Steps towards a key to CJ verb morphology

(I imagine that everyone who didn't already scroll past this post based on the title already knows this, but just in case: the following was inspired by and builds heavily on the recent verb-related posts by Ronald over at Ibadairon (and subsequent discussion in comments between Ronald, Amida, Azuma, myself, etc.)

(Update, a few hours later: I reorganized the tables a bit for clarity. No actual details have changed.)

(Update 2, later yet: removed all evidence of my appalling "yo"-related blooper, apparently before anyone noticed it (phew). Fortunately, this actually makes the theory neater and more consistent. In return for this Orwellian act, I have prepared (on the train this morning) a marvellous application of this morphology to all (all? yes, all!) auxilliary verbs, plus adjectives for good measure, which is too long to write in this margin but which I shall post tonight.)

First: there are 3 verb classes + 3 irregular verb sets

  1. Consonant-Stem Verbs (C-type) -- equivalent to 4-dan in traditional analysis, CI in Ronald's
  2. Vowel-Stem Verbs (V-type) -- equivalent to shimo/kami-1-dan in traditional analysis, V in Ronald's
  3. Double-Stem Verbs (D-type) -- equivalent to shimo/kami-2-dan in traditional analysis, CII/CIII in Ronald's

  4. aru-type -- ra-gyou in traditional analysis, basically C-type with an SS form that's merged with RY
  5. sinu-type -- na-gyou in traditional analysis, basically D-type without a vowel-ended stem (see below)
  6. se/ko -- su and ku in traditional analysis, basically D-type but they use vowel-ended stems for RY forms (see below).

C-type verbs have stems that end in consonants: 咲 sak-, 思 omoh-, 打 ut-

V-type verbs have stems that end in vowels: 見 mi-, 蹴 ke-

D-type verbs have stems with two allomorphs, one ending in a consonant (sC), the other in a vowel (sV): 過 sug-/sugi-, 当 at-/ate-

For each D-type verb, the consonant-ended stem is always used for RT, SS and IZ, and the vowel-ended stem is always used for everything else.

The verbs conjugate like this:

Stem ->C-typeD-type (sC)D-type (sV)V-type

* Plus a single extra rule to prevent consonant clusters:
  • ~C.rV -> ~C.u.rV
(This rule is actually only relevant to RT and IZ forms of D-type verbs, which would otherwise be ~Cru (e.g. sugru) and ~Cre (e.g. sugre) respectively.)

We now have a Small Chart letting us conjugate virtually every verb in Japanese if we know whether it is C-, V- or D-type.

(Note that I am using "raru" to stand in for "sasu" and "rayu" as well, since they follow basically the same rules)

There are two types of allomorphy

Allomorphy driven by preceding phoneme (end of verb stem)

after consonantafter vowel
a (MZ)0
i (RY)0
re (MR)yo

Allomorphy apparently driven by type of verb (C-type vs D-type or V-type)

C-typeD-type or V-type
u (RT)ru
e (IZ)re

I don't really like the RT and IZ forms being driven by verb type rather than phonetic rules -- especially given that the result is always a consonant cluster which triggers the "add a /u/" rule -- so let's soften the blow by calling it "driven by phonetics, with a special exception for D-type verbs."

That lets us simplify the above Small Chart to this Even Smaller Chart of Endings:

formallomorph 1
(following a consonant)
allomorph 2
(following a vowel)
force "following a vowel" allomorph
for D-type verbs?

* Not forgetting, of course, the consonant cluster-avoiding rule:
  • ~C.rV -> ~C.u.rV
Dealing with the irregular verbs

This is the fun part.

aru-type verbs are, like I said, identical to C-type except for their SS form, which appears to have been borrowed from their own RY form. Bo-ring.

se/ko (su and ku) are standard D-type verbs (stems: s/se, k/ko -- ko is the ONLY verb to have a stem allomorph that ends in an o) except that they use the consonant-ended stem when creating their RY form. (Note that according to the allomorphy rules above, since the stem ends in a consonant, the RY "-i" doesn't turn to 0. Neat!)

sinu-type verbs are the interesting one. They are, as I said, basically D-type verbs without a vowel-ended stem. What this means in practice is that in places where phonetically-driven allomorphy would normally change a D-type ending (e.g. a -> 0 because it follows a vowel), the ending does not change because it follows a consonant instead. This all checks out.

Summary: how to build a given form of any CJ verb -- a Dyad of Small Charts

1. Figure out which stem to use

C-type (and aru-type) verbs, V-type verbs and sinu-type verbs have only one stem, making this step easy. For D-type verbs and se/ko, this table sums it up:

Type of verb -> D-typese/ko
for MZvowel stemvowel stem
for RYvowel stemconsonant stem
for SSconsonant stemconsonant stem
for RTconsonant stemconsonant stem
for IZconsonant stemconsonant stem
for MRvowel stemvowel stem
for raruvowel stemvowel stem

2. Attach to this stem the appropriate allomorph of your desired form's ending

formending after
ending after
special notes for
D-type verbs
RTuruforce "after vowel" form*
IZereforce "after vowel" form*

* Plus consonant cluster-avoiding rule:
  • ~C.rV -> ~C.u.rV
3. Finally, check for the one remaining irregularity
  • aru-type verb, SS -> use the RY form
And I do believe that covers everything -- yes, EVERYTHING -- about conjugating CJ verbs. He said arrogantly, late at night.

Obvious questions, in order of interest to me:

  1. What's the deal with the D-type verb RT and IZ forms -- why can they force an "override" of the phonetic rule, even though this leads to the invocation of a special consonant-cluster-catching rule? (Could it be something to do with building on the SS, as Ronald suggests?)
  2. Were sinu/inu originally full D-type verbs that lost their sV stems at some point? If so, how did that happen? If not, how come they so closely resemble incomplete D-type verbs now?
  3. What's the deal with the aru-type verb SS form? (These words, especially "ari" and "nari", got serious use in CJ, so this is not a minor detail in the language.)
Now, good friends, tear it to pieces!

Popularity factor: 8


Well...I'm still ruminating. But impressed, nonetheless.


: )


You really bothers me.

I mean, that RT/IZ u, of course! : )

Our future perspective shows us that the system was undergoing a change whereby the RT was replacing the SS. This process had already completed in the V-stems & yodan C-stems (but wouldn't finish in the others until the Muromachi Period). These two latter indicate to me that the RT & IZ endings were (or were trying to become) -(r)u and -(r)e. (Hence the merger of the RT & SS?) We can't know what earlier versions of the RT & IZ of the V- & yodan C-stems looked like, but if in the D-stems the endings were attached to the SS as I have suggested, it would represent one more instability in the system pushing it toward the reorganization that eventually occurred (e Ds becoming V-stems and i Ds becoming either V-stem or C-stem (stems in h b m r))

Still ruminating.... (Moo)


Nice work. I'll put my own thoughts in a post later. Though I have to say, does it really bother you that verb classes drive the endings? I kind of like that. I love the idea of verbs having their own personalities and natures you have to get to know slowly. As a puzzle, I can appreciate the frustration, of course.


How about this as a story: D-type verbs (excuse me using my own terms) originally only had the vowel stem. Then, for whatever reason, the vowel stems started turning into consonant stems in certain forms. However, the allomorphs that originally followed vowels were still used to create the endings, even if that led to the creation of sub-allomorphs via rules like ~C.rV -> C.urV etc. Oh, but ooh! Wait! I know how to make your theory work!! Thanks!!

(I like shifting all of the weirdness onto the SS form like this, because SS is also the location of the exceptions which distinguish aru-type verbs and su/ku from their related verb families. Cordon the whole form off, I say, it's a disaster area.)


Azuma: it doesn't bother me that CJ ended up the tantalizingly almost-but-not-quite-logical Big Chart that it did, but I do like the idea that the whole thing can be derived from simple rules. To me, that makes it seem all the more incredible... the trick is not to take it so far that you end up with a linguistic version of Schoenberg's idea that all of western music is just variations on E D C (if you notice that this is also the opening phrase in "three blind mice", it becomes difficult to take Schoenberg seriously ever again.)


Oh, no, that doesn't work. Damn.


Huh? Which? Where? Why? What?!?

If the D-verbs had originally been V-stems, how/why did they develop differently than the mainstream V-stems?

I think Azuma's comment elsewhere about Old Irish verb prefixes is relevant here, only turned around for suffixes: it's hard to tell what these endings are remnants of. Just think of the changes we can see in the record (nari from ni+ari, -katta from -ku+atta (not verbal, I know)).

It's like we have a series of snapshots of the area around a kaisatsuguchi in Tokyo Station; we can see individual people coming from various directions and falling into lines to pass through the gates, but we can't say anything about where they came from before they entered our view.

(I know: Go to bed, Ron!)


Basically, I visualise them as originally V-stem verbs suffering from creeping consonant stem rot, because that lends drama to their heroic struggle against consonant stems, culminating in their triumphant rebirth as fully-fledged MJ V-stem verbs based on their own RY forms, which they had with them all along.

(The h/b/m/r stems which became MJ C-stem verbs are the Lost Tribe.)

As for how... collision of slightly different language groups, perhaps? Is it plausible that if you go back far enough in Japanese history you might find people who say sug'ru and sug're living not far from people who say sugiru and sugire? Ah, speculation.

Comment season is closed.