My really, really long post on the Classical Japanese Verb
I have to thank Matt for his excellent posts on this. If there were any boxes on the Big Chart I had only half-memorized before now, those days are gone forever. Staring and staring, and this is what I've come up with.
Matt divides the verb up into three, like Gaul:
Consonant Stems - omoh(u), sak(u), and so on, all six stems are merely variations on the vowel added to the root. [yodan, NA-hen, and RA-hen]
Dual Stems - vowel roots ending in I or E, but taking a U in the renyoukei, rentaikei, and izenkei [kami ni-dan, shimo ni-dan, SA-hen, KA-hen]
Vowel Stems - vowel roots constant throughout the six stems
[kami ichi-dan, shimo ichi-dan]
This classification is traditional and sensible, and I accept it. So I thought of lots of ways to try and neaten or simplify the charts we know and love in ways that didn't rely on good knowledge of modern Japanese. Alas, not much progress. So I stepped back and thought on the three classes I started with.
Consider the V-class. Now it makes up the majority of Japanese verbs, but it started out really small back in the day. For E-roots there's only 蹴る keru. And for I-roots a mere 11 verbs, all monosyllabic roots (excluding obvious compounds). It seemed to me that this class had to be newer than the other two. No-sword's post on keru hints that there were historical reasons leading to its isolation, but when I played around with my dictionary, I found that at least with 干るhiru, there was an earlier hu in the Nara days, a D-class presumably conjugating like E-monosyllables 寝nu, 経hu, and 得u. That is to say,
Perhaps one or more of the eleven kami-iti V-roots had personal reasons like lonely keru for founding the class, but it's clear that it proved popular for other verbs like hu so short they lacked a sense of a definite stem. For similar reasons I'd bet that 寝nu, 経hu, and 得u were the first to join keru. This is where it gets interesting.
The more I looked at the new V-stem class, the more I realized how much it resembled not the D-class, but the C-class. For one, there's a constant unmodifiable root. Second, it also eliminates the distinction between the shushikei and rentaikei, a trend we know eventually led to their fusion in modern Japanese. All the irregular verbs (su, ku, na-row, and ra-row) as well as the older D-roots preserve the SSK and RTK distinction. Only the C- and newer V-roots do not. As a matter of fact, the only reason NA-row irregular verbs exist as such is because they differ from C-roots here:
I don't know whether C-roots always ignored the distinction, and NA-rows are prickly outliers, or whether NA-rows are the only holdouts (there are only two NA-row verbs, after all) against an innovation that swept the rest of the C-roots. I suspect the latter, but even going against history, if for simplicity's sake, we consider the basic RT ending to be ru, and the IZ re, we find it holds for all verbs of all classes! Even su, ku, and RA-row verbs. In fact, on a fancy, it would make sense if first the RA-verbs dropped an extra RT ru and IZ re, considering that it would lead to a repetitive aruru and arure, and the C-roots followed suit just in time for the beginning of recorded written Japanese.
So, first step in a unified chart: (1)SS = +u; (2)RZ = +uru, and IZ = +ure, except for C-roots whose final consonants carry the burden of the U and E on their own. (The NA-row twins being remaining fashionably nostalgic).
On the same principle of seeking a basic ending with explainable (as opposed to memorized) exceptions, we can say the RYK ends in +i, unless usurped by an existing vowel stem (i.e. E-root V-class verbs). This also works for the "irregular classes". Accepting that though the inherent vowel of 来ku is O (MZ kozu), and of su, E (MZ sezu), they take ki and si in the RYK, this leads me to think that once upon a time, the RYK of the D-class as well may have ended in +i.
Which leads me to the mizenkei. No stem is as weird as the MZK, and the C-roots take the only A on the chart here. There may be no connection with the A of ancient nominal bound forms that survive even today (ama-gutu rain boots, vs. ame rain), but if we assume there is, we can conclude that the MZ stem ending is actually ZERO. The consonant verbs would use the binding A to avoid an impossible consonant cluster, and vowel verbs from ko to se to mi to tabe all get to keep their original inherent vowel. The corollary to this would be that the MZ reveals the original character of the verb most clearly, a fact to take advantage of later.
This leaves only the imperative. At first I was confused by the seeming existence of two different endings: e for C-roots, and yofor the rest. But consider 来ku and su. The ko and se of their MZK's, provisionally accepted by me as the most original form of the verb, reappears only here. Why not, then, consider the imperative ending to be zero as well? A zero-ending imperative has a lot of cross-lingual precedent, after all. I feel further out on thin ice in saying this than anything else, but perhaps C-root e and V-root yo, like later ro, are all the various results of fused emphasis particles. shin- + binding a + i could reasonably become shine. And my chart has ko or koyo for the imperative of 来ku, which lends some support to the idea or an (originally) optional particle.
So, if anyone has read this far, here we have it:
MZ : ZERO
RY : -i
SS : -u
RT : -uru
IZ : -ure
MR : ZERO + emphasis parrticle *i or *yo
With deviations explained thus:
(1) C-stems use the binding a to avoid a consonant cluster in the MZK
(2) C-stems rely on their own final consonant to support the u and e of the RT and IZ. (except the two NA verbs)
(3) All vowel roots (except ku and su) give precedence to their inherent vowel in the RY. Which practically only means, E-roots keep the E.
(4) The non-changing vowel roots of the new V-class give precedence to their native vowels in ALL stems. This has two consequences (1) RT=Vru, IZ=Vre, and (2) to preserve a SS ending in u, they must resort to ru, justified by the precedent of C-roots, which also have identical SS and RT stems.
And that takes care of everything, I believe.
If that looks like all of Japanese verbery devolving from what most closely resembles the irregular su/ku paradigms, I think it's only natural. Two of the most common verbs in the language are likely to preserve older forms and prove more resistant to innovations and simplifications. Heck, even nowadays it's still konai.
Pedagogically, I would suggest the following. The above system is accurate, but requires prior knowlegde of existing verbs to make sense. As a method for generating or recognizing forms, it won't help much if the student only learns the usual SSK, or even the more common RYK. I suggest (yay Classics!) a principal parts system. The MZK provides the most irregular forms of all verbs, and also makes it immediately clear if a verb is a C-root or not by the -a. The only ambiguity there is between D- and V-roots. For that the traditional SS should suffice. This also gives the irregular but very frequent RA-row SS ri. In fact, the only forms unpredictable from these two are the RT and IZ of 死ぬshinu and 往ぬinu, the two NA-row irregular verbs. This should pose little challenge to memorization, I think. My suggestion: the SSK, and the zu-negative, in that order.
saku, sakazu [yodan, C-root]
sugu, sugizu [kami-nidan, D-root]
tabu, tabezu [shimo-nidan, D-root]
miru, mizu [kami-ichidan, V-root]
keru, kezu [shimo-ichidan, V-root]
ku, kozu [KA-hen, D-root]
su, sezu [SA-hen, D-root]
ari, arazu [RA-hen, C-root]
shinu, shinazu [NA-hen, C-root *must memorize RT shinuru, IZ shinure]
For those of us who have studied MJ, I think this is useful too, since the big difference after all is in the SSK. ku, not kuru, ari, not aru, tabu, not taberu, and so on.
Feedback wo o-negai shimasu!