Some linguistics observations while in Japan: Kansai origins of formal Japanese

Something I became more aware of than before is that most of the humble as well as honorific verb forms that we find in modern Japanese, along with several fixed expressions come off as completely irregular within Tokyo Japanese, but are actually the regular outcome of such verbs in (an early version of) Kansai Japanese. This is no surprise, as Kyoto and the larger region was the start if Japanese high culture, and most of the written history of Japanese is in fact in this dialect. That expressions and forms would be borrowed, especially in the formal registers and fixed phrases is thus hardly surprising.

Adverbs of i-adjectives

In SJ (Standard Japanese) the adverbial form of the so-called i-adjectives is -ku, e.g. hayaku 'early, quickly'. In the Kansai dialects however, intervocalic *k is lost before *u/i (regularly? or irregularly? at least regularly in this type of adjective) yielding *hayau which subsequently monophthongized regularly to *hayō. This then gives an explanation for a whole set of formal expressions that don't make sense in SJ:

o-hayō gozaimasu 'good morning' < *o-hayaku goza(r)imasu
o-medetō gozaimasu 'congratulations' < *o-medetaku goza(r)imasu
arigatō gozaimasu 'thank you' < *arigataku goza(r)imasu

The negation of the 'polite' suffix -masu

The suffix -masu (or perhaps more properly: polite auxiliary verb) is added to verbs to make them more polite: aru 'to exist' arimasu 'to exist (polite)'. Its negation, however, is irregular. Rather than the expected **masa-nai we get -mase-n, even in modern Kansai dialects that negation would be unexpected, we would rather expect *masa-n, however historical insight into the suffix shows that it was originally a "shimo ni-dan" verb. Most shimo ni-dan verbs end up in the SJ ichidan class, verbs that have an unchanging base-stem to which in the unmarked form a suffix -ru is added such as tate-ru tate-masu tate-nai 'builds; builds (polite); doesn't build'. However, there used to be forms where there was some overlap between the normal verbs, namely in the conclusive form.

So in Middle Japanese a verb like tateru would have been tat-u in its "conclusive" form, but tate-ru in its attributive form. When the conclusive an attributive merged (as they already had in the standard verbs) this overlap was lost. But for the polite suffix, rather than becoming mase-ru in both the conclusive and attributive form, it rather kept the conclusive form for both: mas-u but its negation still betrays its old (and non-SJ origin); -mase-n is the regular negation of a Kansai dialect shimo ni-dan verb.

The volitional form of the 'polite' suffix -masu

Another leftover of the shimo ni-dan origins of the masu verb, is its volitional form. The volitional used to originally be made by adding -(a)u to the verb. -au subsequently contracted into ō. Hence we get *hanasau 'let's talk' > hanasō.

But the volitional form of -masu is not **hanasi-masō but hansi-masyō. As we discussed in the previous blog post, eu yielded yō in SJ. Hence a stem *mase-(a)u yielded > maseu > masyō. This conjugation has been completely reformed form the shimo ni-dan verbs in SJ, where a verb like mise-ru 'to show' would have a volitional form mise-yō not **misyō. The Kansai dialect still has a leftover of this behaviour in the volitional form of suru which is syō the regular outcome of *se-(a)u, rather than the SJ siyō.

The conjugation of honorific verbs with stem-final -r

A element that must be related to the honorific/polite system being based on the Kansai dialect rather than the Tokyo dialect of SJ must also lie at the basis of an unusual irregularity of the stem final -r verbs. However, the exact development of this is not completely clear to me. The regular conjugation of verbs that have a stem-final r followed by the polite suffix -masu is simply -r-i-masu. e.g. ar-u > ar-i-masu. However, a group of honorific verbs have a different conjugation, where the r is lost:

gozar-u > goza-i-masu 'to exist'
kudasar-u > kudasa-i-masu 'to give'
irassyar-u > irassya-i-masu 'to go/come/exist'
ossyar-u > ossya-i-masu 'to say'
nasar-u > nasa-i-masu 'to do'

All of these verbs are presumably originally univerbated combinations of a verb + the auxiliary ar-u 'to exist. And while ar-u is an irregular verb that behaves different from all other verbs of this shape in Middle Japanese, the irregularity is not the loss of an r before i. To my knowledge, this is not regular behaviour of the verb ar-u in the Kansai dialect either; but I would love to be corrected. Either way, finding the conjugation of a verb type being different specifically in its honorific verbal system, does point to the strange 'layered' effect of the honorific/polite system that seems to be importing its morphology from a different linguistic stratum than the rest of the verbal system.

Comments

[Matt]

Re your last section, this is actually one weird thing about Japanese formal language that doesn't seem to involve dialect mixing as such. All of those verbs (and I can add another, rarer one: tsukawasaru) do appear with /-arimasu/ forms in the early Edo period, but these gradually gets worn down to /-aimasu/ forms (and sometimes much further, e.g. gozansu, gossu). Even after that point, though, the original /r/ forms were still available, and IIRC considered slightly more "proper."

Incidentally, the same is true of the imperative forms /kudasai/, /nasai/, etc. - these were originally /kudasare/ etc. but the /r/ was lost and the vowel shifted a bit.

Re the origin of these words - most of them are not, in fact, something + /aru/, or at least not directly. /gozaru/ is (御座 + aru), /ossyaru/ might be (maybe from 仰せ + aru, maybe from 仰せらる). The other /-aru/ endings ultimately go back to the passive auxiliaries of the form /-(r)are-/ (used for honorific purposes) reanalyzed as quadrigrade verbs with /r/ stems. (The original lower bigrade versions survived alongside these new quadrigrade ones for a good while, though, well into the late Edo period at least.)

So maybe it was something to do with this origin that inspired the shared r-less -masu forms, with gozaru coming along for the ride by analogy ("honorific verbs ending in /-aru/") - or maybe it was the other way around, with the extremely common gozaru getting worn down for purely phonetic reasons and the others coming along for the ride by analogy. If I had to guess, I'd say the latter.

[PhoeniX]

Thanks for thinking along with my ramblings! Fascinating that this loss of r seems to be so late. Now that you mention it it's so clear that these forms are old passives.

And in the case of Kudasaru, I suppose that's technically from kudasu, which is probably an old causative of kudaru with a suffixation of the auxiliary -su, yes?

So, I think you're probably right to suggest that it's no coincidence that the r-less forms spread within the honorific verb system. It would only have to have happened once in one of these verbs to give rise to the rest. One still wonders, however, how it would have happened in any one of these. Every now and then, irregular sound changes of course take place, and it's difficult to figure out why it happened; but it'd be nicer, I think if we could find some kind of solution that doesn't involve assuming irregular shifts... how about this:

Could it be that there were two forms of gozaru that competed with each other? one using the auxiliary *aru and the other using the auxiliary *iru? So you'd have goza-(a)ru besides goza-iru. The goza-iru form eventually won out in pre-masu (and perhaps imperative?) form; whereas the goza-(a)ru form won out in the unmarked stem.

From there the gozaru vs. gozai-masu distribution was interpreted as a feature of honorific verbs and was spread to the other verbs.

I kind of like that idea, but I have no idea if there's any evidence for it, and you'd probably expect actual evidence for it to be present in the written record.

[Matt]

I'm definitely sympathetic to the aversion to irregular sound changes! But I think in this case it's justified and I'll tell you why:
- First of all, no evidence of "gozairu" at all as far as I know (and I just checked a couple of dictionaries and found nothing). That doesn't rule it out as a purely spoken form of course, but...
- There is lots of irregular change attested around "gozaru" anyway:
-- The shift from "goza aru" to "gozaru" is arguably irregular (certainly not unusual for an /a.a/ sequence to do that in Japanese, but since "goza aru" existed for centuries before "gozaru" appeared, it clearly wasn't automatic like OJ vowel merging)
-- From gozaru/gozarimasu we have: gozansu, gozamasu, gozarinsu, gossu, possibly asu (!!)... and those are just the ones I can remember!
-- Also worth noting that there's a parallel gozariyasu / gozaiyasu pair ("yasu" itself is probably originally from "arimasu"...)

Since there is evidence for all kinds of irregular changes to "gozaru", probably because it was so common, for me it seems more likely that "gozaimasu" was a result of that process too than an unattested other form of the verb.

But I also wouldn't want to rule out the change starting in the other verbs, because of the way you see the /i/ in the imperative forms too. Like, if kudasai < kudasare, then why not kudasaimasu < kudasaremasu? Maybe the /i/ forms are somehow a hangover from when the verbs were bigrade. If so, it's a bit suspicious that the /i/ forms originally weren't very common and only gradually became dominant, but I suppose stranger things have happened.

As for kudasaru, yeah, it's one of those ru/su intransitive/transitive pairs. It's not quite regular enough to be the auxiliary /-su/ itself, rather a lexicalized fossil of something that was apparently productive in the pre-OJ days... but it does seem highly likely to be related.

[PhoeniX]

Historical Japanese is pretty fun. Haha.

Thanks for the detailed answer.

It's pretty striking actually how much of the honorific/polite speech has all kinds of highly colloquial contractions that are not actually fully regular in other parts of speech.

I'm thinking, for example, of itterasshai; which much be from itte-irasshai, with the common e-i > e contrations as in ochite-iru > ochiteru; But, correct me if I'm wrong, but the form itte-irasshai does not seem to exist at all.

It feels counter-intuitive that honorific speech would be more prone to undergoing really extreme (irregular) reductions, but perhaps it is exactly because they are part of the highly formulaic speech that is said a lot that they undergo rather haphazard reductions.

[David Marjanović]

C'est dû à la fréquence.

In 19th-century Russian, "lord" /suˈdarʲ/ had shrunk to a politeness clitic /s/.