The dictionary form of Japanese verbs always ends in a -u syllable. Ichidan (one row or single-step in German) verbs will always end in -る (-ru, e.g. 食べる, taberu, to eat) while godan (five rows or five-step in German) verbs can end in any of the following:

  • -う (-u) as in 会う (au, to meet)
  • -く (-ku) as in 書く (kaku, to write)
  • -ぐ (-gu) as in 泳ぐ (oyogu, to swim)
  • -す (-su) as in 話す (hanasu, to speak)
  • -つ (-tsu) as in 立つ (tatsu, to stand)
  • -ぬ (-nu) as only (!) in 死ぬ (shinu, to die)
  • -ぶ (-bu) as in 遊ぶ (asobu, to play)
  • -む (-mu) as in 読む (yomu, to read); or
  • -る (-ru) as in 有る (aru, to exist/to be/to have)

It is notable that -う (-u) is actually from -wu, which can only be seen in the (w)a-form (used e.g. in negation 会わない rather than *会あない) as all other w- syllables have merged with their respective Ø- forms. Furthermore, it is interesting that among voiced/unvoiced pairs only -く (-ku) and -ぐ (-gu) coexist as verb endings; -ず (-zu) and -づ (-dzu) do not form verbs to the best of my knowledge and neither do -ふ (-fu) or -ぷ (pu). Verbs ending in -ゆ (-yu) or an -う (-u) which is not actually a historic -wu seem not to exist at all. Finally, according to my grammar book there is only a single verb that ends in -ぬ.

What are the origins of this perceived imbalance? How come only one voiced/unvoiced pair exists in contemporary Japanese and are there any mechanisms that explain the absence of the other missing syllables e.g. by mergers? And why is it -ぶ but -つ and -す?

  • 5
    The restriction you mention was already in place in Old Japanese, whose yodan verb roots could only end in /p, t, k, b, g, m, s and r/ (p-roots eventually becoming w-roots). 死 /sin-/ was an irregular verb that merged into the yodan class when its dictionary form was levelled from /sinuru/ to /sinu/ in late Middle Japanese. Other irregular consonant roots ended in /k, r and s/, patterning with the yodan class. The cause of this restriction might be found in Proto-Japonic reconstructions, but my cursory search didn't find anything Nov 6, 2019 at 23:04

2 Answers 2


Japanese verb roots show a strong tendency to be native rather than borrowed vocabulary. In native Japanese vocabulary, original *p was lenited to [w] between vowels (based on the historical spelling of syllables with this [w] sound and the alternative development of *p to a voiceless fricative in word or morpheme-initial position, this medial [w] from *p seems to have passed through an intermediate stage of a voiceless labial fricative [ɸ]).

Because of these sound changes (*p > ɸ followed by VɸV > VwV), there is no regular source of intervocalic /p/, /h/ or /ɸ/ in native Japanese vocabulary (although these sounds do show up intervocalically in some kinds of borrowed words, reduplicated mimetic words, or compound words). I think that explains fairly well the absence of -fu and -pu verbs.

I don’t know whether there are explanations for the other gaps that you mention. I remember reading that some Japanese verbs have switched classes between time periods: I don’t know whether that is relevant.


I'm not qualified to comment on the rest, but I do want to note that both -zu and -nu are verb conjugations in classical Japanese. For ichidan verbs, both these forms are formed by replacing the -ru with -zu/nu:

  • 見る miru "to see"
  • 見ず mizu "without seeing" (modern Japanese 見なくて minakute)
  • 見ぬ minu "not seeing" (modern Japanese 見ない minai)

As you can see, if -nu or -zu verbs were common, they would easily be confused with the negated forms of ichidan verbs. Godan verbs, though, are conjugated the same way as -nai, so there's less potention for confusion:

  • 死ぬ shinu "to die"
  • 死なず shinazu "without dying" (modern Japanese 死なないで shinanaide)
  • 死なぬ shinanu "not dying" (modern Japanese 死なない shinanai)
  • I don’t subscribe to your bolded sentence. Chance overlaps between different grammatical forms are common and not a problem because context exists and because different inflections are acknowledged in our brains. For example the French verbs dormir and finir each have a form where the final r is replaced by t: il dormit versus il finit. However, the first is past tense and the second is either present or past tense. (Present tense of dormir is il dort.) Likewise, Japanese are fine distinguishing between 来る、切る and 着る even though their forms overlap.
    – Jan
    Dec 3, 2019 at 12:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.