Japanese is famous for its very few irregular verbs, but there are some cases where verb-forms are missing and other verbs/adjectives are used instead.

For example, (in standard Japanese) the verb ある "aru" (to be/exist) does not have a non-polite negative (which would be あらない "aranai") and instead the adjective ない "nai" is used. Still, ある is usually considered regular.

Why is that? Is it just a matter of definition, or is there some reason that ある cannot be considered an irregular verb which has undergone suppletion?

Another example is する "suru" (to do), which does not have a potential (can do). In stead, the verb できる "dekiru" is used. (する is already irregular, though.)

2 Answers 2


I agree with jlovegren. that it probably has to do with definitions — in fact, I have seen aru called an irregular verb. On the other hand, I hazily remember reading something by Kindaichi Haruhiko which contrasted the behavior of aru with "the adjective nai", so, clearly, there are mainstream voices on the "not irregular" side too.

To judge from Alex B.'s response, there is no "sharp line", so the question is, which explanation seems more probable?

Let's say we declare nai part of the aru paradigm. This leaves us with the problem of what to do about the independent adjective nai. If our argument is that it no longer exists, we are left with the difficulty of proving this. For example, nai clearly existed independently of aru in old expressions like (1) below.

1) Naki ni shi mo arazu

Now consider a fairly direct modern translation of (1):

2) Nai (wake) de wa nai

What can we offer as proof that the first nai in (2) must be considered a conjugation of aru? I can't think of anything. (I'm sure if I am wrong someone will let me know in comments.)

So let's say that nai still exists as an independent adjective. In that case, we have the opposite problem. What is our proof that the final word of (2) must be considered a conjugation of aru? We can offer the polite form (3):

3) Nai (wake) de wa arimasen

... where the link to aru is clear -- but in fact, the following is also acceptable:

4) Nai (wake) de wa nai desu

... which is what we would expect to see if the final word in (2) is "adjective nai" rather than "conjugation-of-aru *nai*": if we have no reason to go back to an aru-related form, we create a polite form just like any other adjective.

I'm sure some people would argue that (4) is not grammatical, but then we are back to definitions. It is undeniable that nai desu is very common in casual speech. It cannot be ruled unacceptable except by fiat. And note that we are also seeing forms like nakatta desu encroach on what "should" be arimasen deshita, another example of nai acting more like an independent adjective than something that is "underlyingly" a form of aru.

It seems to me that the most logical explanation is to view aru as a defective verb, or at least a verb with a nonpolite negative form that is "suppressed" by the use of independent-adjective nai instead, because:

  • It frees us of the need to make the weak, apparently arbitrary distinction between adjective and negative-of-verb required by examples like (2)
  • It explains why nai-like forms are also encroaching on the polite negative form of aru as seen in (4): because the word is increasingly not perceived as being part of aru

One argument against my second point is that we see forms like ikanakatta desu (instead of ikimasen deshita) as well. Maybe desu is just turning into a non-conjugating "politeness particle", and so the encroachment on arimas- is part of a bigger change involving the loss of -mas- forms, rather than something specific to aru. (Counterarguments to that might include "Yeah, but this is itself driven by the influence of the adjective nai on the negative verb ending (or "auxiliary", etc.) -nai, so ultimately it comes down to the triumph of adjective nai anyway".)

  • Thanks a lot for your elaborate answer. Very interesting reading. However, I am not sure I understand all your arguments. You mention "nai desu", but to me, the fact that "nai desu" is semantically equivalent (some might disagree) with "arimasen" proves that "nai" is the negative of "aru" (at least semantically). I.e. arimasu vs. arimasen = arimasu vs. nai desu = aru vs. nai.
    – dainichi
    Commented Feb 15, 2012 at 6:58
  • Cont'd: Also, even if "nai" also exists as an independent adjective, I don't think that is an argument not to treat it as a negative of "aru". For example, the preterite of verbs "ser" (be) and "ir" (go) in Castillian Spanish are both "fui" (etc) with obvious suppletion. So the fact that e.g. "dekiru" has other meanings (e.g. nikibi ga dekiru), should not stop it from being a potential of "suru". And yes, "nai desu" is not specific to "aru", but not because "nai" is gaining ground, but because non-polite negative + desu is gaining ground.
    – dainichi
    Commented Feb 15, 2012 at 6:58
  • The issue with multiple instances of "nai" isn't that they couldn't coexist but rather that there doesn't seem to be a way of telling them apart. Essentially, if we can't show that a given "nai" must be an inflection of the verb "aru" as opposed to the adjective "nai", then what basis do we have for insisting that any are? (Excluding "We define them to be so," of course.) The fact that they are semantically equivalent is exactly the problem, as I see it.
    – Matt
    Commented Feb 15, 2012 at 7:27
  • Re "nai desu", that was probably an unnecessary distraction. I just wanted to show that the "but it politifies to arimasen" argument isn't watertight, because it can also politify to "nai desu". (Which leads to the same problem: no way of distinguishing "[aru + neg] + polite]" from "[[nai] + polite].)
    – Matt
    Commented Feb 15, 2012 at 7:36
  • 1
    Well, the difference vs "go/went" is that "go" has "went"-shaped gaps, and vice versa. "nai" has no gaps -- we only want to hijack it because we are uneasy with a verb paradigm with no negative, which is arguably begging the question. Anyway, I think we have reached an impasse here. Based on semantics/syntax, you argue that they are unitary, so why not put "nai" in the "aru" paradigm? Based on morphology/etymology, I argue that they are separate, so why not leave them that way? What we need is an objective way to compare the explanatory power of our positions, but I can't think of one...
    – Matt
    Commented Feb 15, 2012 at 12:52

I think the question being raised is whether there is a sharp line in inflectional morphology between lexemes (usually verbs) which have defective paradigms, and verbs which exhibit suppletion in their paradigms. (For an approach to inflectional morphology centered around paradigms, see Stump 2001).

I think that to a certain extent it is a matter of definition, but nonetheless there are a large number of uncontroversial cases that furnish examples of suppletion and of defective paradigms. Traditionally, the term suppletion has been applied to alternations in a paradigm between verb forms which are etymologically unrelated (cf. Rudes 1980), although most modern analysts would say that the real question is whether the forms are phonologically related or not (cf. Corbett 2007; Fertig 1998; Mel'cuk 1994). Textbook examples of suppletion, however, almost always draw upon etymologically unrelated forms. A defective paradigm, on the other hand, is one where an expected alternation (expected based on the behavior of other verbs in the language) is simply unavailable.

While I am not familiar with the Japanese facts, I'll go through two simple examples from English and Latin so that you can get a flavor of the type of argumentation required, and then probably make a judgment on the Japanese for yourself. Because you refer to conventional forms that seem to fill a hole in the paradigm, my first guess would be that these are suppletive.

First, take the English forms go, went, going, gone. Under a defective paradigm analysis, we would claim that there is simply no past tense form corresponding to present tense go. There is, however, another verb, went, with a severely defective paradigm (having only this one form), which tends to get used where a past-tense analogue of go is needed. (See OED entries for go, wend, for more details) What makes this analysis improbable is that the alternation between go and went is very robust and regular, and the two words have no appreciable meaning difference other than a difference in tense. Phrasal verbs using go, for example, for their past tense with went. (e.g. go crazy ~ went crazy).

Now consider the Latin deponent verbs (e.g. locutus sum "I have talked"), which have passive-voice forms, but active voice meanings. For these verbs, there are no forms corresponding to a passive voice meaning, related or otherwise. From glancing at the meanings of some deponent verbs (conor "to attempt"; polliceor "to promise"; orior "to arise"), we see that passive meanings corresponding to these verbs probably would not arise very often in the course of language use. Verbs having suppletive paradigms, on the other hand, tend to have a high frequency (see Veselinova 2006 for details), and the meanings expressed by the suppletive forms tend to arise frequently.

Greville G Corbett. Canonical typology, suppletion, and possible words. Language, 83(1):8–42, 2007.

David Fertig. Suppletion, natural morphology, and diagrammaticity. Linguistics, 36(6):1065–1091, 1998.

Igor Mel’cuk. Suppletion: Toward a logical analysis of the concept. Studies in Language, 18(2):339–410, 1994.

Blaire A Rudes. On the nature of verbal suppletion. Linguistics, 18:655–676, 1980.

Gregory Stump. Inflectional Morphology: A theory of paradigm structure. Cambridge: CUP, 2001.

Ljuba N Veselinova. Suppletion in Verb Paradigms: Bits and pieces of the puzzle. John Benjamins, Amsterdam, 2006.

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.