Japanese Pitch Accent [split from Q&A]

A forum for discussing linguistics or just languages in general.
Salmoneus
MVP
MVP
Posts: 3030
Joined: 19 Sep 2011 19:37

Re: Japanese Pitch Accent [split from Q&A]

Post by Salmoneus »

Creyeditor wrote: 28 Jan 2021 20:03 I thought the arguments rests on the counterexamples with monomorphemic dipthongs accented on the unexpected vowels.
As I read it, Clawgrip made too arguments - one theoretical, one factual. The theoretical argument said that GrandPiano's analysis was misguided even if their facts were correct. The factual argument said that "then there's also the matter of..." possible counterexamples to GrandPiano's purported facts.

I have no opinion on the factual argument, as I don't speak Japanese - and it doesn't seem that interesting, other than for people really interested in Japanese. But the theoretical argument has a much broader application, and I did have an opinion about that.

I took these to be distinct arguments, because Clawgrip introduces the theoretical argument first, and only later adds, almost as an afterthought, "then there's also a matter of...", and introduces the factual argument. If Clawgrip only wanted to say that he'd thought of two possible counterexamples, then their post would presumably only have been that one sentence, since all the rest appears to explicitly make a theoretical argument.
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: Japanese Pitch Accent [split from Q&A]

Post by Creyeditor »

He can probably best answer it himself. I don't want to put words in anybody's mouth and I will patiently wait [:)]
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
clawgrip
MVP
MVP
Posts: 2257
Joined: 24 Jun 2012 07:33
Location: Tokyo

Re: Japanese Pitch Accent [split from Q&A]

Post by clawgrip »

Maybe my argument was flawed, so let me try explaining in a different way.

From what I read, GrandPiano's premise is that, of the 25 possible two-vowel sequences in Japanese, 21 (or 22) are to be considered "as two monophthongs in hiatus", while the remaining four (or three) (/ai/, /ae/, and /oi/, and possibly /ui/) are to be considered diphthongs (If long vowels/sequences of like vowels are to be excluded from this premise, then the numbers of monophthongs in hiatus is 16 (or 17)).

This premise is supported by this evidence: "it seems like the vowel sequences ai, ae, and oi (maybe also ui?) only permit the accent on the first mora when they occur within a single morpheme, and other vowel sequences allow the accent on either mora."

Counterexamples of these sequences, such as 老いる oìru and 考える kangaèru are presented and disqualified on the grounds that the two vowels belong to different morphemes and thus are not applicable.

I am not sure about the final conclusion, but it seems that there are two possibilities:
a. all instances of these vowel sequences are to be considered diphthongs, or;
b. all instances within a single morpheme are to be considered diphthongs

In the case of a., why would a conclusion whose evidence is based entirely on excluding counterexamples comprising multimorphemic sequences thus apply to multimorphemic sequences? If, for example, we conclude that /oi/ in オイル òiru is a diphthong because /oi/ in 老いる oìru is actually two monophthongs in hiatus and thus not comparable, then by what reasoning can we consider 老いる oìru to contain a diphthong?

In the case of b., if we conclude that /oi/ in オイル òiru is a diphthong, while /oi/ in 老いる oìru is not a diphthong but rather two monophthongs in hiatus on account of the two vowels occurring in different morphemes, then by what reasoning can we consider /oi/ in オイル òiru to be a diphthong, but not the /oi/ 黒い kuròi, (in which /o/ and /i/ belong to different morphemes) for which the phonological realization is entirely identical in all respects (vowel quality, vowel length, pitch)?

Looking at Japanese vocabulary in general, when an accent occurs on one of two monophthongs occurring in sequence, that accent is much more likely to appear on the first vowel than the second. An accent on the second vowel is comparatively rare, and consequently, the absence of examples of certain sequences is not particularly conspicuous. With that in mind, I think that an absence of examples is much more cleanly explained as simply that, an absence of examples, rather than by introducing an entirely new phonological feature. This is why I considered it far-fetched. With the addition of the other counterexamples I provided, the premise becomes even more untenable.
GrandPiano
mayan
mayan
Posts: 2080
Joined: 11 Jan 2015 23:22
Location: USA

Re: Japanese Pitch Accent

Post by GrandPiano »

clawgrip wrote: 26 Jan 2021 13:06Respectfully, I find your reasoning doubtful. You claim that "ai and ae only allow the accent on the first mora," and invalidate verbs that violate this premise by claiming that "the -eru and -iru of ichidan verbs are treated as morphemes." You then use this as grounds to "analyze /ai/, /ae/, and /oi/ as diphthongs and other sequences as two monophthongs in hiatus."
I think you're misunderstanding the basis of my argument. The primary basis of my argument is verbs like 帰る káeru and 入る háiru, which have the accent on the third-to-last mora even though they belong to an accent category that normally places the accent on the second-to-last mora in the dictionary form (and also noticing that this only happens with certain vowel sequences, e.g. 倒す taósu has the expected accent placement). To me, the natural explanation for this is that accent placement is actually based on syllables, not moras, and these verbs have two syllables. It also seems to me that this analysis can explain an apparent difference in accentual behavior between godan and ichidan verbs: accented godan verbs with endings like -airu, -aeru, and -ooru have the accent on the third-to-last mora, but accented ichidan verbs with endings like -aeru, -oiru, and -iiru have the accent on the second-to-last mora. Under this analysis, a simple explanation is that the suffix in the dictionary form of an ichidan verb is actually -eru/-iru, not -ru, and so there is a morpheme boundary between the two adjacent vowels (I suppose this would also synchronically explain why ichidan verbs can only end in -eru or -iru).
clawgrip wrote: 26 Jan 2021 13:06I think that if a dependent morpheme has phonetically merged with the verb stem such that the two adjacent vowels are now to be considered a single diphthong rather than a sequence of two monophthongs in hiatus, then the original grounds, that "-eru and -iru of ichidan verbs are treated as morphemes ... for the purpose of accent assignment," would no longer seem to be valid. Namely, I would argue that if two morphemes have merged into a single phonological unit, then the now-phonetically irrelevant morpheme boundary can no longer be used to determine anything about the phonological composition of that unit.
Phonetically merged how? In any case, my argument is about phonology, not phonetics.
clawgrip wrote: 26 Jan 2021 13:06Then there's also the matter of words like マイカー maìkā, 喘ぐ aègu, and ツイン tsuìn, which also run against this theory, as the two vowels are within the same morpheme.
Admittedly, I wasn't aware of 喘ぐ. I suppose it's the same sort of situation as 通る tóoru vs. 催す moyoósu. The only solution I can think of is to analyze 通る as /toː.ɾu/ but 催す as /mo.jo.o.su/, and similar to analyze 帰る /kae.ɾu/ but 喘ぐ as /a.e.gu/, though I recognize that this isn't a very satisfying explanation. Still, if you disagree with my analysis, I'm curious how you explain the accentual behavior I described above.

What's your source on the accent for マイカー? I checked the NHK accent dictionary, and it says that マイカー is maikáa. I don't think ツイン is necessarily a counterexample. I wasn't able to think of an example for /ui/, so it could be that /ui/ just isn't one of the sequences that shows this behavior.
clawgrip wrote: 26 Jan 2021 13:06When considering the diversity of Japanese dialects, which vary widely in their pitch patterns and conjugations, it seems a little silly to insist that single morphemes that have remained in the same place for at least 1,200 years are preventing the alteration of an accent in some way.
I'm not sure I understand your point here. I don't think dialects are relevant, because this is only about Standard Japanese, and I'm not quite sure I see why "remaining in the same place for at least 1,200 years" would affect whether a morpheme can be relevant to accent placement.
clawgrip wrote: 29 Jan 2021 01:26From what I read, GrandPiano's premise is that, of the 25 possible two-vowel sequences in Japanese, 21 (or 22) are to be considered "as two monophthongs in hiatus", while the remaining four (or three) (/ai/, /ae/, and /oi/, and possibly /ui/) are to be considered diphthongs (If long vowels/sequences of like vowels are to be excluded from this premise, then the numbers of monophthongs in hiatus is 16 (or 17)).

This premise is supported by this evidence: "it seems like the vowel sequences ai, ae, and oi (maybe also ui?) only permit the accent on the first mora when they occur within a single morpheme, and other vowel sequences allow the accent on either mora."

Counterexamples of these sequences, such as 老いる oìru and 考える kangaèru are presented and disqualified on the grounds that the two vowels belong to different morphemes and thus are not applicable.
I think you may have misunderstood the premise of my argument, but hopefully I explained myself a little better above.
clawgrip wrote: 29 Jan 2021 01:26I am not sure about the final conclusion, but it seems that there are two possibilities:
a. all instances of these vowel sequences are to be considered diphthongs, or;
b. all instances within a single morpheme are to be considered diphthongs
b. was essentially my conclusion, although it seems like there are a few words like 喘ぐ and 催す where sequences like /a.e/ and /o.o/ occur within a morpheme as sequences of monophthongs rather than diphthongs/long vowels, as I described above.
clawgrip wrote: 29 Jan 2021 01:26In the case of b., if we conclude that /oi/ in オイル òiru is a diphthong, while /oi/ in 老いる oìru is not a diphthong but rather two monophthongs in hiatus on account of the two vowels occurring in different morphemes, then by what reasoning can we consider /oi/ in オイル òiru to be a diphthong, but not the /oi/ 黒い kuròi, (in which /o/ and /i/ belong to different morphemes) for which the phonological realization is entirely identical in all respects (vowel quality, vowel length, pitch)?
I wonder if there's some confusion here about what is meant by "diphthong"? I think of a diphthong as being a phonological entity, not a phonetic entity. As I understand it, a diphthong is a sequence of two phonetic vowels which phonologically behave as a single unit. There isn't necessarily a phonetic difference between a diphthong and two monophthongs in hiatus (there might be a difference within a given language, but I don't think there's a universal difference).

For オイル vs. 黒い, I suppose one could argue that there's evidence from other words that diphthongs occur within morpheme boundaries but not across morpheme boundaries, and so it follows that オイル contains a diphthong but 黒い does not.
LinguistCat wrote: 18 Dec 2020 22:51And I'm not even sure what it would mean for someone to claim there's a difference between an accent before a sokuon vs after it (aside from maybe /s:/), though I have seen some resources differentiate, for example, ma'tte vs asat'te.
Unrelated, but I was looking back through the thread and I noticed this. I've seen that accent for asatte in some places as well, and I suspect that it's a mistake. The NHK accent dictionary gives asátte. As far as I know, the sokuon can never carry the accent.
clawgrip
MVP
MVP
Posts: 2257
Joined: 24 Jun 2012 07:33
Location: Tokyo

Re: Japanese Pitch Accent

Post by clawgrip »

GrandPiano wrote: 30 Jan 2021 02:39 I think you're misunderstanding the basis of my argument. The primary basis of my argument is verbs like 帰る káeru and 入る háiru, which have the accent on the third-to-last mora even though they belong to an accent category that normally places the accent on the second-to-last mora in the dictionary form (and also noticing that this only happens with certain vowel sequences, e.g. 倒す taósu has the expected accent placement).
It does appear that somehow I missed the part of your argument where you mentioned the verbs having either no accent or having an accent on the penultimate mora of the final morpheme. I apologize for rambling on and on without having read that portion clearly.
To me, the natural explanation for this is that accent placement is actually based on syllables, not moras, and these verbs have two syllables. It also seems to me that this analysis can explain an apparent difference in accentual behavior between godan and ichidan verbs: accented godan verbs with endings like -airu, -aeru, and -ooru have the accent on the third-to-last mora, but accented ichidan verbs with endings like -aeru, -oiru, and -iiru have the accent on the second-to-last mora. Under this analysis, a simple explanation is that the suffix in the dictionary form of an ichidan verb is actually -eru/-iru, not -ru, and so there is a morpheme boundary between the two adjacent vowels (I suppose this would also synchronically explain why ichidan verbs can only end in -eru or -iru).
clawgrip wrote: 26 Jan 2021 13:06Then there's also the matter of words like マイカー maìkā, 喘ぐ aègu, and ツイン tsuìn, which also run against this theory, as the two vowels are within the same morpheme.
Admittedly, I wasn't aware of 喘ぐ. I suppose it's the same sort of situation as 通る tóoru vs. 催す moyoósu. The only solution I can think of is to analyze 通る as /toː.ɾu/ but 催す as /mo.jo.o.su/, and similar to analyze 帰る /kae.ɾu/ but 喘ぐ as /a.e.gu/, though I recognize that this isn't a very satisfying explanation.
This explanation is begging the question. The premise that 喘ぐ aégu must have a morpheme boundary between /a/ and /e/ while 帰る káeru need not is only valid if we have already accepted the conclusion that /áe/ can occur in a single morpheme while /aé/ cannot. Therefore, it cannot stand as evidence of that conclusion. (I'll get to toosu/moyoosu below)
Still, if you disagree with my analysis, I'm curious how you explain the accentual behavior I described above.
My main issue is that you appear to be suggesting a complete change to Japanese phonology based on an irregularity found in a handful of words at most, and it may not even be an irregularity at all.

First off, I want to say that after rereading your posts, I will go back on another thing I said: namely, I will accept that verb suffixes can affect the accent pattern in the way you say. But let's look at our examples:

倒す taósu "to throw down"
This comes from the Classical Japanese 倒す tafusu, a compound of ta (not sure what this is) plus 伏す/臥す fusu "to lie down". So this is not a single morpheme. The fu became o instead of expected u probably to dissimilate it from (because tau would almost certainly contract to ).

入る háiru "to enter"
This word does not appear to exist in Classical Japanese; instead, 入る iru was used. Although I am having difficulty finding an etymology for hairu, it's almost certainly a compound of some root ha with iru.
However, 参る máiru can be traced to to Old Japanese mawiru, so I think the stem is a single morpheme.

帰る káeru "to return"
This can be traced back to Old Japanese kaperu, but I can't find anything else, so it is likely a single morpheme. It does bear a resemblance in sound and meaning (go back to a previous place/state) to 変える/換える/代える/etc. kaeru "change; replace; substitute" but I found no link in the bit of research I did.

通る tóoru "to go past"
Like kaeru above, this can be traced back to Old Japanese as toporu, so I believe the stem to be a single morpheme.
There's also the word 遠い tooi, which seems to allow for varying pronunciations, including toói. I also have to wonder though if these are the same root, since the stems are identical aside from accent, and "far" and "pass through" seem like they could be somewhat semantically related.

Here are the counterexamples presented:
催す moyoósu does appear to be across two morphemes. Although rare, 催い moyoi does appear to be a word, or at least a suffix (a quick search shows that it seems primarily to be appended to 雨 ame, 雪 yuki, and 船 fune). Therefore, this counterexample is invalidated.

喘ぐ aégu can be found in Classical Japanese as 喘く ahaku/aheku, with the aha/ahe stem being an onomatopoeic representation of a panting sound. This is clearly a single morpheme.

ツイン tsuìn is borrowed directly from English twin and is obviously a single morpheme.

What's your source on the accent for マイカー? I checked the NHK accent dictionary, and it says that マイカー is maikáa. I don't think ツイン is necessarily a counterexample. I wasn't able to think of an example for /ui/, so it could be that /ui/ just isn't one of the sequences that shows this behavior.
My source is actual usage, but here is another source. But anyway, I searched for マイカー on YouTube and checked the first result, and actually found something unexpectedly interesting.
The first time she says マイカー (at 0:05) she pronounces it with the pattern you suggest (maikáa), but every subsequent time she pronounces it with the pattern I suggest (maìkaa) (0:18, 0:24, 0:55, 0:59, maybe more but I got tired of watching the video at around the halfway point). The first instance pronounced maikáa is probably to introduce the topic, as putting the accent on kaa makes sense because that's the thing actually being discussed. But on subsequent mentions she reverts back to maìkaa, which in my experience is the normal way to say it.

Also, notice that she says sheàkaa "share car" and rentàkaa "rent-a-car", which have the exact same pattern, placing the accent on the mora immediately before kaa. There are many other words with the same pattern, such as ダンプカー danpùkaa "dumptruck", エコカー ekòkaa "eco-friendly car", コンパクトカー konpakutòkaa "compact car", スポーツカー supootsùkaa, etc. etc., so the fact that the accent falls on the i of mai is strong evidence that it is not a diphthong.

So the evidence káeru has counterevidence aégu;
The evidence máiru has counterevidence maíkaa;
The evidence tóoru has counterevidence toói (though admittedly as a pronunciation variant)
clawgrip wrote: 26 Jan 2021 13:06When considering the diversity of Japanese dialects, which vary widely in their pitch patterns and conjugations, it seems a little silly to insist that single morphemes that have remained in the same place for at least 1,200 years are preventing the alteration of an accent in some way.
I'm not sure I understand your point here. I don't think dialects are relevant, because this is only about Standard Japanese, and I'm not quite sure I see why "remaining in the same place for at least 1,200 years" would affect whether a morpheme can be relevant to accent placement.
I recant on the morpheme thing, and the dialect thing made a bit more sense to me when I was arguing against something that you never actually posited.
clawgrip wrote: 29 Jan 2021 01:26In the case of b., if we conclude that /oi/ in オイル òiru is a diphthong, while /oi/ in 老いる oìru is not a diphthong but rather two monophthongs in hiatus on account of the two vowels occurring in different morphemes, then by what reasoning can we consider /oi/ in オイル òiru to be a diphthong, but not the /oi/ 黒い kuròi, (in which /o/ and /i/ belong to different morphemes) for which the phonological realization is entirely identical in all respects (vowel quality, vowel length, pitch)?
I wonder if there's some confusion here about what is meant by "diphthong"? I think of a diphthong as being a phonological entity, not a phonetic entity. As I understand it, a diphthong is a sequence of two phonetic vowels which phonologically behave as a single unit. There isn't necessarily a phonetic difference between a diphthong and two monophthongs in hiatus (there might be a difference within a given language, but I don't think there's a universal difference).

For オイル vs. 黒い, I suppose one could argue that there's evidence from other words that diphthongs occur within morpheme boundaries but not across morpheme boundaries, and so it follows that オイル contains a diphthong but 黒い does not.
I recognize that morphology can influence phonology, but phonology itself is meant to describe the resultant material. Even if two different morphological processes produce the same result, that result is still a single phonological product. I'm not convinced that there is a way to describe in purely phonological terms why /ói/ is a diphthong while /ói/ is a sequence of monophthongs.
LinguistCat wrote: 18 Dec 2020 22:51And I'm not even sure what it would mean for someone to claim there's a difference between an accent before a sokuon vs after it (aside from maybe /s:/), though I have seen some resources differentiate, for example, ma'tte vs asat'te.
Unrelated, but I was looking back through the thread and I noticed this. I've seen that accent for asatte in some places as well, and I suspect that it's a mistake. The NHK accent dictionary gives asátte. As far as I know, the sokuon can never carry the accent.
As far as I know, you're correct that sokuon never carries an accent.

Also I hope you recognize that there is no hostility in what I say. I wouldn't spend this much time looking up and writing all this stuff if I didn't think there was merit in discussing it.
User avatar
Sequor
sinic
sinic
Posts: 438
Joined: 30 Jun 2012 06:13

Re: Japanese Pitch Accent [split from Q&A]

Post by Sequor »

I'm just going to say I'm enjoying this exchange so far, in spite of not quite fully understanding it, largely due to my shaky grasp of Japanese.

I'd just like to gently prod you guys about a question I asked earlier on, if any of you can say anything about it:
While you're here, are you aware of any known instances where (Middle or modern) Chinese tone was adapted in Japanese with a certain pitch accent pattern that matched it in some way?
hīc sunt linguificēs. hēr bēoþ tungemakeras.
clawgrip
MVP
MVP
Posts: 2257
Joined: 24 Jun 2012 07:33
Location: Tokyo

Re: Japanese Pitch Accent [split from Q&A]

Post by clawgrip »

Sorry for taking so long to get back to you. To be honest, I know of no such examples, but I am not particularly knowledgeable about Chinese tones. The Chinese readings of individual characters themselves in Japanese do not carry pitch information unless the Chinese reading of a character can stand as an independent word. So e.g. 菜 sai bears no specific accent, but in the word 野菜 yasai it has a heibon accent (i.e. no accent).

Chinese cities read in Japanese tend to have the exact same accent pattern, even if the characters are pronounced with nonstandard readings to match the Chinese pronunciation more, and this even shifts if you add -shi "city", meaning the Chinese tone is entirely ignored:

武漢 Bùkan, 武漢市 Bukàn-shi "Wuhan"
北京 Pèkin, 北京市 Pekìn-shi "Beijing"

Looking at some relatively recent food borrowings, some with non-standard readings to match the Chinese pronunciation more:
焼売 shāomài - shūmai "shumai (dumpling)"
小籠包 xiǎolóngbāo - shōrònpō "xiaolongbao (steamed bun)"
炒飯 chǎofàn - chā̀han "fried rice"
八宝菜 bābǎocài - happṑsai "babaocai (type of stir fry)"

There appears to be no correlation, and a strong preference for placing the accent on the penultimate character.

I can't say for sure whether there are any words influenced by the original Chinese tones, but considering the examples above, the fact that the characters were borrowed over a millennium ago, as well as the fact that Japanese people are about as good at learning Chinese tones as English speakers, it seems quite unlikely.
User avatar
Sequor
sinic
sinic
Posts: 438
Joined: 30 Jun 2012 06:13

Re: Japanese Pitch Accent [split from Q&A]

Post by Sequor »

clawgrip wrote: 15 Feb 2021 23:55I can't say for sure whether there are any words influenced by the original Chinese tones, but considering the examples above, the fact that the characters were borrowed over a millennium ago, as well as the fact that Japanese people are about as good at learning Chinese tones as English speakers, it seems quite unlikely.
All of that was very interesting. Thanks!
hīc sunt linguificēs. hēr bēoþ tungemakeras.
Post Reply