Handbuch der orientalistik section eight central asia

It is therefore very probable that voicing (prenasalization) in Proto-Japanese depended on the tone (pitch) of the syllable: high tone caused prenasalization while low tone did not. Since - at least in the series of labial stops - this phenomenon is clearly correlated with voic­ing in Mongolian (see above), we may safely project this prosodic fea­ture on the Proto-Altaic level, and reconstruct high tone (pitch) where Japanese has prenasalization, and low tone (pitch) where it has none.

It is most certain that this conclusion will have far-reaching conse­quences. Several phenomena (pitch on non-initial syllables in Korean, vowel length on non-initial syllables in Tungus-Manchu; loss or pres­ervation of final vowels in Turkic, Mongolian and Korean) may possi­bly be explained using this information. But this remains work for the future.

2.2. Development of the PA consonantal system in the daughter lan­guages.

2.2.1. Tungus-Manchu.

The TM system appears to be the most archaic. Only the following changes took place:

A. Voicing of initial unaspirated dentals:

1. *t-> *d-, *c-> *3-

B. Spirantization of the velar *kr

2.*kf>x

C. Loss of the distinction of aspirated vs. unaspirated consonants

*p'-, *tf-, *cf- > *p-, *t-, *c-

*_p. > -b-, *-c- > -s-

CHAPTER TWO

79

[Note: the latter rule probably means that the affricates in early PTM were phonetically fronted: otherwise we would expect a merger of *-c- with *-s-, not with *-s-.]

5. *-p'-, *-t'-, *-c'- > *-p-, *-t-, *-c-

D. Loss of *z-:

6. *z- > *s-

E. Loss of resonants in some structures of the type CVR(V):

7. *CVn-, *CVr-, *CV1-, *CVj- > *CV-

F. Loss of palatal *1, *f

8. *1, *i > *1, *r

Note that rules 1-5 are successive; a change in their order would lead to different events. Rule 7 must also precede rule 8 (since *1 and *f are never lost, their change to *1, *r must have occurred already after the original *1 and *r were lost). But in relation to each other, the groups of rules 1-5 and 7-8, as well as 6 (*z- > *s-) are independent, and could have occurred in any order.

2.2.2. Turkic.

The following processes must have happened resulting in the recon­structed PT system:

A. Loss of *s

*s- > *cf- before back vowels

*s > *s elsewhere

B. Lossof*pf-

3. *pr- > *h- (still present in PT to judge from the Khalaj data, see above)

C. Deaffricatization of *c-

4. *c- > *t-

D. Loss of initial resonants
5a. *ri- > *3-, *i- > *d-

5b. *n-, *1- > *d-, *m- > *b-, *rj- > 0-

E. Palatalization of *d-, *z- and *y

6. M-, *z- > *3- > *j-

F. Loss of aspiration contrast

*p- > *b-, *t- > M-, *k- > *g-

*-p- > -b-, *-k- > -g- [the latter only before -r-]

*pf > p, *kr > *k, *cr > *c, *tf > *t [occasionally *tf- > *d- before % *i, *1]

Rules l-5a are the earliest, because they are common for Turkic and Mongolian (see below); this is the main reason why we think that the initial resonants were lost not simultaneously, but in two successive steps (first the palatalized, then the rest).

80

INTRODUCTION

Rules 5-8 are specifically Turkic and have to be ordered exactly this way, because otherwise the final system would look quite differently.

2.2.3. Mongolian

Here we must suppose the following sequence of events:

A. Loss of *s

*s- > *c'- before back vowels

*s > *s elsewhere

B. Loss of *pf
3.*pr->*h-

C. Deaffricatization of *c-

4. *c- > *t-

D. Transformation of resonants

5a. *ri- > *3-, *1- > *d-

5b. *rj- > *0-/ *n-, *g- (depending on the following vowel, see above) 5c. *-f- > *-r-, *-l- > -1-, *-ri- > -n- or -j- [the latter with still unclear distri­bution]

E. Loss of *z

6. *z- > *s-

F. Palatalization of dentals before *i

7. *tf>*c', *t>*c,*d>*3

G. Fricativization of *-b-, *-g-, *-rj-

8. *-b- > *-w- [except for positions in clusters and before *k, *g]

*"§" > *-h- [except for positions in clusters and before *g]

*-n- > *-h- [except for positions in clusters where it stays as *-ng-; after

*n- where *-n- > -g-; and before *b, *g where *-rj- > *-m-] H. Intervocalic Lautverschiebung

9. *-p- > *-b-, *-t- > *-d-, *-k- > *-g- [but not *-c- > *-y\]

10. *-pf- > *-p-, *-tf- > *-t-, *-kr- > *-k-, *-c"- > *-c-
I. Accent transformation of *p
H.*pV>(*fV)>*hV

J. Initial Lautverschiebung 12. *p-, *t- > *b-, *d-13.nr-,5fkf-,5fcf->n-,'fk-,*c-

Rules l-5a are common Turko-Mongolian (see above).

Rules 5b-6 are in fact independent and unordered; they could also be positioned anywhere in between any of the rules 8-13 or even after them.

CHAPTER TWO

81

Rules 7-8 also are independent of each other and their order could be reversed; but they both had to precede the ordered group of rules 9-13.

2.2.4. Japanese

A.Lossof*l-,*l-l.*l-,*l->*n

B.

2- -g- > -y- in the 3d syllable

C. Transformation of affricates
3.*c>*cf-,-s-

*z, *s > *s

*c > *tf, *3 > *d

D. Aspiration rules [established by I. Gruntov; Z here denotes any

voiced consonant]

6. *CVCV, *ZVCV > *C'VC'V; *C'VCV (not *CVZV) > *CVCV

E. Palatalization rules

*b > *b, *d > *d, *g > g after *i-diphthongs and before -j-

*n- > *ri- before *i

F. Prenasalization rule

9. *-C-, *-Z- (not *-C-!) > *-nC- in non-initial syllables with high pitch

G. Voice shifts and mergers

*b- > *p- before low vowels, *d- > *b- always

(*k > g, *p > b), *t > *d [but *d before front vowels]

*b > *p, *d > *b, *g > *y

*C,*Z>*C

H. Transformation of resonants

*n- > *0-

*f > *t before -i, -u

*1 > *nl before *-rV-, *r > *nr before *-rV-

*nr > *nt, *r > *t (sporadically)

*I > *s, *f > *r, *1 > r, *ri > *n

*n > m-, -m- / -n-

I. Disappearance of voiced fricatives

20. *p > *b-, -w-, *6 > *d-, -j-, *-y- > *-0-

The final rule may not in fact be necessary: it depends on our inter­pretation of the reconstructed PJ system. S. Martin, e.g., prefers to re­construct *-b- and *-d- in intervocalic position as well, even though OJ and all dialects reflect -w- and -j-; on the other hand, it may be argued

82

INTRODUCTION

that PJ did not have *b- and *d-, but only *w- (*p-) and *j- (*6-), even though Ryukyu dialects have b- and d- - these all are non-distinctive features.

The rules of phonetic development in Japanese are rather compli­cated and involve a hypothesis about several intermediate steps with assimilations, prenasalizations and palatalizations. Moreover, only rules 1-2 (*1- > *n- and weakening of *-g- in the 3d syllable) are common to Japanese and Korean and distinguish this subgroup both from Turko-Mongolian and Tungus-Manchu. Since both prenasalized con­sonants (clusters like *-mp-, *-nd- etc.) and palatalized consonants (*d, *g) are quite common in Austronesian languages, one might speculate that most phonological changes in the history of Japanese occurred al­ready after the very early split of Korean and Japanese (around the 3d millennium BC) and the subsequent migration of Proto-Japanese to the Japanese archipelago, under the influence of substratum Austronesian languages.

2.2.5. Korean

A. Loss of *1-, *1-

l.*l-,*l->*n

B.

2. -g- > -y- in the 3d syllable

C. Transformation of affricates
(3a) *-iu- > *-u-

3b. *si- > *s- > *h-

4. *z, *s > *s

D. Voice shift

*t, *k > *d, *g

*-b-, *-d-, *-g- > *-w-, *-r-, *-y-7.*C,*Z>*C

E. Resonants

*i)-, *n- > *n-

*r, *r, *L *1 > *r

10. *-jR- > *-j-

F. Final dialectal developments
ll.*-y->-h—-0-

12. *-n- > -h- ~ -0-

Groups of rules C, D and E are independent of each other and could have happened in any order - but after groups A and B (the only two rules common to Japanese and Korean) and before group F.

CHAPTER TWO

83

2.3. Consonant clusters

Initial consonant clusters are highly atypical for modern Altaic lan­guages (although occasionally they emerge due to vowel loss, as in some Southern Mongolian dialects, Korean and modern Japanese dia­lects), and were certainly absent in Proto-Altaic. However, medial con­sonant clusters were fairly common.

The most frequent medial clusters are nn, rk, ng, jb, jr, lg, lb, rg, nt, jkf, gd, jl, bl, rp', Ik', Ik, nn, kt (each reconstructable at least in five roots).

The most typical cluster types are "resonant" + "occlusive", but we also meet "occlusive" + "resonant", "resonant" + "resonant" and "oc­clusive" + "occlusive".

Here is a complete chart of PA consonant clusters and their reflexes.

PA	PT	PM	PTM	Kor.	Jpn.	Roots
*rp	(r)p	rb	rp	rp	P	*arp'a, *sarp'a, *sarp'e, *sarpri,*sirp'a
*rp	rp	rb	rb	r?	(m)p	*siarpu, *kiurpe, *sarpa
*rb	rp/b	rb	r		t	*t'arba, *t'erbd
*rm	r	rm	m	m		*kfirma, *siorme
*rtr	rt	t	[rt]			*tertfa, *ertra
*rt	rt	d	rd	r	(n)t	*kiorteme
*rd	r(V)t	rd/(r)3	r[d]		t	*burdV, *kerdu
*rc?	re	re	c	nc	ns	*sarcu, *sercu
*rs	rs	rs	rs	s	s	*borso(krV), *karsi, *kirsi
*rkf	rk	rk	rg	k	k	*borkri, *sark'V, *sedurk'V, (*tferk'o) 1
*rk	rk	rg	rk(/gd)	(r)k	(n)k	*curka, *korke, *riiarke, *prerkV, *erka,
						*gerki, *kr!rka, *s[e]rko,*triarko, *pfarkV
*rg	r(V)g	rg,r(V)y	rg	r(h)	nk	*erga, *iargi, *miurgu, J