Handbuch der orientalistik section eight central asia

2.1.43. PA initial *rj-

PA Turk. Mong. Tung. Kor. Jpn.

*i)- *0-,*j- *n-,*0-,*j-,*g- *rj- n- *0-,*m-/*n-

Notes.

PA *rj- is best preserved in TM.

In Turkic, the usual reflex is 0-, but there are a few cases of j- before original diphthongs with -i- and *-e-, cf. *jal- < *rjelu, *janu- < *rjenu-f *jar-in < *rfera, *jebeg < *r)iabi, *jid < *rjiuja. In one quite exceptional case we have *n-, reconstructed in Turkic in just one word, viz. the inter­rogative pronoun *ne < PA *t)iV.

Mongolian has a quite complicated distribution, depending on the following vowel. It must be stressed that the distribution must be quite

74

INTRODUCTION

recent, because the vowels that follow have the timbre acquired al­ready in Mongolian, after the complicated changes of the original PA system. Thus Mongolian has:

a) *j- in several cases before original diphthongs: *jeyu- < *rjidbi, *ja-(yu-)

< *!)iV

*g- in front of -u-: *gub- < *ijupu, *gura < *T}urV, *gu < *r)[iu]

""0- in front of -6-, -ii-: *ole < *rjiole, *undii-su < *j]iurftre, *ocu- < *r]6)cuf

*iine' < *i)dnr}i~, *osugeji < *i]usi, *ojekeji < *j]iije

d) *n- in all other cases: *na-m- < *rja, *naji- < *r)Kni, *neyu- < *r}eni,
*nai]~$i- < *r}inu, *naran < *r)lra, *nagcar-kaj < *r)idkca, *noka) < *r)iak'u,
*nolig < *i]bla, *nojir < *ijuju.

Japanese has normally *()-, but in three cases *n- before the follow­ing *-n- (original or secondary), probably due to assimilation: *nas- < *rjansa, *nan-ka- < ^olaf-kV) and *nV {*na-ni) < *T)iV. It appears, how­ever, to have a reflex *m- < *n- before diphthongs, cf. *mura-(saki) < *t]iolef *mdtd < *r)iiir)tre, *mh < *rjiabi, *mdmi < *rjiar]e. This evidently means that the combination *ni- > *ri- in early Proto-Japanese, because m- is the standard reflex of PA *ri-, see above.

2.1.44 PA non-initial *rj

PA Turk. Mong. Tung. Kor. Jpn.

*-n- *n *n, h *n *n, 0 *n,*m

Examples of PA *-n- can be found in the following entries: *arja, *arju, *ar\V, *arjo, *burje, *ar}V, *crat]o, *dirfe, *et]V, *erjV, *gdrji, *giarju, *gorjV(fV), *irp, *irjV, *drje, *ior)0, *karjV, *kerji, *ker)V, *kbrja, *k6rje, *kdi]a, *ku\)e, *kur)i, *k[ia]r\e, *krai]a, *krer\a, *k'ioT)a, *k'i6r)i, *kriurju, *kriurju, *kfor)i, *lar)i, *lior}e, *lur]a, *liir)u, *mai]i, *mer)a, *mer}a, *merju, *?nerje, *mirja, *miurjO, *m67]i, *mla]r)i, *ndj]a, ^nidrja, *narjo, *nerju, *niar]e, *niorje, *niorje, *nidj]u, *nurju, *r}iarje, *drji(cV), *6rje, *piur}u, *perji, *pir)a, *p'iur}i, *priuj]i, *prdrje, *sarje, *sar)0, *sarju(nV), *serja, *sirju, *siat]e(rV), *siurjV(-kV), *siurje, *siurje, *s[iu]iju, *sarjuf *sot]e, *tarja, *tarjo, *turja, *turji, *teai]a, *trdrju, *trarje, *trerfa, *trerjo, *trerja, *tTior)e, *triur}0-(bV), *trorje, *foi]erV, *trur]e, *Vurji, *urje, *3iorfe, *siur]u, *nutje, *nurju, *sidr}a, *t'uTje, *pdr)a, *ciar)u, *sior)e.

Notes.

PA *-n- is a quite frequent phoneme, but its reflexes are not easy to establish, because they have to be separated from the (also frequent) clusters like *-nk-, *-ng- and *-nn-, *-nrj- (on which see below).

In Turkic, *n normally stays unchanged, except a few cases of as­similation (*getnurgen < *gerjurgen < *giarju, *bon-cok < *bor]-cok < *mdrji).

CHAPTER TWO

^ , . , " „ iave tne timbre acquired al-

The same is valid for TM, which generally jd ch of ^ aJ pA

but has occasional cases of palatalization *-rj-

*buni < *bQrje, *xoni < *k'oVi). h »• < m.

In Mongolian, *-n- has complicated reflexe;

in syllable-final position — with occasional asi*nwry *QU < * r /

fore dentals: *men-d« < *mcVu, perhaps also »n1-_s- < ^J^- < * -

vocalic position it is reflected just like PA *-g-, i * ^.

in the vast majority of cases. We should noli ... < ^ .^ < . ^

Mongolian in these cases frequently has not -g.^. < , ..^ < .^

usual PM *-h- ( < PA *-g-), almost always rer

Mongolian. Some other types of reflexes are ah

ronments: - cases *n- before the follow-

a) PA *-rj- is reflected as *-g- after *n- (rule estr due to assimilation: *nas- <

*nagaj < *lur]a, *nagacu < *ner)u, *nige(n) < *rf-«0 < *Viv- lt appears, how-some variation in this position: we have b^phthongs, cf. *miira-(saki) < and *nigul ( > Khalkha niigel) from *niarje; < *WW- This evidently means nos) and *nogur-su (Dong. noGosun etc.) fron-roto-Japanese, because m- is ( = *neyune, Khalkha niine) from *lidt]e.

b) before the following *-y- and *-b- it is reflec

*siur]V(-kV), *temeyel < *trer)o, *kamar (but a]

with standard reflexes of the same ro<
*kayu-rga-sun. y Kor. Jpn.

c) in a few cases before the following -ir-, -ur- n' u n' m

stead of the expected -y j-: *arjgir < *arja the following entries: *aija,

*(h)orjgur < ydrje, *or}gul < *drje; note c]e> *ei]Vf *erjV, *garji, *giarju,

WMong. norjya-su < *nor)gur-su together w^f, *Ker)V, *kdrja, *kor)e, *kdt]a,

*nogur-su and *nowur-su, see above. In these^Yfyf, *kfiiiT]u, *kriur)u, *k'oi]i,

to suppose any original clusters, so there JPmerja, *merju, *merje, *mir)a,

variation between *-y- and *-ng- in this posi, *W^M/ *nidr)e, *nior}e, *nior)e,

Korean normally has n in syllable-final posr *-prem, *mr\a, *vriuni, *vciur\if

-'- at the syllable boundary) or -n'- in intervc^T^OV), *siur)V(-kV), *siurje,

cases we also encounter assimilated reflexes -iirjar *tut]i, Yarja, *traiju, *tearje,

-h- (*kin- < *ker)i, *kin- < *korje) or -n- (*kaie, *trorjerV, *tur)e, *trurji, *ut]e,

*siurje-lV). fya, *ciar)u, *sioTje.

Japanese can reflect *-n- as -n- or -m-, with so far unclear (as in the case of PA *-ri-, see abo

Like other resonants (*r, % *m) *-n- can cut lts reflexes are not easy to nominal stems before original velar suffixes: */ted from the -*set](i)-gV (cf. TM *serj-gi).

76

INTRODUCTION

2.1.45. The problem of voicing (prenasalization) in Japanese and its consequences for Altaic

While discussing the fate of PA *p' we have paid attention to the fact that the PJ prenasalized reflex *-mp- is correlated with the PM voiced reflex *-b- and that both the voicing in Mongolian and prenasalization in Japanese may have been due to prosodic factors. We have also seen that the majority of cases with -mp- are associated with high pitch (in cases when it can be reconstructed), while the majority of cases with -p-are in syllables with low pitch.

Let us now try to examine the whole Japanese evidence and see if these conclusions are valid elsewhere, not only for Altaic *-p'-. The following preliminary remarks must be made here: Unlike the first syllable, the pitch in the second and following sylla­bles has several restrictions in Japanese:

1. All possible values of pitch are found in disyllabic nouns;

In tri-(and more)-syllabic nouns, with very few exceptions, high pitch on the first syllable can only be followed by high pitch;

In verbs, with very few exceptions, high pitch can only be followed

by high pitch

4. In tri-(and more)-syllabic verbs low pitch can only be followed by

low pitch;

5. In adjectives high pitch can only be followed by high pitch, and low

pitch can only be followed by low pitch.

Voicing (prenasalization) in polysyllabic forms also has one general restriction, namely: two voiced (prenasalized) consonants within one stem are not allowed.

It follows that several cases of apparent exceptions may be actually due to the above restrictions: the pitch of the second syllable is irrele­vant (neutralized) in trisyllabic nouns like *CVCVCV (*CVCVCV is not allowed), in verbs like *CVCV- (*CVCV- is not allowed), in adjectives like *CVCV- or *CVCV~ (neither *CVCV-, nor *CVCV- are allowed); ab­sence or presence of nasalization is irrelevant (neutralized) in a struc­ture like *CVCVnCV (*CVnCVnCV is not allowed).

Below we shall restrict our observations to relevant structures only. 1. PA voiceless aspirated or voiced stops : PJ voiceless stops a) low pitch: *krap'a > *kapi; *krepre > *kdpd-; *oplkV> *pukupiiku-$i; *t'epra > *tapa-; Yap'e > *tdpbr-; [*efa > *papa\, *liapa > *napai; *sdpl > *sipa; *t'upi > Hiipak- (but modern tsubaki - with secondary voicing?); *jdp'e > *ddpdrd; *sepru > *supa-dai; *caba > PJ *tapara; *eba > *hpu-) *kreba > *kdpa; *krdbani > *kapind; *taba > PJ *tapi; *ibe > PJ *\pua; *tabu > PJ *tiipijdi; *flbi > PJ *(d)lpia; *nibi > nipdp-; *batfo > *bata; *kefo > *kdtii;

CHAPTER TWO

77

*kibtre > *kbtav, *meteo > *mdtb-; *iotea(mu) > *atama; *patra > *pdtdr-; *piatre > *pbtbp- / *putuk-; *p'atta > *pdtd (but also *pata); *prat'a > *pdtak~; *priitra > *pdtd; *slfo > *sitbmi; *pdt'o > *pdtd; *bbda > *bhtd) *gedi

*kitd; *kfudi > *kutu-pikl; *pada > *pdtd; *najadi > *nditd-m-; *eco > *btbr-; *necre > *nbti; *piucfa > *pdtur-; *krace > *kbtb; *keca > *katd; *okra > *dkudjdi; *crik'a > *tdkdrd; *cidkre > *tdki; *gokfi > *kukv, *gukfa > *kdkdr-; *kakro > *kdkl; *kok'e > *kdkdrd; *k'iujkre > *kiikui; *ok'e > *dkdr-; *pekro > *phkdr-) *vokre > *duki; *c'iigu > *tukd; *dagi > *(d)ikusd; *tega > *tdkdi; *kugu > *kukiipi; *pdgo > *pdkd', *segu > *siiku-jaka; *stga > *sdkdpi; *siga

*sdkuri', *togi > *tukd; *trdga > *tdku-; *trbgi > *tukd;

b) high pitch: *kep'a > *kdpua; *pepra > *pdp(u)i; *trepra > *tdpu-; [*prepfa > *pdpdi\; *kiiip'u > *kupd; *ldpi > *nipd; PA *keba > *kdpi) *k'ibu > *kupd; *ldbo > *ndpd; *tubu > *tupi; *bi6fe > *putd; *patra(-kV) > *pdtd; *padi > *pitu; *kidcu > *kuti; *mdcrV > *mdtud; *muce > *mdtu; *sidci > *situ; *iico > *dtd-; *^eci > *(d)iti; *uk'e> *bdkd; *d[e]gi > *(d)ikd; *muga > *mdki; *pdgo > *pdkud; *soga > *sdkd-i; *sugo > *sdkd; *toga > *tdkd; *fege > *tdkd; *uga > *dku; *6tre > *dtd-nd.

2. PA voiceless aspirated or voiced stops: PJ voiced (prenasalized) stops

a) high pitch: *crtpr[u] > *tumpd-mdi; *gopfa > *kdmpu; *kepri > *kimpi; *sapri

*slmpd; *sapri > *simpdi; *triupro > *tumpua; *fopru > *tumpu-ra; *kdpa

*kdmpu-; *nep'e > *ndtnpu-; *trepa > *trepra > *tdmpd', *kdbo > *kdmpid; *kobu > *kumpud; *labo > *ndimpu; *sabi > *simpi; *trube > *tudmpi'f *tabu

*tiimpunai', *gebo > PJ *kdmpi; *tibulka > PJ *tumpdki; *kufa > *kdntud; *latra > *ndntd; *nitra > *mntd-; *preta > *pretra > *pdntd; *eda > *dntd; *%ade > *dantd; *kecra > *kdntud; *pfuci > *punti; *borso-krV > *bdsdnki; *ciokro > *tmkd-; *manukrV > *mundnki; *mukce > *mdnkurua; *bdge > *bmkdm-; *sdgu > *sunkui; *sigi > *sinkurai; *suga > *sanki;

b) low pitch: *gapra > *kdmpd-; *kupV > *kumpd-r~; *krep'o > *kbmpbr-;

*lajprV> *ndimpar-; *sipi > *simpdr-; *upi > *urnpa-p-; *saba > *sdmpak-;
*treba > PJ *tdmpi; *batri > *pinti; *bot'e > *pbntbk-; *k'ad[u] > *kuntiir-;
*udu > *untai; *kiucu > *kuntbk-; *iikru > *unkat~; *biuk'i > *plnkatn-;
*lejkca > *niankap-', *nik'[u] > *nbnkbp-; *treki > *t'ekri > *tinkir-; *kraku >
*kcak'u > *kiinkutu

The general picture which emerges is quite curious. We see that there are many more examples with low pitch and voiceless stops (67 cases) than with high pitch and voiceless stops (30 cases); and with high pitch and voiced stops (34 cases) than with low pitch and voiced stops (19 cases). Moreover, of the 30 cases with high pitch and voiceless consonants 22 cases are disyllabic nominal structures of the type *CVCV'r and of the 19 cases with low pitch and voiced consonants 15 are disyllabic verbal structures of the type *CVCVC-.

78

INTRODUCTION

We may with high probability suppose that there was a tendency in Proto-Japanese for a metatony *CVCV > *CVCV in disyllabic nouns (note that, as we have said above, trisyllabic nouns with low pitch after high pitch are already completely absent); and for a metatony *CVCVC-> *CVCVC- in disyllabic verbs (again, as we have said above, trisyllabic verbs with high pitch after low pitch are already completely absent). If we exclude those cases we get the following distribution of tones and voice (prenasalization) on non-initial syllables:

	Hi£h	Low
Voiced	22	4
Voiceless	8	56