Japanese/Korean/Tungus

[Alexander Vovin]

Dear Stefan,
 
     Thanks for support -- and, of course, all the questions you ask are
right to the point and need to be answered -- I'll try to do it below.
Note that I have to comply with our moderator directions, and therefore,
can leave only essential part of your posting in order to make it short.
 
On Fri, 28 Mar 1997, Ralf-Stefan Georg wrote:
 
> Since I have been addressed in Sasha Vovin's posting, let me throw in a few
> remarks. Basically, Sasha, this time, I'm on your side, I have no quarrel
> with the notion that Japanese and Korean are related (and there might be a
> relationship with Tungus after all). Your Diachronica paper helped much to
> form my opinion on this, since there you managed to show morphological
> evidence, just the thing which - as you correctly observe - I want to see.
>
> For Japanese you list such forms as (I believe the unasterisked forms
> represent attested Old Japanese):
>
> -ey < *(a)Ci< *(a)gi  (to be compared with Korean -Gi etc.)[in Diachronica
> XI you give it without the bracketed initial *a, what does that mean ?]
> -u < *-wi < *-bi        (to be compared with Korean * -bi, Tungus *-bi
> etc.), and some minor cases where the actually compared form on the
> Japanese side is a reconstruct.
>
> Could you please help to clarify for us non-Japanologists, how the
> proto-Japanese reconstructs are arrived at ?
 
A.V.:
Reconstruction of -ey as *(a)Ci is based on the following facts. We do
know from internal evidence that OJ ey comes from *aCi (cf. sakey 'rice
wine', but saka-dukyi 'cup for sake', examples of this sort are abundant).
We have to posit a *C between the vowels, as no vowel clusters are allowed
in the proto-language. The proposal that *C goes back to *g belongs to
Martin, and is based, if I remember correctly, on observation that this
while only two other consonants (*r and *m) can be lost in intervocalic
position, their loss is accountable for (Whitman's law), while we can't
accont for *C disappearance here. In addition, *g is exactly the stop
lacking in the proto-system (only p, t, k, b, d are reconstructable), so
we pose *g for pre-Japonic. Now, *(a) is included to account for the pairs
like tuk- 'be attached' : tukey- 'to attach' (Unger reconstructs the stem
*tuka- for the intransitive pair here, and I do not accept this final *-a,
for the reasons too long to list here, but you can trust me that there are
other counterarguments that in my opinion outweight benefits of
reconstructing *-a for the majority of PJ verbal stems).
    Reconstruction of final -u as -bi I believe might be warranted by
final -mi in some  Ryukyuan dialects (cf. Okinawan -N) corresponding to
-u in OJ. There is also -mi in OJ which in certain cases may be argued as
quality verb final, although the evidence is more slippery here.
 
S.G.:
> I'm not so happy with *-gi "causative relic" and your Manchu example. Why
> don't you cite straightforwardly Evenki -gi: (with a long vowel, which is
> probably  of proto-Tung. status) 'transitivizer' ? It does the job much
> better, since the Manchu verb algi- you give doesn't mean "let know" at
> all, but rather "to become or be known, famous", rather an
> intransitive/anticausative formation to the root found in ala- "to report,
> say".
 
A.V.:
Yes, you are right about Evenki -gi: -- it slipped my mind, as I said that
I was typing the list on the basis of my memory, and I know Manchu better
than any other Tungusic language. As for Manchu algi-, you are right, it
means "be known", but known that this -gi- can still be compared with
Korean 'transitivity flipper", since the latter goes both ways.
 
S.G.:
  > The
proto-Tungus negative marker (an independent particle in all
> probability) is given in Diachronica XI, by you as *aana (a:na, with a long
> vowel), I think that reconstruction is the one which should be preferred,
> since Evenki a:chin 'not, yok' seems not to leave another choice. Btw., do
> you have any idea about those obscure elements as Manchu -kU or Evenki (and
> other Tungus lgs.) -chi(n) here ?  In the same Diachronica paper you give
> the MK (Middle Korean ?) equivalent as /ani/, here as an(h); which is
> correct ?
 
A.V.:
You are right again about the vowel length in Tungusic, should be *a:na. I
would compare J and K forms to Even a:n etc. (TMS 1.41a), rather than to
Evk. a:cin, although they are probably also related. As for Manchu akU, I
have demonstrated recently that Manchu -k- < *-nk- (forthcoming in JSFOu),
that gives us *ankU. It is not possible to separate -KU as such, but I
trust that we can of course have *ank-U on the basis of the other Manchu
negative waka. MK ani appears to be the original form, and anh- (verbal
stem) is a contraction of ani-ho- "not-do", although there is, I believe
some scanty evidence fro dialects permitting to pose two of them as equal
variants.
 
 
S.G.:
>
> On the correspondence pattern:
> -kyi < *-ki                MK -e/-a ?< *Ga/Ge      Manchu xa/xe/xo < *kV
> retrospective              perfective              perfective
>
> you give in Diachronica XI (*if* this is the same cognate set, I may of
> course be totally wrong here),
>
> as equivalent to Manchu -ha/-he/-ho < -*kV a Japanese "past marker" *-iki :
> is this the same thing as "retrospective -kyi < *-ki"? In the same list you
> give as MK (unasterisked, therefore attested ?) -ke-, here -e/-a ?< *Ga/Ge,
> could you please clarify what is actually attested, what is reconstructed
> and on what reasons ?
>
> On the cognate set
> -te- < *-ta-Ci-            -te-/ta-                ----
> perfective                 perfective
>
> we read in Diachronica XI: Jap. *-ita-Ci-, MK -te- (and Tungus *-taa-,
> which I don't know well, there seem to be several dental suffixes you could
> have in mind, would you mind giving an example ?). Again, which is the
> state-of-the-art ?
 
A.V.:
The discrepancy here is due to the fact that I have changed some of my
views recently. My original position was that  retrospective and
perfective markers in OJ do not follow the infinitive (gerund) marker -i,
therefore I posited *-iki and *-ita-. Now, I am persuaded by majority's
point of view that they do follow *-i, and therefore, I presented *-ki and
*-ta-Ci- here.
 
S.G.:
>
> Furthermore, is the MK "realis participle" -(V)n the same thing as the
> *-na- suffix from Diachronica XI, which is there compared with Jap. *-in-
> "perfective" ?
 
A.V.:
Stefan, I believe you were looking at the wrong column in Diachronica XI,
which compares PJ *-in-, MK -n (which is the same as -(V)n here, and PMT
*-na- (which I did not include here this time).
 
S.G.:
> The "tentative/volitive" suffix from your posting (Jap. -ama-, Tung. -Vme-)
> looks in your paper again a little bit differently: Jap. *-am- (OK, I
> think, the latter is the reconstruct, whereas the former the attested form,
> or ???) and Tung. *-m-. I'm not too familiar with that latter suffix, so
> one or two examples might help a bit.
 
A.V.:
If OJ subjunctive -amasi is not build on tentative *-ama- and -si (also
appearing in negative tentative -azi < ?*-an-si), then tentative should be
reconstructed as -am- rather than -ama-, but this is a difficult choice to
make. I forgot  to add Korean promisory -(V)ma to this list, btw.
    Tungusic - *-m- or (V)me- (likely the first one, as I told I was
typing from my memory this time) is taken from O.Sunik's "Glagol v
tunguso-man'chzhurskikh iazykakh", which is at home at present time, so I
can't give you the exact reference to the page (you'll find it in the
section on mood markers). Again, from my memory, Even has -mna-
for tentative, and -mci- for subjunctive (cf. Evk. subjunctive -mca:-, I
don't remember the Evk. tentative form), on which Sunik builds his
reconstruction.
 
     Finally, I might be overcautious with Jurchen data (you are right,
there are markers that resemble -me and -re both phonetically and
functionally), but this is the same point I've already brought once in our
discussion of the "Secret History": until the careful Chinese-oriented
reconstruction of Jurchen is done, it might be premature to operate with
Jurchen data, in the sense that preference should be given to Manchu,
since our understanding of Manchu phonology is better than that of
Jurchen.
 
I  hope this will clarify the matter,
 
Cheers,
 
Sasha