How do I read pitch accent notation (alternatively, why does my pitch notation in yomichan and OJAD disagree)?

Question

I've installed this Anki plugin that adds pitch accent info. The author has helpfully provided some guidelines about the notation:

• Overline: Indicates "High" pitch
• Downfall arrow: usually means stressing the mora/syllable before.
• Red circle mark: Nasal pronunciation、e.g. げ would be a nasal け.
• Blue color: barely pronounced at all.

Now, I also gather from Kanshudo that

• 0 is heiban
• 1 is atamadaka
• 2 is nakadaka
• 3 is odaka

and Reddit tells me that

2 and later can't really be converted to nakadaka / odaka directly.

The 0, 1, 3, ... syntax just explain where the last high pitch mora is. Naturally, 0 is converted to Heiban because it has no "last" high pitch mora, and that's the exact definition of Heiban. It never goes low. 1 converts to atamadaka because it means that the last high pitch mora is the first mora, which is the definition of atamadaka - "head (or start) is high".

But as for nakadaka, the definition is just "the last high pitch mora is in the middle of the word". It could be 3 or 5 or anything really depending on how long the word is. It just can't be the first or the last mora.

The same goes for odaka.

Ok; now, for 中国 yomichan (and the Anki extension) tells me that the pitch accent is:

ちゅ[うごく]{HHH} [0]

So it's heiban? Then I listen to yomichan's audio, and think "wait, there's definitely a high-low dropoff after ちゅ".

So I go on OJAD and Japanese Accent Study Website and, lo and behold:

[ちゅうごく]{HHLLL} (actually, OJAD shows it just dropping off after ゅ, but I don't know how to markdown that)

And now I'm confused.

0. Do I have a bug somewhere in my setup?

1. Do I not understand how Anki and yomichan denote pitch accent? Do the linked resources actually agree?

   • Then, I'm asking for an explanation of how to read the accent notation used by the AJT plugin (and yomichan, I suppose)

Or,

2. Is Anki's and yomichan's reading wrong (despite the audio linked)? yomichan correctly identifies it as "China" (so it's the right word) and it's not some obscure vocabulary (so there have had to be other people spotting this)?

   • Then, is it correct to say that 中国 is actually atamadaka?

Answer 1

The problem you've encountered, albeit badly notated, is that the spelling and reading 中国【ちゅうごく】 is shared by two different terms.
(Definitions here as according to the Daijirin dictionary.)

• 中国【ちゅうごく】 - pitch pattern 0, 平板【へいばん】
  • The middle of the country. The area where the imperial court was historically located. The 畿内【きない】 area.
  • The 中国地方【ちゅうごくちほう】 (Chūgoku region).
  • An old name for the 山陽道【さんようどう】 region.
  • In the old Ritsuryō system, the third of four regions of Japan, as apportioned based on area and population.
• 中国【ちゅうごく】 - pitch pattern 1, 頭高【あたまだか】
  • China

So ultimately both pitch patterns are correct -- just for different meanings.

---

Note: Broadly, I've noticed that Chinese-derived terms tend to have 頭高【あたまだか】 pitch accent patterns. For terms like 中国【ちゅうごく】 that are inherently Chinese-derived, we see that the sense more closely tied to China still has the 頭高 pitch, while the more Japan-specific senses now use the 平板 pitch pattern.

---

Addendum

You'd also noted:

ちゅ[うごく]{HHH} [0]

So it's heiban? Then I listen to yomichan's audio, and think "wait, there's definitely a high-low dropoff after ちゅ".

I realized that there's a possible note of confusion here that I want to address.

In the 平板【へいばん】 pitch accent pattern, there is no downstep, which is why dictionaries might annotate this using the number zero, indicating a lack of downstep.

This is not to say that the pitch doesn't decline at all in 平板 words -- the speaker's pitch will rise after the first mora, and then gradually decline until the next word or phrase that has another high pitch in it. (Note: This is for "standard" Japanese, i.e. 標準語【ひょうじゅんご】 or Tokyo-based broadcaster-ese.)

For words with a downstep, meanwhile, numerical dictionary notation indicates the mora right after which the speaker's pitch drops more markedly.

Compare the following contrastive pair:

• 変【か】える: pitch pattern 0, 平板【へいばん】: [かえる]{LHH}
  Generally speaking, the え mora will have a slightly higher pitch than the る mora.
• 帰【かえ】る: pitch pattern 1, 頭高【あたまだか】: [かえる]{HLL}
  Generally speaking, the か mora will have a markedly higher pitch than the え mora.

In your term 中国【ちゅうごく】, the 平板【へいばん】 pronunciation will still show some slight decline in pitch after the rise on the う mora, so the ご and the く will be slightly lower. The 頭高【あたまだか】 pronunciation will show a marked decline in pitch after the ちゅ mora, much more of a drop in pitch than you hear between the high-pitched morae in the 平板【へいばん】 pronunciation.