Thanks for taking the time to explain this in detail. Much appreciated.
So I was wrong in thinking that it had something to do with the も + negative following the counter?
That is, it was heiban only because it was acting as an adverb, rather than as an item counter (い\ってん) or a standalone point score treated as a noun (いって\ん).
Wishing you a speedy recovery.
Yes, as far as I am aware. Words with も often do end up with heiban pitch like 誰も and 何も (even though some aren’t like いつも), but I think for counters its more just like a coincidence that when theres a も its usually being used in an adverbial way. As far as i’m aware usually this is just when its odaka or with a drop in the last syllable when it ends with ん. So its pretty interesting because the drop placement can change depending on numbers. Like a weird one is 人
六人
七人
八人
九人 (きゅうにん)
all have the drop in between the number and にん.
三人
四人
五人
九人（くにん)
all have the drop on the に. Thus those 4 get turned into heiban when adverbs and the other 4 don’t. Likewise we have
ひと↓り and ふたり↓. The latter becomes heiban when an adverb and the former doesn’t. Kinda complicated, but theres order to the chaos.
I thought the pitch never drops after an extended vowel sound. For 九人 (きゅうにん) wouldn’t it have to drop after きゅ or after に instead of after う?
Yes! Sorry, I mentioned 前部末型 earlier when I said
全部末型 as that counter which means it drops on the last mora of whatever comes before it save for general relucance for the drop to be on an ん or long vowel and stuff.
But forgot to explicitly mention what 前部末型 is which would make it apparent that falls in the same category. You would be shifting the accent one back for きゅうにん so its きゅ↓うにん. Same with じゅうにん which I didn’t mention. Thats just a general property of all pitch though and for 前末部型 the accent moves to the previous mora as you might expect.
Yeah, I got what you mean now. You just meant that it drops after the first mora before に that is allowed by the general rules of pitch accent.
Yeah. Compounds seem like a pain, and they kinda are, but one of their really nice rules are that if there is a drop in the first part it has to be as late as possible. And if theres not a drop in the first part, then the latter half can be heiban, atamadaka, or just keep its pitch which is things get a bit harder I guess.
Sorry, that’s rude to ask because I am sure you explained that somewhere else. It’s just I started to watch your video but don’t have 2 hours in the nearer future I can set aside to listen to the end. Where or how did you learn the basics of pitch accent?
I started getting serious about pitch accent after I made that video anyways lol. Well, I guess it depends on what your definition of basics are. Long ago I learned about the 4 general patterns and stuff which is like the super basics, but for things like:
I used Tips for Japanese Pitch Accent · GitHub
高低アクセント - Google Docs
and dogens pitch series.
Personally I think dogens pitch series is good for getting a general idea, but I thought it was pretty lackluster for getting a really good grasp like I wanted to (although partially thats just because I don’t think video format is great for memorizing rules). Its in english though and probably a lot more accessible. But first I went through the google doc, then I went through the git hub page. Then for general stuff that didn’t fit the model I had built in my head, I relied on google to correct some misunderstandings I had. Then nowadays I use the shinmeikai and nhk accent dictionary but thats probably going a bit beyond the basics. As far as general things are concerned, those two links should have everything you need.
(Don’t mind my stream-of-thought response.)
I guess this varies by the part of speech as well. For compound nouns I usually assume the pitch drops after the first mora of the second part unless I learn otherwise. That seems to be the most common, especially if it’s a longer word. I was just looking through that Google doc you sent before in the 複合名詞のアクセントの傾向 section to see if it explained that pattern or others, but frankly I’m unable to make sense of it. The leading sentence seems to confirm what I thought (that the most common pattern is to fall after the first mora of the second half, followed by the last mora of the first half which I hadn’t really considered). But the examples given are making less sense.
Specifically, I thought I understood what it was saying, but then I looked up the pitches for words using the examples they give and it contradicts my understanding of the section. For example, if the second half is 1-2 mora then most fall after the last mora of the first half. But then it gives ～感 as an example, and when I look up words ending in 感 most of them are heiban. But now thinking it through more, I suppose most of those words ending in 感 aren’t actually compound words. They are just words that happen to end with the kanji 感. So now there’s another issue, which is not knowing enough words to (consistently) recognize when a word is a compound word or a standalone word. So that’s fun… I can probably use a heuristic based on the first part’s length I guess. If it’s just one kanji / 1-2 mora then it’s probably not a compound word. If it’s 2+ kanji then it’s probably a compound. Seems to work for the 感 case at least. I’ll have to check out some words from those other examples in that line of the doc and see if I can generalize that heuristic.
Yes 感 is 前部末型 as you can see for words like 親近感, 緊張感, 嫌悪感, etc…
Yes 3-4 mora latter halves that aren’t nakadaka (which is a vast majority of them) is basically guaranteed to make it 後部一型. The longest latter half ones actually usually end up preserving its own pitch, though.
Assuming you’re referring to stuff like 穏やか, from what I’ve seen this pattern follows for nearly every な-adjective ending in か (excluding words that obviously don’t fit this pattern and just coincidentally end in か) and the few real exceptions should be easy to memorize. For example, 静か and 詳らか follow the same pattern of falling after the third mora from the end even though they aren’t the most common length of four mora and don’t have the や before か. (Seems like the Google doc mentions this one as well. I really should read the whole thing at some point.)
Yeah I don’t know why I typed やか rather than か lol, good catch. I don’t know if I’ve come across a real exception yet, but I have only been SRSing words I’ve been using really so there might be some I just haven’t seen yet. And fyi its 3-4 mora so しずか is still game
Thank you! Until now I was feeling a bit intimated to learn pitch accent because the whole idea to memorize every word from a dictionary seemed to be so overwhelming… But in any case it is good to know at least the basics. I don’t know why I can’t get myself to subscribe to Dogen, I even read his Japanese book.
Yeah tbf you definitely don’t have to go that far with it. I personally am, but I wouldn’t say I necessarily suggest other people to do the same. Its just I have a tendency to shoot for mastery and go hard when its something I enjoy, so really its more of a hobby at this point.
I would say just being aware of it and knowing general rules will have a good return on investment. And then if you do learn the pitch accents for words on a one by one basis, when you know all the rules surrounding them and sentences as a whole already, its really easy to integrate that knowledge into practice imo.
According to Yomichan, 些か can both follow pattern or be heiban for some reason. And 密か can also fall after the second mora besides following pattern. But since both of those can follow normal pattern, it seems fine to ignore the weird alternate options. However, 細か seems like a legitimate exception, as it can fall after the second or third mora but not after the first. Just found those by scanning the first three pages of a jisho search for *??か #adj-na.
I know しずか is still game (that was my point since I think some people probably associate this pattern with 4 mora words at first), though I would say it’s really 3-5 mora or even just 3+ mora. There are certainly much fewer 5 mora words that fit the pattern, but there are still enough to say it belongs I think. From a quick search:
And a handful more that I’d never seen before and didn’t feel like listing. I even saw one with 6 mora! (つつましやか)
They’re probably less weird than you think, actually.
If I’m not mistaken this one is a [-3] adjective. 些か just also happens to be an adverb though which I believe is 平板, hence the [0].
This ones also [-3], but devoicing naturally shifts the accent.
This one is more interesting because I couldn’t think of a reason so I searched real quick and found this
ことばの話2798「『細かな』のアクセント」
2年目のMアナウンサーが、お昼のニュースの前にした読みをしていました。
『NHK日本語発音アクセント辞典』 には、 「細かに」 が、
「コマカニ(LHLL)」
「コマカニ(LHHL)」
という 「中高アクセント」 が2つが載っていました。これに従うと、
「こまかな(LHLL)」
「こまかな(LHHL)」
になりますね。でも、私が入社した頃に教わったアクセントは、
「こまかな(HLLL)」
という 「頭高アクセント」だったのです。おかしい！もう時代遅れになったのか？ それとも、 そもそも間違ったアクセントを教わった のか？
念のため、もう一冊 『新明解日本語アクセント辞典』 を引いてみると、 「細か」 で載っていました。そこには、
「コマカ(LHL)」
「コマカ(LHH)」(尾高)
の2つのアクセントの後に、( )が付いて、
「古はコマカ(HLL)」
として「頭高アクセント」も載っているではなりませんか！
そうかやっぱり伝統的なアクセントだったんだな。アクセントも、もちろん時代とともに変わっていくんですねえ。いまさらながら感じました。
So it actually used to follow the pattern and shifted over time for some reason! The person writing it learned it as [-3] so theres definitely still people around who say it that way.
Ah, okay. Makes sense on both counts!
Even bigger mystery now! I wonder what could have caused a word like this to shift. Like you said, there’s no alternate part of speech or devoicing in this case to explain it away. So weird!
I wonder if it’s related to 細かい also being a word, but I don’t know. I can’t think of any other words off the top of my head that have both い-adjective and な-adjective versions. (The similar-ish situation with 大きな and 小さな wasn’t particularly enlightening after looking up their pitches.) I suppose if 細か was derived from 細かい it could potentially explain the exception, but that’s less convincing if that one resource is accurate in saying 細か used to follow pattern and actually changed over time. Looking at other examples of い/な-adjective pairs would help if you can think of any.
As of recently, if I’m not mistaken words moving away from atamadaka is a thing. Especially in favor of heiban. That, in combination with the pitch for 細かい might be enough for people to tend towards こま↓か or こまか↓ as of recently, but other than that I can’t think of anything. Maybe when I take a break from studying later I’ll take a look into it deeper.
柔らか and 柔らかい come to mind and sure enough the pitch accent says [-2] and [-1] on my yomichan dict but the audio is [-3]. This has audio for both and the [-3] sounds better imo
I finally managed to start the Phonetics lecture and it is a bigger blow than I was expecting.
Now I know that “the brain deteriorates quickly with age”, I might “sound like a mushroom farmer” and I could be a “non- native grammar Nazi” impossible to fix the bad habits I accumulated so far, what a charming man Dogen really is. 
Anyway, I am ready to take all offenses and start shadowing. Does anyone have a recommendation for an app that is useful for this purpose? And how do I know if I sound like a mushroom farmer?
A shadowing app?
Well, let me ask, what do you want to focus more on: pronunciation or pitch? I think for either one, the best thing you can do is recording yourself and listening to it back (and iirc dogen agrees), so just any old recorder app will be able to do that. Its just that you have to first develop a solid awareness of what things should sound like to do either. So really, its entirely possible that you could sound like a mushroom farmer and not even know it despite listening to yourself. I think just lots of listening helps with knowing what the proper pronunciation sounds like.
For pitch though, I feel like just listening isn’t usually enough. You kinda have to train your ears to hear it in my experience and get input while kinda being aware of it and know what the pitch of a sentence should be like in order to gauge how bad you are. The latter half of that can be taken care of by shadowing, but the first half can’t.
For shadowing though, you can use any normal audio playback app, but the important part is hearing your own mic feedback too. All apps are pretty bad for this because when it goes through your computer, theres like a lag that makes it really hard to speak and listen to yourself at the same time. For this reason, people usually get a mic with monitoring capabilities. I got this one
https://www.amazon.co.jp/SAMSON-マイク-ポータブル-USB-コンデンサ/dp/B001R76D42