I was reading a tanka which appears in the Man'yōshū (my translation):

When I eat a melon, thoughts of my children come to my mind.
When I eat a chestnut, thoughts of them come unbidden to me even more.
From where did they spring?
They flit before my eyes, and grant me no sleep.

My question is, how much awareness do native speakers have of the pitch accent of archaic verbs, for example 食む "to eat" or 悔ゆ "to regret" or 増ゆ "to be increased"? I hear recordings of similar poems by native Japanese speakers, but from where do they deduce the pitch accents of archaic verbs?

Perhaps, in cases where the Classical Japanese verb has a modern reflex, are there certain implicit rules that can be applied — for example, if [増える]{LHL} exists in Modern Japanese, do native speakers intuit that the pitch accent should be [増ゆ]{HL}?

Is Classical Japanese pitch accent even agreed upon in the Japanese linguistics literature?


2 Answers 2


Well, I understand the question to be in two parts.

(1) Whether the actual pitch of Old Japanese verbs is known (to researchers);

(2) How do modern Japanese people apply modern pitch when reading Old or Classical texts.

These are properly two different questions, so I will try to deal with them separately.

(1) The historical pitch is really well-known from the tone marks assigned to many words in dictionaries and texts. The go-to reference here is Elisabeth de Boer's The Historical Development of Japanese Tone (2 Vols., Harrassowitz Verlag, 2010).

Usually, in literature dedicated to Japanese tones only nouns are considered, but the tonal marks occurred on verbs as well, and there are listings of those.

For example, in Samuel E. Martin The Japanese Language Through Time (Yale University Press, 1987), which is the most accessible place actually giving historical tones, on pages 192-197 there is the complete chart of all verbs marked with tonal marks in the Ruiju Myōgishō dictionary (12th cent.), in several forms. (Note that Martin assumed the tonal marks to denote the opposite of how they are treated now, so read all, say 'HL' as 'LH' and vice versa.) For example, from Ruiju Myōgishō we know that the verb 置く 'to put' has final form [おく]{LL}, coordinate converb [おきて]{LHL} without onbin or [おいて]{LHL}, [おいて]{LHH} with one, infinitive (converb) [おき]{LH} and deverbal noun [おき]{LL}.

Of course, all of the aforementioned refers to the period of tone marks, which is late 10th to 14th century, corresponding to Early Middle Japanese; however, it is not assumed that Old Japanese of the 萬葉集 was drastically different in this regard.

(2) The pronunciation of Classical verbs in modern, Tokyo, pronunciation is a separate matter, more prescriptive than descriptive. For this, there are norms, and dictionaries actually explain that.

アクセント習得法則, the preface to the Shinmeikai accent dictionary, actually has Table 2 of Section 42 to help making the Tokyo pitch for a Classical correspondence to a current verb. In general, most forms are made as in current language, as for the unique forms:

  • The final (終止形) and attributive (連体形) forms are unstressed or stressed according to the verb (泣く is unstressed in modern, so [なく]{LH}, [なくもの ]{LHHHL}; 起きる is stressed in modern, so [おく]{HL}, [おくるもの]{LHLLL}). Post-final らん is stressed on ら if the verb is unstressed, allows the stress on the verb otherwise ([なくらん]{LHHL}, but [おくらん]{HLLL}).

  • The a-form (未然形) is as follows: negation ず does not enforce any stress ([なかず]{LHH}, [おきず]{HLL}), same with its attributive -ぬ; conjectural -ん (-む) enforces the stress before it, overriding all else ([なかん]{LHL}, [おきん]{LHL}); conditional -(a)ば makes stress before it if unstressed ([なかば]{LHL}) but gives way to existing stress ([おきば]{HLL}).

  • Coordinate converb -て enforces stress before it only on unstressed ones ([なきて]{LHL}, but [おきて]{HLL}; the dictionary also says there is a tendency to say the -て-form of unstressed verbs unstressed as [なきて]{LHH}); perfective -ぬ enforces stress before ([なきぬ]{LHL}, [おきぬ]{LHL}); -たり enforces stress before it only on unstressed ones ([なきたり]{LHLL}, but [おきたり]{HLLL}) - here also [なきたり]{LHHL} is possible, with stress on た in case of unstressed verbs.

  • As for the 已然形 form, -ども enforces stress before it only on unstressed ones ([なけども]{LHLL}, but [おきども]{HLLL}).

  • Finally, in the imperative, the unstressed verbs get final stress (or before -よ, if present: [なけ ]{LHL}, [なけよ]{LHL}), the others retain stress ([おき]{HL}, [おきよ]{HLL}).

  • This is precisely the answer I was looking for, and well-sourced besides.
    – jogloran
    Commented Oct 30, 2022 at 16:39
  • Do you know what evidence caused the consensus to switch regarding the interpretation of the tone marks in the Ruiju Myōgishō?
    – jogloran
    Commented Oct 30, 2022 at 17:05
  • 1
    The whole length of de Boer's book is dedicated to that: Vol. 1 is modern dialects, Vol. 2 is historical texts. The book is actually available freely on the web for perusal. Commented Oct 30, 2022 at 18:00

For ordinary people, it's mostly a guesswork. What I think most would do is to find a similar sounding word in the modern vocabulary and borrow its accent pattern (which might or might not be accurate). For example, はむ can be pronounced like かむ (噛む).

There are NHK programs in which broadcasters read classics. One example is 古典購読. I believe their accent doesn't sound too strange to most native speakers.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .