even in the humanities, let alone linguistics).

Sadly I don't.

My claim is not that Nihon-Siki is all that accurate, merely that it is more accurate in the case of しゅ and it's brothers. Hepburn is certainly more accurate with し, which is more accurately syi, but Nihon-Siki is not attempting to be a pronunciaation guide, but a spelling guide.

I simply feel that Hepburn is so inaccurate to begin with, also with things such as <zenbu> which should be <dzembu> to most accurately reflect how it's pronounced that it seems silly to use it for any purpose.

I also don't like that Nihon-Siki uses double constants for <っ>, I would have much præferred something like consistently using <t̮>, a t with an underdot, such that it becomes <Nit̮poṃ> — it is not even pronounced with a doubled p to begin with. I would præfer to consistently render <ん> and <っ> as <ṃ> and <ṭ>.

 

[sholum]

@trapsarebetter

Thanks for the clarification on the speakers.

That <しゅ> was spelt as <しゆ> before 1940 was merely orthography, it was already pronounced the same way as it was today; the orthography was simply ambiguous and they resolved this by introducing smaller versions of many characters at that time.

I seem to remember that 'Ye Olde' Japanese (from around the introduction of katakana, and presumably before) is thought to not have those sounds, thus the orthography didn't reflect it. I'll leave that alone though, as I don't know that much about it.

I simply feel that Hepburn is so inaccurate to begin with, also with things such as <zenbu> which should be <dzembu> to most accurately reflect how it's pronounced that it seems silly to use it for any purpose.

Trust me, you won't find me advocating Hepburn as it is (Traditional or Modified). I think I said this in a previous post, but I despise the use of macrons for extended vowels; it's completely unnecessary, especially for native English speakers. There isn't an existing, official system that I think efficiently transcribes Japanese for a non-Japanese audience.

[...] also with things such as <zenbu> which should be <dzembu> to most accurately reflect how it's pronounced that it seems silly to use it for any purpose.

It's funny you say that, because older versions of Hepburn romanized ず as 'dzu', which is why we call the plant 葛 'kudzu' instead of 'kuzu'.
I don't think that extended to the rest of the ザ行 though.
I don't prefer this romanization though, because it encourages a prominent 'd' sound where it doesn't belong. Or, in the case of 'kudzu' makes English speakers pronounce it 'cud-zoo'. 'Zu' is close enough, IMO.

I also don't like that Nihon-Siki uses double constants for <っ>, I would have much præferred something like consistently using <t̮>, a t with an underdot, such that it becomes <Nit̮poṃ> — it is not even pronounced with a doubled p to begin with. I would præfer to consistently render <ん> and <っ> as <ṃ> and <ṭ>.

I agree that there is a lack of distinction for ん. When needed, I've used 'n'' (i.e. gen'in). Using 'm' with some modification, as you did, makes a bit more sense than 'n' phonetically, though it's never been a big complaint of mine other than that it should be transcribed the same in every case. My only complaint with those underdots is that it's a pain to type.

Double consonants get the idea across well enough, IMO. I don't mind distinguishing it with its own typography, since っ is its own character, but it doesn't seem like either would convey pronunciation better than the other for people without much knowledge of Japanese.


What's become apparent to me is that we are weighing the importance of certain sounds and symbols for pronunciation more or less than others. At least we both agree that the existing systems aren't that good.

 

[Neushaartje]

@sholum

I seem to remember that 'Ye Olde' Japanese (from around the introduction of katakana, and presumably before) is thought to not have those sounds, thus the orthography didn't reflect it. I'll leave that alone though, as I don't know that much about it.

Yes, but that is very far back, think 900's. /sja/ did not arrive from contracting /sija/, but from loans from Chinese; at this time /kwa/ and similar sounds were also introduced that eventually disappeared again.

Note that Hepburn and Nihon-Siki in the 1980s already included <sha> and <sya> even though what is now <しや> and <しゃ> was spelt the same at the time, relying on context; Romanization in the 1980's was thus far less ambiguous than the kana script — this was of course not really a problem since Japanese is and was mostly written with Chinese characters to begin with.

Trust me, you won't find me advocating Hepburn as it is (Traditional or Modified). I think I said this in a previous post, but I despise the use of macrons for extended vowels; it's completely unnecessary, especially for native English speakers. There isn't an existing, official system that I think efficiently transcribes Japanese for a non-Japanese audience.

It's funny you say that, because older versions of Hepburn romanized ず as 'dzu', which is why we call the plant 葛 'kudzu' instead of 'kuzu'.

Not entirely, it Romanized <づ> as <dzu> and <ぢ> as <dji> in older versions, <ず> and <づ> are of course pronounced identically in modern standard Japanese.

I don't prefer this romanization though, because it encourages a prominent 'd' sound where it doesn't belong. Or, in the case of 'kudzu' makes English speakers pronounce it 'cud-zoo'. 'Zu' is close enough, IMO.

The same problem exists with <tsu>, which would inspire an English speaker to pronounce <mitsubishi> as as “mit-su-bi-shi”, which a Japanese speaker then could re-analyse as “ミトスビシ”

One will be understand using a [z] everywhere indeed, but one will also be understood pronouncing <し> as “si”, and Japanese persons can't easily hear the difference between the English words “she” and “see” anyway.

I agree that there is a lack of distinction for ん. When needed, I've used 'n'' (i.e. gen'in). Using 'm' with some modification, as you did, makes a bit more sense than 'n' phonetically, though it's never been a big complaint of mine other than that it should be transcribed the same in every case. My only complaint with those underdots is that it's a pain to type.

Neither Hepburn nor Nihon-Siki have them consistent, both employ the same rule that <n'> be used before a vowel or <y>, and <n> in all other cases as it's not ambiguous any more this way.
This scheme ironically creates an ambiguity in extended Nihon-Siki which is used to transcribe the newer katakana combinations for loaned segments that uses <'i> and <yi> to distinguish some things that were never in native Japanese, so <an'i> can now either stand for <あんい> and <あねぃ> though <ねぃ> is only theoretical; I've never actually seen it anywhere.

Double consonants get the idea across well enough, IMO. I don't mind distinguishing it with its own typography, since っ is its own character, but it doesn't seem like either would convey pronunciation better than the other for people without much knowledge of Japanese.

Do they get it across well? It's not pronounced as a double consonant except for <ss>; and most importantly some dialects of Japanese can feature <っ> in positions before another consonant. Some dialects have words like <くっ>, how would one transcribe that?

What's become apparent to me is that we are weighing the importance of certain sounds and symbols for pronunciation more or less than others. At least we both agree that the existing systems aren't that good.

Agreed, but there's one difference: I believe that what you desire to achieve is impossible and therefore shouldn't be focused on. You want a pronunciation guide; I believe that an English speaker with no education of Japanese phonology can never reproduce Japanese accurately to begin with and that furthermore the pronunciation isn't stable from the perception of an English speaker.

https://www.youtube.com/watch?v=58u5UzEADHg#t=0m09

I believe that the way this Japanese speaker pronounces this word will be heard as “ftari” the first time, and “hutari” the second time by an English speaker. I believe that this word can be pronounced by the same speaker as something an English speaker will hear as “ftali”, “futali”, “ftari”, “futari”, “hutari”, “hutali” depending on random factors.

Even if we ignore the r/l distinction there are still three different ways an English speaker can hear it as “ftari”, “futari”, and “hutari” — this problem cannot easily be solved and I believe it to thus be impossible to have a system that indicates pronunciation well from the perspective of English speakers.

 

[sholum]

@trapsarebetter

Not entirely, it Romanized <づ> as <dzu> and <ぢ> as <dji> in older versions, <ず> and <づ> are of course pronounced identically in modern standard Japanese.

As far as I know, kudzu wasn't called クヅ in the past.
Not exactly the best results, but here's what I found on wiki:
Some other names for kudzu:
https://ja.wikipedia.org/wiki/%E3%82%AF%E3%82%BA#%E5%90%8D%E7%A7%B0
Old Hepburn romanization differences:
https://infogalactic.com/info/Hepburn_romanization#Obsolete_variants

Not that important to our conversation, I just thought it was interesting.

The same problem exists with <tsu>, which would inspire an English speaker to pronounce <mitsubishi> as as “mit-su-bi-shi”, which a Japanese speaker then could re-analyse as “ミトスビシ”

True.

This scheme ironically creates an ambiguity in extended Nihon-Siki which is used to transcribe the newer katakana combinations for loaned segments that uses <'i> and <yi> to distinguish some things that were never in native Japanese, so <an'i> can now either stand for <あんい> and <あねぃ> though <ねぃ> is only theoretical; I've never actually seen it anywhere.

That's pretty interesting. Romanizing these new forms is definitely a challenge. Thankfully, I don't think there'll be much of a need to do so, since these are all for foreign sounds.

Do they get it across well? It's not pronounced as a double consonant except for <ss>; and most importantly some dialects of Japanese can feature <っ> in positions before another consonant. Some dialects have words like <くっ>, how would one transcribe that?

Again, 'well enough'. In English, gemination isn't much of a thing; it only shows up in certain cases of double consonants. It's workable.

For the odd case you provide, I'd say it's a point in favor of removing the double consonant convention in favor of something else. Only for completeness though, as I don't think that's something that will ever really come up.

Agreed, but there's one difference: I believe that what you desire to achieve is impossible and therefore shouldn't be focused on. You want a pronunciation guide; I believe that an English speaker with no education of Japanese phonology can never reproduce Japanese accurately to begin with and that furthermore the pronunciation isn't stable from the perception of an English speaker.

I agree that it is impossible to get a perfect pronunciation guide, especially since no one will know how to pronounce the vowel sounds without learning them first (though that's easy enough). In my experience, the majority of the mistakes English speakers make when pronouncing Japanese are the vowels. Next is not separating morae correctly (like what you were talking about with Mitsubishi) Next is parsing things like 'kyo' as if it were 'ki-yo'.
The regular consonants in romanized Japanese ('r' excluded, of course) aren't mispronounced that badly, so I don't see a reason to change them to things that are completely unintuitive.
Assuming someone took five minutes to prime themselves on Japanese pronunciation, I think the better romanization system would still be the one with less things to remember: thus using familiar constructs for the consonant sounds.
But yeah, people will continue to call Nakamura as 'knack-uh-myur-uh' and Tokyo as 'toe-key-yo' if they don't have any knowledge of pronunciation.

I see absolutely zero purpose for romanization in the modern world other than to convey pronunciation to people who can't read Japanese. I see a few minor reasons in academic settings before the advent of UTF-8 and the ubiquity of fonts (it'd be a pain the butt to write a paper on Japanese in English without using romanization, for instance), but nothing today. That's why I keep stressing that it should be as intuitive to the audience as possible while maintaining a one-to-one transcription.

As far as I can tell, using nihon-shiki or kunrei-shiki only makes sense for the group of people who know Japanese (and thus know that 'du' doesn't sound like 'do' and 'zi' doesn't sound like 'ze' in 'zebra'), but aren't writing in Japanese... Which isn't a very common use case. The only other use case I know of is for official documents, and in that case, it's only being used because that's what they chose to use.

 

[Neushaartje]

@sholum

Again, 'well enough'. In English, gemination isn't much of a thing; it only shows up in certain cases of double consonants. It's workable.

Well, it isn't in Japanese either except with /s/; this notation arouses the impression that it should be geminated which is not how it's pronounced in Japanese, an apostrophe such as <tyo'to ma'te> could also work.
It arouses the impression that Japanese has actual geminated consonants such as, say, Finnish; this is even a popular myth amongst phonologists and they are often described as such.

In my experience, the majority of the mistakes English speakers make when pronouncing Japanese are the vowels. Next is not separating morae correctly (like what you were talking about with Mitsubishi) Next is parsing things like 'kyo' as if it were 'ki-yo'.

I think that might be influenced by your perception that is influenced by Hepburn. I very frequently see English speakers make mispronunciations that seem to arise from Hepburn such as pronouncing “全部” as “zenbu”, not as “dzembu”. Another common thing is the situation with <fu> — Japanese speakers will universally say that if one can't make the proper sound in between [ h] and [f], then simply using an [ h] is far præferred.

The regular consonants in romanized Japanese ('r' excluded, of course) aren't mispronounced that badly, so I don't see a reason to change them to things that are completely unintuitive.

I disagree, the /z/ is mispronounced almost always and English speakers don't seem to realize that the neutral pronunciation is [dz], and [z] is a variant that almost only occurs after a vowel — it is better to always use [dz] than never.

Assuming someone took five minutes to prime themselves on Japanese pronunciation, I think the better romanization system would still be the one with less things to remember: thus using familiar constructs for the consonant sounds.

Is it less to remember? I believe I've adequately demonstrated that <tya> and <sya> are closer to the actual sound than <cha> and <sha>, though it's not as if it can ever lead to communication problems as I doubt that JApanese speakers could even hear the difference between “tues” and “choose” and “suit” and “shoot”.

I see absolutely zero purpose for romanization in the modern world other than to convey pronunciation to people who can't read Japanese.

Then are you also opposed to “Muḫammad” rather than “Mohammed”? I think it's more meaningful to convey structure since they'll completely mispronounce it anyway, no matter what one do.

If one wish to convey pronunciation, one should use i.p.a..

As far as I can tell, using nihon-shiki or kunrei-shiki only makes sense for the group of people who know Japanese (and thus know that 'du' doesn't sound like 'do' and 'zi' doesn't sound like 'ze' in 'zebra'), but aren't writing in Japanese... Which isn't a very common use case. The only other use case I know of is for official documents, and in that case, it's only being used because that's what they chose to use.

See it thus: Certainly we can agree that there is merit for a unified global standard of romanized Japanese. It would be annoying if the same name were romanized differently depending on target language, which is actually what happens at times, for instance “Vladimir Putin” is traditionally romanized as “Poetin” in Dutch.

But not the entire world speaks English, so for that purpose it would perhaps even be better to use a Spanish-based system? I however think it would be better to use a system based on Japanese phonology itself — this is already what is happening in languages that are written in the Latin alphabet from the start. One would not know how to pronounce “Arnold Schwarzenegger” without a working knowledge of French, but that is how the name is written, not “Ahnolt Shvahtsenegah” or something like that, to communicate pronunciation.

I'm still not sold on the merit of communicating pronunciation since they mispronounce it anyway. I very commonly see the “e” in “Shinzo Abe” being silent, as if it were “Ape” but with a [ b] — so one might as well simply write “Sinzou Abe”.

I'm also not sure at all whether Hebpurn's judgements on what is closest to the pronunciation are accurate — remember that this was but one man who decided this, based primarily on Dutch and Portugese second-language speakers. We are now left with “Godzilla” because an English speaker heard “ゴジア” and heard “Godzilla”, not “Gojira” as Mr. Hepburn would prædict.

I'm not at all so convinced that English speakers when hearing “し“ and “じ” hear “shi” and “ji” when not primed for it — I've seen various examples when they not be primed that suggest they simply hear “si” and “(d)zi”.