ffix+=A ffiv!';( tf;Elrt )
BTD Revisited:
Han BuddhistA Reconsideration of the
Transcriptional Dialect
W. South Coblin
Repinted fromThe Bulletin of the Institute of History and
Academia Sinica
Vol. L)flII, Part IV
Taipei, Taiwan, China
L993
t xrry*.ftE L#Ery.ninF:.{ffi*+a*, frw+*ifi-a$ : fi"Pl,,r.f =+)Ln
1
BTD Revisited A Reconsideration
of the Han Buddhist
Transcriptional Dialect
W. South Coblin
This study confronts anew a number of issues raised in earlier studies concern-
ing the Han Buddhist transcriptions and the dialect whose phonology they are
thoughttorePresent'Newmaterialsandapproachesarebroughttobearandapplied,andacompletelynewsetofphonologicalreconstructionsisproposed.Alarge set of sample reconstructions is offered in the second part of the paper'
EEt +++*ffiH (fti4$IEffiS ) ("Notes on the dialect of Han Buddhist
rranscriptions,,) -H " At#gffiffi ffiF! ffi'X,fi i*tfi1-tr *rfr.ffJANf.H^t+#1ki*HiE$jEFfift*fr!"fiE he" #**ffi " -#a H- #'i&.ffiHrnlBf tiltsEB!AeE{t*+ffi# "
I. Introduction
1.1 In two earlier studies (1981; 1983) a number of Chinese Buddhist
transcriptional forms from late Han times were gathered emd evaluated as
evidence for reconstructing the sound system of the language on which they
may have been based. In the intervening years new materials and new
approachestotheproblemhavecometolight'suggestingthatareconsidera-
tion of the earlier work should be attempted'
-867-
W. South Coblin
1.2 The data to be used here are of three types:
1) The material published in our earlier studies, with certain corrections
and emendations. A sizeable body of forms from these data could not be
identified, and a list of these problematic cases was circulated among
interested individuals, who were then very helpful in suggesting further Indic
equivalents. I am particularly grateful to Professors E. Zidlcher and P-
Harrison for identifying a number of these forms and also for correcting
errors in the already published data. And in addition, I should like to thank
Professor Harrison for kindly allowing me to use the new Lokakpema data he
has collected from T 624. But in registering these acknowledgements I must
emphasize that any errors or deficiencies in the material cited here are solely
my responsibility.
2) One hundred eighty-five new forms gathered from texts of the Three
Kingdoms (TK) period (220-265). These data have not been published
elsewhere and are cited selectively here. They derive primarily from the
following sources, identified by translators' names and text numbers from the
Taisho Tripitaka (T):
Kang Senghui HfBg T 152, 2Oo
Kang Sengkai H€fH T 432
Tandi E=# T 7433
Zhiqian {ffi T 54, 76, 169, 185, 198, 2lO, 225, 362, 474, 493, 632,
790
According to Ziircher (1959:55), Kang Sengkai (a Sogdian) and Tandi (a
Parthian) began working at Luoyang f&W in about 250 A.D. Kang Senghui
was of Sogdian extraction, but was born in the area of modern Hanoi, of a
family which had lived for generations in India. He settled at Jianye B X ,
in the state of Wu E , in 247 (Ziircher 1959:51). Zhiqian rtras a person of
-868-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
Indo-scythian ancestry who .was a native of Luoyang but moved to the wu
area shortly before 22O (Zwrcher 1959:48)'
3) The large late Han and TK transcriptional corpus published by Yu
Min (1984). In our own selection of transcriptional materials, we have
limited ourselves to texts considered by Ziircher to be genuine. Yu has cast
his net more widely and finds a number of forms which we have not seen'
The fact that these are usually cited by him as isolated syllables rather than
full compounds sometimes makes them difficult to interprel and use, but
many of them are nonetheless of great value and interest'
1.3 ln our earlier efforts to reconstruct the Han Buddhist Transcriptional
Dialect (BTD) we took as our point of departure the Qieyun gJffi system
(QYS) reconstruction of Bernhard Karlgren' In the present study we
continue to cite Karlgren's forms, as emended by F' K' Li' but their sole
purpose here is to serve as an algebraic reference to the sound categories of
the QYS. They do not form the basis for any of our reconstructions' Our
BTD forms are on the contrary backward projections of Old Northwest
Chinese (ONWC), which is our reconstruction for a set of closely related
dialecrs spoken in northwest china at ca. 400 A.D. (Coblin 1991a). The
major attested varieties of ONWC were spoken in the neighborhood of the
city of chang'an R* and, in the Gansu corridor. They are assumed to have
been ancestral to various stages of the Chang'an dialect of the Tang period
and to the shazhou i4 i'l\J (sZ) dialects, which ulere spoken in the
neighborhood of Dunhuang in late Tang and Five Dynasties times' For recent
studies of these later forms of northwest. Chinese, see Takata (1988) and
coblin (1988; 1989; 1991c; Ms.1). All transcriptional examples given here are
cited in the ONWC reconstruction, unless otherwise indicated. Only t'he
reconstructed BTD target forms are starred.
-869-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
after: YJ), who was born in what is now the Peking area and grew up in a
monastery near Mount Tai (Coblin 1991b). Where possible this material can
be drawn upon here to elucidate moot point.s in the BTD reconstruction.
Another transcriptional corpus which can be adduced for comparison is the
Mahamayuri translation of Sarighabhara (S), which probably represents a
dialect of sixth century Jiankang (Pulleyblank 1979; Coblin 1990).
1.5 Our views on the nature of the language or languages underlying the
original BTD texts have been set forth at some length earlier (1981; 1983:31-
33). In essence, we continue to hold with Ziircher (1977:779) that these
"may be Sanskrit, any kind of Prakrit, or even some Central Asian idiom."
This stance has been censured by Pulleyblank (1983:85) because he sees in it
a rejection of what he calls the "Gandhari hypothesis," i.e. the ideas of H.W.
Railey "nd J. Brough on the role of the northwest Middle Indian dialect (i.e.
Gandhari) in certain early Chinese transliterations of Indic material. Now, as
I read it, this theory meant for Bailey that there were "scattered traces of
the same Middle Indian dialect in Khotanese, Tibetan, Agnean, Kuchean, the
earlier Chinese Buddhist transliterations, as, in particular, in the remains in
Sogdian, Uigur Turkish, and in Mongol (in living use), and also in Manchu
texts" (Bailey 1946765). For Brough it involved '"the possibility that the
originals of some [emphasis added. WSC] of the earlier translations of
Buddhist works into Chinese were written in Gandhari" (Brough 1962:50).
These cautious forrnulations, if read as written, do not seem to be fundamen-
tally at variance with Ziircher's position. Is it possible that Pulleyblank holds
a more extreme view and would assume that all early transcriptional material
is based on Gandhari texts? This seems unlikely. For example, he has on a
number of occasions cited the BTD form ft (QYS bjwom-; ONWC buam-),
Skt. brahma, [o support arguments that Chinese qusheng syllables in -m had
-877-
W. South Coblin
final breathiness of some sort in late Han times (1962; 7978:774). And yet,
it is well known that the cluster -hm- became -mm- in the northwest
Prakrits, giving us Gandhari bramma (or brama) in place of Skt. brahma, a
fact of which Pulleyblank himself is aware (see 1962.II, p. 231). In our view,
Pulleyblank's use of the Sanskrit rather than the Gandhari form is perfectly
legitimate here.It is simply an affirmation of the possibility that the late Han
Chinese may have had access to texts which were written in something other
than "pure Gandhari." And this is hardly improbable, for we know that the
earliest Buddhist Hybrid Sanskrit (BHS) texts were probably "pre-Christian by
more than one century" (Edgerton 1953.I:5), and that in BHS texts
"sanskritisms are constantly present cheek by jowl with Middle Indic forms,
and often with hybrids which strictly are neither one or the other" (ibid.
p. 4). It would seem that we should, as Pulleyblank has himself done, remain
flexible regarding the Indic original underlying arrry particular Chinese
Buddhist transcriptional form. In summary, it is probable that, in practical
terms at least, there is considerable consensus among different investigators
on this point, rather than substantive disagreement.
1.6 In our earlier studies of BTD we took the Eastern Han rime
categories of Luo and Zhou (1958) as a primary basis for interpretation of
fundamental relationships among syllable finals. Reconsideration has shown
that this was fallacious and that it is on the contrary the WeiJin (WJ)
period categories posited by Ting Pang-hsin (1975) which are the more
appropriate framework within which to work with the BTD finals. The reason
for this may be that, in comparison with the sound system of spoken
dialects, the EH scheme was already obsolete and archaic in the second
century A.D. That it is the \4J system which is most helpful may reveal that
poetic riming in this period was out of step with phonetic reality by about a
-872-
BTD Revisited--A Reconsideration of the Han Buddhist TranscriPtional Dialect
century or so.
1.7 In using the WJ rime categories, we have in a number of cases
departed from the now popular "one rime group' one phonemic vowel"
principle of reconstruction (i.e. the "rime principle"). Two theoretical
justifications can be offered for this. The first is that there is a rather good
chance, in our opinion, that at least some of the larger wJ rime groups were
"inherited" as part of Han poetic tradition rather than being true reflections
of current. phonological reality. These inherited groups may have
incorporated finals which rimed in certain poetic dialects of earlier periods
but which were not perfect rimes in the BTD period. Secondly, if we were
absolutely certain 1) that the riming standards of Ting's rime categories were
based directly on the dialect underlying the BTD transcriptions' and 2) that
ONWC was a direct descendant of BTD, then it might be reasonable to
adhere more strictly to the rime principle. But the first of these points is
moot, and the second is an historical impossibility. For these reasons we have
chosen to be guided but not constrained by the rime principle. We
contemplate here the possibility that poetry written according to the
standards deduced by Ting from his WJ data might not have rimed perfectly
or "naturally" when read aloud in the BTD period progenitor of ONWC, or
for that matter even in BTD itself.
II. The Initials
The BTD initial system of late Han times has been discussed in detail in
our earlier studies, and it is probable that at least some points in that recon-
struction reflect generally held views and are relatively uncontroversial. There
are however, a number of matters which require furt.her discussion, because
-873-
W. South Coblin
of fundamental revisions in the system and/or because of the different
starting point (i... ONWC rather than Karlgren's QYS) for the backward
projection to the BTD stage. In the present section we shall concentrate on
changes and revisions and make only brief mention of matters on which
nothing new is suggested.
2.1 The Labials. The ONWC labials, P-, Ph-, b- and m- (= QYS p-, ph-,
b- and m-), are well represented in the transcriptional data and generally
correspond to Indic p, ph, b - bh - v, and m respectively. The ONWC
forms can be projected back to BTD unchanged.
A point of interest can be raised here regarding the following example:
T 13.241.1 ILE Uii - bii- khu; Skt. bhikqu; P. bhikkhu, Gd. bhikhu
The character LY, has another reading, pii:. Pulleyblank ( 1983:79)
identifies this reading as "traditional" and prefers it in the compound iL E .
However, it should be noted that, for the northwest dialects at least, it was
clearly one of the voiced initial readings which was used in this compound.
This is proven by the Tibeto-Chinese transcriptions of the Emituojing [-r]ffifE
ffi (now usually called Texts O and Oa by tibetologists), where the
compound occurs and is spelled 'byi-khe'u or 'byi-'khe'u. The chinese
dialect reflected in Texts O and Oa was conservative and preserved the initial
voiced/voiceless distinction well. Consequently, rtre can safely assume that the
syllable Lt in L?,8 was pronounced with a voiced initial in the northwest.
2.2 The ONWC dentals, t-, th-, d-, and n- (= QYS t-, th-, d-, n- and t-,
[h-, d-, and U-) probably had at least three allophonic variants. In syllables
having the ONWC vowels e'and a, which correspond to the various Division
II vowels of the QYS, the allophones were probably quite retracted and may
in fact have been true retroflexes. They are frequently used in the ONWC
materials to transcribe Indic cerebrals. When standing before the ONWC
-874-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
volMels i, e, a, o, and u (phonetically perhaps [y] or [yu] in the pertinent
environments), these sounds can freely transcribe foreign dentals or
retroflexes in the ONWC sources. They rilray have been phonetically
"post-dental" in some way, but were probably not as retracted as the phones
standing before 6 and ii. Elsewhere, this set. of sounds in the main transcribes
Indic dentals. The behavior of this series of initials in the BTD materials is
similar to that found in the ONWC data, and I project it backward to BTD
unchanged. In syllables having oNwc 6 and ii, I shall posit BTD *-r-, which
is to be read as a syllabic feature of retroflection or rhotacism rather than as
a discrete medial consonant.
ONWC l- is used to transcribe Indic l, I , and r. It can be projected
back to BTD unchanged.
2.3 The ONWC Sibilants were ts-, tsh-, dz-, s-, and z- (= QYS ts-, tsh-, dz-,
s-, and z-). ONWC s- is very common in the BTD materiirls and most often
transcribes Indic s. ONWC ts- (or possibly dz-) occurs in the following
example:
T 224.470.1 ffiIEF]# gan do ho tsin: - dzin: Skt. gandhahastin
ONWC dz- definitely appears in the following:
T 152.42.2 +qF* kuo luo - lou dzin Skt. krakucchanda
An apparent example of tsh- is the following case:
T 280.446.1 HrrI tshei- ho Skt. saha-
This, however, raises the question of the curious role of the syllable fiin the early transcriptions. Compare the following:
T 362.300.1*+A *tshei- gat
T 362.300.2 HlF.* ri ia- rshei-
Skt. svagata; P. sagata
Skt. $eya
T 362.300.3 *+6J4 tshei- kuo dzim Skt. -samkusum[itabhyudgatal
In addition, at T 362.77.1 we find a long list of names in which ffi
-875-
W. South Coblin
occurs frequently, e.g.
*ffi ( '{R.>) W.@* tsuo dei po T uei- tshei-
i#ffiEq[# pui- pa- do ia tshei-
( ffi >) f/^[EJ* si ?o tshei-
f tr!fl+* si li- gun tshei-
In the Skt. version of this text, all names in the list are given in the
genitive singular, and it would seem that ffi is used here to transcribe the
masculine genitive singular ending -sya. This ending is realized as -ssa in Pali
and many of the Prakrits. It would aPpear, then, that in the early Buddhist
transcriptions * $ras to be pronounced something like "sa," a reading which
is not attested in received dictionaries and glossaries, so far as I know. In the
Jiyun ft ffi and other later lexica, H is said to be used as a variant writing
for a word sat (ONWC sat), meaning "to scatter, throw away; banish;" but it
seems questionable whether this reading could be involved here.
ONWC z- appears in two examples:
Skt. jyotipprabhasya
Skt. puqpadhvajasya
Skt. simhasya
Skt. Srikutasya
T 4f8.917.3 EftA iuo zuin
T 474.524.3 lR.6J pa zuin
Skt. yojana
Skt. papiyan
Pelliot (1933) has devoted a special study to the second of these, in which
he reconstructs the underlying Prakrit form as *paven (1933:92). The ninth
cenrury glossist Huilin ffiI# , in notes discussed by Pelliot (pp. 88-89), states
that Gl in the text was earlier written @ luen:,- (QYS riwen:,-). Pelliot
rejects this note as rank speculation, forwarded entirely in response to the
phonetic difficulties posed by E.I, with its offending initial zu-. He then cites
a number of examples where ffi zue (QYS zjwe) and E (which Pelliot reads
as QYS dua: and everywhere emends to ffil ) render foreign ve, vai, etc., in
order [o demonstrate that for some reason Chinese zu- could indeed render
foreign v- syllables. Now, E occurs in the BTD data (see section 2.8 below),
-876-
BTD Revisited--A Reconsideration of the Han Buddhist TranscriPtional Dialect
where it renders foreign ve-, etc. But, contra Pelliot, I believe its correct
reading here is ONWC huie (QYS hjwiea). Examples of m do not occur in
our own BTD data. Yu (1984:317-318), however, lists several cases transcrib-
ing foreign va and vai. And Prof. Harrison's T 624 data yield the following
example:
T 624.357.3 frEH zue lam Skt. vairambha
In any case, if one would, like Pelliot, carry out wholesale emendations in
those sources where it does appear, then why not emend ffi to ffi huie,
rather than vice versa, and solve the problem this way?! As regards Huilin's
editorial gloss, Pelliot.'s summary dismissal seems on the face of it rather
arbitrary. But in his favor we can cite evidence of a Yery different type.
In sound glosses of the Eastern Han period, QYS z- interchanges
primarily with s- and with other sibilants such as ts- and dz-. However, there
is an interesting body of exceptions to this (examples cited from Coblin 1983:
51, with ONWC values substituted for QYS forms):
Zheng Zhong 72
Xu Shen 480
tt42
Zheng Xuan 165
AII hun
ffi ?uen-
tE zuan
'HJ zuln
SJll zuin
trJ zuin
E uan, 'Iuen
fr kuiin
The z- initial syllables in these glosses seem to have an affinity for words
with guttural fricative initials followed by the vowel u, and it is noticeable
that our problematic word @ occurs in two of the examples. Perhaps the
situation in the Eastern Han dialects reflected here was similar to that
otrserved in our slightly later transcriptional language. To account for this
state of affairs, we shall tentatively suggest that ONWC zu- in such cases be
derived from earlief *w-, followed by the vowel *-i- Plus some other vowel'
As will become clear in section 2.7 below" our BTD *w- is envisaged as
-877-
W. South Coblin
having had t\Mo different allophones, [n*] and I r "]. The shift to later
zv- rnay have been a fronting process similar to that. which occurred widely
in modern northern dialects when earlier *hy- shifted to later cy-. According
to this theory, @ would be reconstructed as BTD *wiin. The word ffi would
be restored as BTD *wian, and m would be *wie.
Pulleyblank (1983:86) has expressed doubts about the use of later z- to
transcribe -j- i, yojana. I am not sure whether these are really justified. But
if they are, then perhaps \Me can do Huilin one better and suppose that. trl
in this form is a corruption of ! iuin! As will be outlined in section 2.5
below, this could be reconstructed as BTD *juin, a step which might help to
account for the transcription of Indic -j- here.
In summar/, we tentatively posit two earlier origins for ONWC z-. In the
majority of cases we can perhaps continue to derive it from *z-. In certain
others we shall restore it as *wi- followed by some other vowel.
2.4 Of the ONWC retroflexes, ts-, tsh-, d7-, and c- (= QYS t$-, t$h-, dz-
and s-), tq- is not attested in the data at all. For dz- we have one example:
T 362.300.3 *y44 tshei- kuo dqim
ONWC q- transcribes Indic g, s, Sr and
T 13.236.3 i4F5 sa mon
T 224.4?7.r lfiD ki qa
T 224.434.1 iffi4 bas ea
T 280.445.1 W,ib tou qa
ONWC tsh- transcribes Indic ks, e.g.
T 418.913.3 fUfU tphat li-
T 224.434.2 Effi t$htn:,- dei
T 224.458.1 [,jffi ?o tehuk
T 458.438.2 Effi dot tshin-
-878-
Skt. -samkusum [itabhyudgata]
6, e.g.
Skt. Sramar.ra; Gd. gamapo
Skt. isana
Skt. bimbisera
Skt. tucara
Skt. keatriya
Skt. ksanti
Skt. akqobhya
Skt. daksina
BTD Revisited--A Reconsideration of the Han Buddhist TranscriPtional Dialect
If the syllables in which these initials occur are reconstructed with the
retroflex feature, *-r-, the entire series can be treated as retroflex allophones
of the sibilant series.
2.5 ONWC syllables beginning with the vowel i- form a group which is
assigned ro a single initiat class, ji- (i.e. the yusi ffi@ initial) in the tradition-
al QYS. Three different uses of such syllables are found in the early tran-
scriptions. The first two of these can be illustrated by the following examples:
I. Transcriptions of Indic Y
T 224.427.3 B,=rl1ii mo ho ian:
T 418.917.3 ffiA iuo zuin
II. Transcriptions of Indic c, j, 6, and s
T 224.4?2.1 ffii+f|J iam bu li-
T 280.445.3 ifrE luot it
T 280.446.1 trfSre huie lou ian
T 458.435.2 iUH ka iap
T 184.464.1 [ElH ?o i
T 196.148.1 {fiEtr iuat dou don Skt. quddhodana
T 152.9.1 ffftJ ia li- Skt. jeli
T 362.300.3 ft'tS{^EE6S iuo lou kuo lou- tshei-
Skt. strraktrla
Interpretation of this material is complicated by the fact that Indic intervocal-
ic y can change to z in some of the Prakrits. And it is also noteworthy that
one and the same Chinese character can represent either y or one of the
consonants such as c, j, etc., e.g.
H
T 224.429.1 trt!fr buam- ka i
T 224.437.1 {H*H ?u bo i
Skt. mahayana
Skt. yojana
Skt. jambudvipa
Skt. vajra; P. and BHS vajira
Skt. vairocana
Skt. kasyapa
Skt. asita
Skt. brahmakayika
Skt. upasika
-879-
W. South Coblin
T 280.446.2 tff*.ffi iui i lo
H
T 224.429.1 yqH kuo ik
T 224.434.1 iEE ka ik
Skt. vicara
Skt. kausika
Skt. kdyika
EE]d
T 224.439.3 H iam Skt. yama
T 196.157.1 +AHtr kuo iam ni Skt. kausambi
To account for the varied behavior of syllables of this type, I suggest that
they be reconstructed with an initial semivowel, *j-, which can be assumed
to have had strong frication, resembling Z. Sounds of rhis type have been
described for Chinese dialects of Hainan by Ting (1986:6), and Woon (1987:
12). Among non-Chinese languages of East Asia they have been observed, for
example, by Matisoff (1973:5-6) in Lahu, a Tibeto-Burman language, and
Svantesson (1988:69) in U, an Angkuic language. Since the use of {'j- to
render foreign affricates and fricatives is particularly common in compounds
where it occurs intervocalically, we can suppose that the assumed frication
was especially prominent in this environment.
Where l'.1- is reconstructed, following *i can be deleted as redundant,
except, in syllables where it serves as main vowel, thus:
* *j"p I iap H *j* > iam X "'ji > i E *jik > ik
The third type of transcriptional application for QYS ji- is represented
in the following examples:
T 626.394.2 ltrBffi iui mo lo Skt. vimala
T 626.404.2 IEHIt iui su do Skt. visuddha
T 196.161.2 ffiqEffi iui ia lie Skt. vaisali
T 196.163.2 ffiffi iui uei- ' Skt. vipasyin
T 474.535.3 [qlffiffiffi ?o iui la d6i Skt. abhirati (= p1,. *avirati ?)
-880-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
T 54.848.2 )Affi'ffitr, ka iui uei- tou
'I 624.363.3 iEtsffiitr iui ma la ndt
Skt. kapilavastu (= Pkt' *kavila- ?)
Skt.. vimalanetra
Skt. ratnaketu
In these examples, ONWC iui (= QYS jiwi) transcribes Indic syllables
beginning in v, whether original or as Prakritic developments from other
labials. Here it seems best to assume that initial iu- was originally *u-, which
broke to iu- by oNWC times. l#, frff, and iH would then all be restored as
BTD *ui. It should be noted, however, that this conclusion does not necessar-
ily give us license to reconstruct ONWC iu- as *u- across the board' On the
contrary, there are cases where it should probably be derived from *jt-
instead, e.g.
T 196.148.1 1nEFtE iuat (< *juat) dou dan Skt' suddhodana
T 224.429.3 ffilX iuat (< *juat) tlha Skt' yakea
T 362.300.3 ft.B{EESS iuo (< *juo) lou kuo lou- tshei- skt. surakula
What should be done in each case must be decided on the basis of
textual evidence, which is unfortunately not always available.
2.6 Among the oNWC palatals, the initials ts-, dz-, s-, fl- (= QYS ts-, Z-,
S-, nL-) generally transcribe Indic c, j, S, and fl respectively. ONWC tSh- is
absent from the data. These consonants can be projected back to BTD
unchanged.
In the following example a syllable having ONWC ts- probably
transcribes foreign k:
^r 224.467.3 ffiffiffi ( &fl >) &# lo lin no tse to
In this case we can posit BTD *kie for ffi and assume that e:rrlier *k was
palatalized when followed by *-ie. In taking this step we build on a general
theory regarding palatatization of early Chinese velars, to be presented in a
forthcoming study by Professor Axel schuessler (schuessler Ms.). BTo *k in
this position may already have been phonetically highly palatalized,
-881-
W. South Coblin
accounting for the following case \Mhere it probably stands for Indic c:
T 602.170.2 ffi*. piek tse Skt. pratyeka; P. pacceka; Gd. prace'a;
Maharaqtri padiekka; Ardhamagadhi
patteya (On the Gandhari form, see
especially Norman 1983:96-8' )
By Tang times ONWC -e in ffi and jf had merged into a general -i final.
The same change had occurred in the Central Plains dialects. In the S dialect
of sixth century Jiankang, k- was still a guttural in words such as ]f and ffi
(see Shi 1983; Coblin 1990; 1991b).
QYS dz-) occur in several BTD
can begin with the following
Words having ONWC 2- (= Karlgren's
forms and present special problems. We
example:
T 626.393.1 #B --- bii P. jhapita
The character ff has two different QYS readings, i.e. duo, dZja (= oNWC do
a1,d ia), and from outside the QY [radition there are further readings: da
and Sjwo (= ONWC da and So). The reading ONWC za could be the valid
one here, and we could perhaps retain it for the BTD period. However, since
we are using ONWC as our point of departure, Itre should note that #
occurs in the works of all three major ONWC period translators, and it is
consistently used by them to render Indic da and da. It would thus seem to
be the ONWC reading de (= QYS da) which was the generally current one
in the northwest dialects. It is possible tha[ BTD, which we suspec[ was a
Central Plains dialect rather than a northwest one, had a palatal-initial
reading here. But to posit a palatal proto-form for the ONWC syllable dzi
seems unwarranted.
The syltable fff zuit occurs a number of times in the data. Let us first
consider the following examples:
-882-
Revisited--A Reconsideration of the Han Buddhist Transcri
T 224.435.1 fr,'{,fuW tou zuit do Skt. tupita
yu (1984:28a): ffi?frEtEFrj ?ou,- Zuit dom iui ho Skt. usijdharmaviha (sic!)
Here frfrf renders skt. git and gij. I believe Yuchi (1985:40) is correct in
suggesting that in both these cases N may be a scribal error for ffi guit. We
can perhaps exclude these two examples from consideration.
Note now the following:
T 313.758.3 ffJflii no- zuit Skt. nayuta
Flere, fff transcribes foreign -yut-. Yu (1984.280) mentions a similar case,
where ffi renders -yut- in ayuta, but he does not quote the example in full'
He plausibly proposes that -yut- was probably *-aut- in the underlying
Prakrit form. This suggests that our ONWC form Zuit can be posited for
BTD as well. A related example can be cited from Prof. Harrison's materials:
't 624.36?.3 f[fEiEfU ( ff , ) ffift ruo do po li Zuit dsi
Skt' vrataParisuddha
Here too it seems probable that Chinese fili 2uit represents Prakrit *-lut',
corresponding to Skt. -sud-.
This, however, brings us to our final and most problematic examples:
T 602.173.1 fibffi' zuit dia. Skt. vidya; Gd' vija, P' vijja
T 362.300.3 [qlfif?fi,Fu ]E# ?o Zuir t6e do gat luo skt. avidyandhakara
Here it seems that ffq transcribes Indic vid-, but why a syllable beginning in
Z- would have been chosen to do this remains unclear. A possibility is that
ffq in these examples is a copyist's error for some less familiar graph, such as
$ uit (< BTD *wit) or ffi huiit. Another approach would be to reconstruct
ffi as BTD *uit and assume that this syllable type regularly yielded later zuit.
Perhaps this Zuit arose first in compound internal position and later every-
where. We might even suppose that early BTD filf *uit (> later BTD *Zuit)
became standard as a rendering for vidya and was later retained for derived
-883-
W. South Coblin
forms such as avidya. These possibilities are offered here for further
consideration.
Finally, Yu (1984:280) gives a very interesting form, fiEfit, transcribing
Indic yasas (See T 196.149.1. Edgerton 7953.II:445 also gives ya(a as a
form of this word.). In the well-known sense "snake" the graph ffi has the
QYS reading dzja (= ONWC 2a), with obscure variant tha (= ONWC tho). It
is QYS dzja which seems to lie behind modern dialect readings, such as
Pekingese she, for "snake." In addition there are rare, literzrry readings, QYS
jia and jie (= ONWC ia and ie) with other meanings. The literary reading ia
would solve our problems here, for we could assume a form *iu iu, in which
the second initial *'.1- had strong frication and resembled 2. But it seems
almost perverse not. to suppose that the transcriber intended ffi here to be
pronounced in the common and universally known reading of the word
"snake." This leads to an interesting possibility. Perhaps the "basic" or
"etymological" reading of "snake" \.vas once indeed *ja > ia. From a northwest
perspective such an assumption is in fact not inappropriate, for the Jiyun
tells us that the word for snake in the Guanzhong (i.e. Chang'an) area was
in fact pronounced as QYS jia ([+.ffi!|] EE+EEEffiEfE). Now, early Chinese
vernaculars prefixed the syllable ft lou: to the names of familiar but disliked
or feared creatures (see Norman 1988:113), and the form ZNt, i" in fact still
current in the Fuzhou dialect (Hanyu fangyan cihui, p. 62). We might specu-
Iate that in early times there existed the vernacular compound fif,! *lou ja
"snake," which was phonetically realized in northern dialects as [louZa]. (W.
make the assumption here that the distinction between *j- and *z- would
have been neutralized intervocalically.) Later, this compound would have
been "unetymologically" reduced to a monosyllabic word *za in many vernacu-
lars. Alternatively, of course, one might simply reconstruct {E as BTD za and
-884-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
leave its function in the frrst syllable of yaSas unexplained. In any case, in
the last. analysis it seems safest to restore ONWC 2' as *2- in BTD, unless we
find evidence to the contrarY.
2.7 The traditional yusan ffi = category (QYS j-) corresponds in oNWC
to syllables beginning with vowels other than i-. By far the majority of these
have ONWC initial u-. The most prominent exceptions to this are the
literary enclitic particles * -.: and ffi -an.
ln the BTD data ONWC syllables beginning in u- are used primarily to
transcribe foreign v, whether original or derived, in the Prakrits, from earlier
labial stops, e.g.
T 150.875.3 i[gffiru ka lo uar Skt. kulapati (= Pkt. *kulavati ?)
T 224.447.2 nffifuruffi [sa ka uat lo Skt. cakravartiraja
't 224.468.2 B;EJffiElffi mo ho iui uat lo skt. mahavaipulya (= Pkt. -vevula ?)
T 280.446.1 WHEI ?ut ton uat Skt. uttarakuravah
T 224.425.2 W*E sot un fla: Skt. sarvajfra
T 280.446.1 #+E put uo dei- Skt. purvavideha
T 362.?17.1 flftr+ no iui uo Skt. nagabhibhu (= Pkt. *nagavivu ?)
Interestingly, such ONWC syllables could only be used to transcribe
"compound-internal" -v-. For absolute initial v-, other transcriptional strate-
gies were utilized (cf. sections 2.5 and 2.8).
A less common application of ONWC u- syllables rras to render
intervocalic -h-:
T 224.435.1 t€Titr iui uo phan
T 184.462.2 W* lo un
T 474.523.2 ffiz; lo un
Yu (1984:319): E uat Skt. halh
Skt. blhatphala; cf. P. vehaPPhala
Skt. rahula
Skt. rahula
ffi uat Skt. hul.
-885-
W. South Coblin
And we also have the following interesting example:
T 362.300.3 'rEW t ftr I TiA suo ia [iui] uo qe Skt. strryaghoga
The character tfi here seems to be a scribal accretion of some sort.
Taken together, these examples suggest that ONWC syllables beginning in
u- may have had an initial consonant of some sort in the BTD period. A
likely value for this sound would seem to be the labio-laryngeal [n-1' In our
reconstruction we shall transcribe this initial as *w- and assume that, in
parallel with BTD *,;-, it had rather strong friction. The following vowel -u-
can be deleted as redundant, except in syllables where it is the main vowel, e'g'
ffi *wat ) uat fi*wu)uln dealing with transcriptional forms where Indic -h- is rendered, it might be
useful to add a phonetic form in brackets, e.g.
ffiE *lo wun [fi*un] > lo un
,HW tftr I +i4 *suo ja wc [h'c] 513 ) suo ia uo qe
The graph ffi -an aPpears in the following form:
T 198.180.3 frffi#R ni -an fla: dei Skt. nirgranthajrata; Cf' P' nigantha;
Ardhamdgadhi niyantha (Pischel 1981: 269)
The let,ter y in the Prakrit form represents the ya-sruti of the Prakrit
grammarians, which, according to Brough (1962:86) "may be considered as
marking only the separation of syllables, presumably without glottal closure'"
Perhaps enclitic or "non-initial" ffi was simply -an in the BTD period. From
the standpoint of the northwest dialects, at least, the two QYS readings for
,E , i.e. ?jan irnd j- (= oNwC ?an and -an) may have been merely posi-
tional variants, with the former occurring in initial position, while the latter
appeared enclitically. (For a similar suggestion, involving QYS reconstructions
rather than dialect materials, see Pulleyblank 1986:8.) It is interesting to note
that. in one Tibetan transcriptional text of the Shazhou period (i.e. Text TD,
-886-
rahula
suryaghoqa
TD Revisited--A Reconsideration of the Han Buddhist T
lines 58-59), enclitic ffi is transcribed in the following ways: ?in, {in, ?en,
yen (contra Takata 1988:276, who writes ?yen for the fourth example)' Here
we see that in the shazhou period enclitic ffi was usually read with a glottal
stop initial, though this apparently could be dropped' The deletion of ?- was
perhaps determined by the speed at which the text was read or the amount
of stress placed on the final particle in a particular instance.
2.8 The ONWC gutturals, k-, kh-, s-, and h- tend to t,ranscribe Indic k, kh,
g - gh, and h respectively. They can be projected back to the BTD period
unchanged. Here we may add that the combination hu- is used in several
examples to transcribe Indic initial v-, e'g'
T 280.446.1 trffiru huie lou ian
Syllables beginning with ONWC ?'
with plain vocalic onset, e.g.
T 602.163.3 frW ?an Pan
This initial can perhaps be retained for BTD'
The last ONWC guttural with which we must deal is t-. From the stand-
point of its behavior in the early transcriptions, examples of this consonant
can be divided into two tYPes:
TypeA.Syllablesin'which?-isimmediatelyfollowedbyoNWCu'
where u is invariably the first member of a diphthong. [n the BTD data
these syllables transcribe Indic v, whether original or derived from earlier
labial stops, e.g.
T 150.877.1 'lFEigl suo"da luon Skt.. sroEpanna; cf. P.
sotaPanna (= Pkt. *sotavana ?)
Skt. sattvaT 224.433.1 ffifU sot 'Iuo
T 280.445.3 ifi8 ?uot it
Type B. Syllables where
Skt. vairocana
usually transcribe foreign syllables
Skt. anapana
Skt. vajra; P. and BHS vajira
'v - is followed by vowels other than u' Here
-887-
W. South Coblin
Indic g, k, or h are transcribed. The following is a
BTD examples cited in our earlier studies:
T 13.233.2 fi|S ?o rom
T 750.877.1 IqlflJfr ?o na'rom
T 150.877.1 ffiWA sie do tom
T 224.4?3.1 tfi ra3
T 313.753.3 BEF$, ma tou lak
complete listing of the
Skt. agama
5K[. anagamrn
Skt. sakrdagamin
Skt. ganga
Skt. mahoraga
Beginning with examples of type B, our original conclusion (1981; 1983)
was that later 1- should here be reconstructed as BTD *g- and that the final
form, for mahoraga, should be considered exceptional. Pulleyblank (1983:82-
83), takes a very different tack. He suggests that in the underlying language
of the Indic texts intervocalic -g- had weakened to -'t- (cf. Brough 1962:86).
QYS r- should then be restored for the BTD period as *'I-. The transcrip-
tion of ganga would then be exceptional and would be accounted for by him
by supposing that it "probably reflects the influence of the more conservative
upper class dialect."
My initial reaction to Pulleyblank's suggestion was incredulity. Indic
intervocalic -g- is fairly well represented in the BTD data, where it is usually
rendered by QYS g- (= ONWC g-). The northwest Prakritic shift to -'v- can
therefore not have been a primary feature of the languages underlying the
Indic originals of the majority of BTD texts. However, a closer look at the
material throws a different light on the matter. The examples in question
here (i.e. the first three in the list above) all occur for the first time in the
transcriptions of An Shigao * tr H , a Parthian who was the earliest BTD
translator (see Ziircher 1959:32-34). No such cases appear in the other Han-
period BTD data. And, in addition to this, though An Shigao's corpus is
rather small, it seems significant that it contains no cases where Indic
-888-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
intervocalic -g- is transcribed by QYS g-. Though, as just noted, these
appear in the works of other Han translators, they are absent from An's
materials. Thus, Pulleyblank's theory can be applied here if we assume that
An's lndic originals were written in a northwest Prakrit and/or his oral
rendition of the texts reflected the pronunciation of such a language. The
texts and/or pronunciations of the later BTD transcribers would presumably
have been of a different type. But what of the problematic form ganga?
Pulleyblank's appeal to an otherwise unknown "upper class diaLlect" seems ad
hoc and unacceptable. But the data themselves suggest an alternate and
plausible solution. The fact is that. ONWC r- in the Chinese transcription of
ganga is in absolute initial position, while in all other forms in our list it is
found in "compound-internal" position, which in effect means that it is
intervocalic in the cited Chinese compounds. (We have no evidence on its
form when it was "compound-internal" and preceded by a consonant, such as
*-n or *-t.) This allows us to assume that in BTD the initial in question
had two phonetic forms. In word-initial position it was [g], while its
intervocalic form was ['y ]. This is a well-known pattern of allophonic
distribution in various languages, north German dialects being a commonly
cited example. The BTD initial, which is in complementary distribution with
our already reconstructed *g-, can also be phonemically tr:rnscribed as *g-,
with the phonetic form added in brackets where this is felt to be useful, e.g.
WW.#h * mo gou [rou] lak ) mo tou lak Skt. mahoraga
tE *g"u > 'Yeu
ln addition to the BTD forms cited
several more:
ft 1ot, for Skt. gat
A rop, for Skt. gup
Skt. ganga
above, Yu (1984:311-312) gives
-889-
W. South Coblin
Both examples come from Yu's Zhiqian corPus. Since we are not given the
environments in which they occur, it. is difficult to determine whether or not
they contradict the hypothesis outlined here. But the following case from our
own TK data is clearly a counter-example:
T 752.42.2 +hJfl\ffi+tr, kuo no 'Iom mou (< *mu ?) ni Skt. kanakamuni
lntervocalic k probably underwent the very common Prakritic voicing to g in
the Indic text underlying T 152, as for example in:
T 752.27.1 tsffi mo gat Skt. makara
There is no evidence that it was here further reduced to -'v- as in the north-
west. Prakrits. However, it should be noted that T 152 belongs to those texts
which were translated in the Wu area rather than in the Central Plains.
Perhaps in this region QYS r - \tras phonetically [S] in "compound-internal"
position in the TK period. If need be, the same argument could perhaps be
applied to Yu's Zhiqian examples, for Zhiqian's transcriptional reuvre is also
supposed to stem from Wu.
Turning now to cases of type A above, we find that these are essentially
different from those of type B. The initial they reflect bears a certain resem-
blzrnce to our *w- in that it can transcribe foreign v. But it differs from *w-
in that it can freely render either absolute initial or "compound-internal" v.
And also important is the fact that it never transcribes foreign h. In ONWC
syllables it occurs exclusively before the vowel -u- as the first member of
diphthongs. It does not appear before the oNWC absolute final *-u (= QYS
-jrr). A guess would be that it was phonetically [r"1. We might suppose that
the element ['y] here was nof satisfactory for transcribing foreign intervocalic
-h-, whereas our laryngeal *w- [hw] could serve this function. Since I r "]
occurs exclusively beforl Lt, and this u is never the main vowel of the
syllable, we can delete u as redundant. Once this step is taken, we note the
-890-
BTD Revisited--A Reconsideration of the Buddhist Transcri
interesting fact that [r"] and *w- are in complementary distribution, with the
former occurring only before the ONWC vowels 6, a, o, o, e, and ii' and the
latter in various other environments. within our BTD system, the two can be
considered allophones of the same phoneme: */w/ th*1, [r*1. Adopting this
solution, we can in conclusion diagram the proposed developments from BTD
to ONWC in the following waY:
BTD ONWC
* / S/ [S] > g- Before front vowels and *u
[S] > 1- Elsewhere
[r] > r-*/w/ [.y.] > .ru-
[n*] > bu=-
III. Finals
The BTD finals will be treated in groups corresponding to the wJ rime
categories posited by Ting (1975). Departures from Ting's system will be
discussed where necessary. At the beginning of each section the pertinent
oNwc finals will be listed, preceded by the corresponding QYS forms in
square brackets. In certain cases, possible Central Plains variant forms will be
suggested in round trrackets.
3.lTheHaiffiGroup.ThisgroupconsistsofthefollowingoNwCfinals:(1) [-ai] -ei - (-oi)
(3) [-ai] -ei
(2) [-uai] -ei - (-uoi)
(4) [-wii] -uei
Final (1) is rare in the early northwest materials. Where present it tran-
scribes foreign e. we have tentatively restored it as [ei1. Phonemically, it
could be written as *-oi or *-ei in our ONWC system. This final is somewhat
-891-
W. South Coblin
more common in the S data, and there too it. transcribes e. In the XZ d,ata it
renders foreign ai and ay (Shi 1983:44). [n the YJ data it transcribes ay
(Coblin 1991b). [t would appear that in the early medieval dialects final (1)
had at least two realizations, one perhaps as [oi], representing the Central
Plains, and the other as [ei] or the like, found in certain other regions.
Final (1) appears twice in the TK text materials:
T 362.300.1 477 pu- ncr:
T 362.300.1 Sffffi non dei:
For the late Han period we have:
T 280.445.2 5{EiE tsam bei- 'Iuon
T 624.363.3 f[[€i&ttj ( ff , ) ffift ruo do po
Skt. ptrrqa
Skt. nanda
Skt. *campakavarna
li zuit dei-
Skt. vrataparisuddha
These examples point to a vowel which may have been similar to Indic
short a. I suggest that this vowel be restored as *a. The T 280 form might
suggest the presence of a final gut.tural in {* , but this is problematic. The
WH Chang-an poets occasionally rimed such qusheng syllables with rusheng
words in -k, but the EH Chang-an poets never used such rimes (Coblin
1986; 1987). It is possible that the transcriptional form has been reduced
and originally contained another syllable representing foreign ka. This inter-
pretation is supported by the T 624 example, where Chinese {t dei- (< *dr)
sternds for foreign -dha.
The entire group can be reconstructed as follows:
(t) *-" (2) -ua
(3) *--ra (4) *-rua
Sample Reconstructions:
I. X *1" > lei ff *tsa ) tsei:,- ffi *he ) hei:71 *ne ) nei: f$ xda ) dei:
EE *the > thei- H *p, > pei- f* *b" > bei-
-892-
BTD Revisitedl--A Reconsideration of the Han Buddhist TranscriPtional Dialect
ft *hua > huei ffi *hue ) hue,i: [f *khua > khuei
III. fr *kra ) kei- E *mta ) nllei \[ *dzro ) dzei 'lE *krua > kudi-
3.2 The Zhi f Group. This group comprises the following oNWC finals:
(1) t-jiJ -(i)a (2) t-jil -i (3) [-iwi] -ui
Final (1) occurs in the following TK examples:
T 474.522.2 #W+ ( Etri&' ) i&ffi kia ia kim po lo
T 474.522.2 ihH++ bii lo dia
Skt. keSakambala
Skt. vairati-
yu (1984:292) lists several more cases, all from the Zhiqian corpus:
ffi tqia Skr. ci ffi tqia Skt. ji f$ die skt. ti (sic) Sft qia:,- Skt' eya
(At 1984:313 Yu has not ti but [i for the third case here. The latter is
probably the correct form, corresponding to our T 474 example above')
This group has frequent contacts with the Hai group, and the distinction
between the two is therefore somewhat problematic (Ting 1975 2O4) '
contacts with zhi E and zhi EH are also common, but there are no
interchanges with the Jie group (section 3.9). The oNwc form *-(i)e can
perhaps be retained here, with the assumption that *-a was phonetically
higher after palatals and *i.
The group can be restored as follows:
(1) and (2) *-(i)e, (3) *-uie (> *-ui)
Sample Reconstructions:
III. * *kia > ki, EE *g,, ) gia Z *tsa ) tsa S[' *thia > thia: J'X *ja ) *ia:
{D) *zia ) zia: H *fr, > ia: ffi *dla ) dLa: ! *dzrie > dlia:
ffi *tsria > tsia E{ *srie ) sia:,- ffi *di, > dia i6 *dia > dia-
* *-. ) -e: (enclitic only) 4 *phia > phi ffi *bia > bi-
fl1 *kuia > kui:
3.3 The You @ Group. This group contains the following ONWC finals:
(1) [-au] -ou
-893-
W. South Coblin
(2) [-jau] -u (3) [-ji"u] -iu
Final ( 1) is well attested in the Han data, where it transcribes Indic u in
75Vo and, Indic o in 25Vo of the cases, e.g.
T 224.4?4.2 dgt$ffi ka lou lo Skt. garuda; cf. P. garula
T 280.446.1 EBre huie lou ian Skt. vairocana
T 626.404.2 ffi[fiffiffigfr lo do no ke dou Skt. ratnaketu
T 418.903.1 FtrEFf[Ftr non dou ?uo non Skt. nandopananda
Final (2) commonly transcribes tr, and occasionally o, in the BTD
materials, e.g.
T 13.241.1 lhE bii khu Skt. bhikqu, Gd. bhikhu, P. bhikkhu
T 224.440.1 HWff\ Su: do uei- Skt. suddhavasa
T 373.753.3 itgBffi ka lu lo Skt. garuda
T 626.404.2 WHIE iui Su do Skt. visuddha
T ?13.75?.3 ts{ S, mo hu lak Skt. mahoraga
For final (3), Yu (1984:296) lists the word 6[ iu transcribing foreigtr y,.,
and more rarely, yo. A case of the latter type from our data is the following:
T 474.527.1 fire iu ian Skt. yojana
For all finals of this group, the ONWC forms can be retained for the
Han period.
Sample Reconstructions:
I. $ *sou ) sou: ff*mou>mou-ffi*kou )kou- ! *khou ) khou:
JE *gou > ?ou: fE *pu > !ou: ft *tsou ) tsou: $ *dou ) dou-
{[. *phou > phou- fS *lou ) lou
E *mou ) mou: S[ *phou > phou: cf.
# *mu > Early ONWC mu ? ) mou $, *mu ) Early ONWC mu ? ) mou
m. fi *gu ) gu /L *ku ) ku: H *su ) iu:. {$ *su ) su g[ xju > iu
E *j, > iu: ffi *zu ) zu- * *thu ) thu: fd *tshu > tshu ff *tsu ) tsu
-894-
BTD Revisited--A Reconsideration of the Han E"ddhit! fgrygl1ptig!3}-Dialect
ff*dzu > dzu: f€ *sru > $u i+ *bu ) bu E *tu > tu ffi *lu > lu
ft *h.. > hu 6 *pr ) pu: T *p, ) pu- H *pu ) pu- # *ps ) Early
ONWC mu ? ) mou f *mu ) Early ONWC mu ? ) mou X *wu > u-
f; *wu > u: Fd *pi, > piu # *miu ) miu fil-[ *kiu ) kiu 8l: *?iu ) ?iu-
3.4 The Hao ffi Group. This group contains the following finals:
(1) [-au] -au
(2) [-au] -au
For this group there is one BTD example:
T 224.425.3 YJtt) mu lji- Skt. [rayastrim6a, P. tavatimsa, Khotanese
Itavatrisa
There are many rime contacts with the Xiao group (see section 3'5),
suggesting that the two were probably rather similar'
Sample Reconstructions:
L S *kou ) kou-ffi *dou ) d.ou: H*tshou ) tshou: H *po'> pou:
ffi *kou ) kou ffi *go, >'Iou- fl *tou ) tou ft *lau ) lou:
il. @, *prou > peu f *m:ou > meu I! *khrou > khiiu: ffi *k:iiu > kau-
ft *grou > 'Iau- ffi *prou > peu- nE *mrou > miiu-ffi *dzrau > dLiiu
3.5 The Xiao fi' Group. This group comprises the following ONWC finals:
( 1) [ -jauu] -au
(2) [-jiaun] -iau
(3) [-ieu] -eu
Two forms from this group are attested in the BTD data:
T 280.446.1 ffiEFBffi tsau- dou mo lo
t- .. )6.b17.2 +Effi ( i9* , ) eig tsau dei sa sg7
Note that Skt. catu- yields ca'u- in the northwest Prakrits (see Brough
1962:299).
And from Yu Min's data we can add (1984:297):
Skt. caturmaharajika
Skt. caturdeSasamgha
-895-
W. South Coblin
ffi kau Skt. kau fffi gau
Final (3) occurs a number of times
foreign deva.
In ONWC the vowel e is assumed to have been rather high and also non-
front, perhaps resembling in some way the vowel e in Lhasa Tibetan (Coblin
1991a). We shzrll retain this vowel in our BTD reconstruction and offer two
hypotheses regarding its phonetic nature in this period. First, where *e
occurs in final position, before velars, or before the vowel *i, we supPose
that it was phonetically either monophthongal, or perhaps diphthongal with a
final "e-like" or "shwa-like" offglide, i.e. [e'] or [ea]. In all other environ-
ments, we suggest that *e was phonetically diphthongal and had as its second
element an "a-like" vowel or offglide, i.e. [e"] or [ea]. This peculiarity
allowed it to rime with finals having the vowel */a/ and also in certain
instances to transcribe foreign a or e with seemingly equal ease. Here one
thinks particularly of the yuan and yue groups (see sections 3.21 and 3.32
below), where, for example, j[ sen renders foreign sen or safi - samj, and ifl
net stands for net or nad. Finally, we propose that in the early Tang period
the vowel [6] was fronted to [i] in the diphthong [ea], resulting, try Shazhou
times, in complete merger of [ca] with already existing -ia-. This can be illus-
trated as follows:
BTD and ONWC Common Shazhou
Skt. gav, gau
in the form ;f d€u, rendering
QYS
-Jau'
-jiziua
-ieu
-janu
-jiap'
*-an
*-ian
-896-
BTD Revisited--A Reconsideration "f tht Hu" B"ddhitt Tt td
-lan
,r,-6g [eatl ---> -iat ---> -iar
Sample Reconstructions:
[II. ffi *kau ) k u ffi *g^u ) gau fl *g,.u > grau t *pau ) pau:
ffi*piau > piau ffi *mau > mau- iS *miau ) miau: I *?au ) ?au {*?iau
) ?iau ffi *th(i)au ) thau ffi *d(i)au ) dau: 6 *6s(i)au ) tsiau
zji *s(i)au ) siau: ffi *tsau > tsau DE *iau > iau ffi *dzau ) dzau
ffi *Sau ) Sau ,9' *Sau ) Sau:
IV. & *k€u ) k6u: fr CCu > OCu ffi *teu > teu- ffi *leu ) leu iE *dCu > deu
ull *k6u > k€u- ,fffi *dcu ) deu ffi *reu ) sdu EilE *teu ) t6u fffi *dCu ) deu
3.6 The iu ft Group. This group includes the following ONWC finals:
(1) [-uo] -o
(2) l-jwol -o (< -io ?)
(3) [-jul -uo
Final (1) appears in the following late Han transcriptions:
-len
-jat'
-jiiina
-iet
T 224.425.5 ,HP=F- suo bo ddi
T 224.425.3 #E bo sot
T 418.908.? =ffi,=#ffi sam miau: sam bo dci Skt' samyaksambodhi
T 184.468.1 [El{nts ?o no mo Skt. anomiya; cf. P. anoma
From the TK materials we have the following examples:
Skt. kah, Pkt. ko "who"
Skt. poeadha
Skt. mahagun-
Skt. pu-spadhvaja
Skt. subhnti
Skt. bodhisattva
T 206.519.1 ffi ko
T 14?2.1042.2 frffi Po- sot
T 362.300.3 iffiryB+ mon: ho gun
T 362.317.2 {#ffiH4ll p..i- pii- do ia
-897-
W. South Coblin
T 474.522.2 L|.,Rffi bii lo die
T 790.729.3 ffi#fr bo lin nsi- - noC
Final (2) is absent from our Han
examples of it (1984:300):
Skt. vairati-
Skt. varanasi
transcriptions. Yu Min cites two
Skt. dharmodgata
Skt. ghoqila
Skt. bghatphala; cf. P. vehapphala
Skt. purvavideha
$ huo, uo- Skt. upa-; Gd. va-*
more numerous in t.he Han data.
9il n, Skt. nya ffi qo Skt. qya
The first of these examples is attributed to Shemoteng ffi B m (Kasyapa
Matanga ?) and Zhu Falan 4{.}*ffi (Dharmaratna ?). It has surely been taken
by Yu from the form tffiEfi{[ Kauldinya, found in an addendum at the begin-
ning of the famous Sishierzhang jing Uq + : H ffi (f 784.722.7, n. 10).
However, the authenticity of this section, and indeed of the text as a whole,
is questionable; see Ziircher 1959:30, notes 61-64. We must handle it with
caution and perhaps with a certain amount of skepticism. The second
example is attributed to Zhiqian. Fortunately the full form has been given by
Yu in a different paper (1989:57): Bffi tnei-qo Skt. tisya. We should note
that the Gd. form corresponding to Skt. -sya would be -S?, while other
Prakrits might have had -ssa or -Ssa. We may guess that the Indic form
underlying Chinese here may indeed have been something like -qya.
Forms having final (3) can be conveniently divided into two groups,
depending on whether they rimed in the Yr, ,ft, or Hou {rt categories of the
traditionzrl Shijing rime system. Four examples of the former type occur in
our Han transcriptions:
T 224.477.1 EffirE dom muo gat
T 196.157.1 EEHFffi guo ei lo
T 224.435.1 tff+iffi iui uo phon
T 280.446.1 #+E put uo dei- - dei-
Here we can also compare Yu (1984:300):
uva-. Examples of "hou-type" finals are
-898-
T 150.877.1 ,HWla suo do 'Iuon
T 224.467.1 ,Hffi suo mie
T 418.917.? fIfl kuo li-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
They render Indic u in about 55Vo artd, o or au (= Pkt. o) in about 45Vo of
the examples, e.g.
Skt. srotdpanna; cf. P. sot-Panna
Skt. Sumeru
Skt. koti
T 196.747.3 Bffi+@flJ mo nom kuo li- Skt. mahanamakoliya
In the TK data we find the following examples:
T 1432.1041.1 Effi'R, dom muo tak Skt. dharmaguptaka
Ziircher (1959.338, n. 168) suggests a possible Pkt. form *dhamma-utta-
ka. A plausible phonetic realization of the combined elements would
presumably be *dhammottaka.
T 362.300.3 rHW ttfrl Ti4 suo ia [iui] uo pa Skt' snryaghoqa
T.362.317.1 ffit€+ no iui uo skt. nagabhibhu; Pkt. *nagavivu ?
T 362.300.3 ffiFfiffift tsan do sok iuo Skt. candra-surya-; cf' Gd'
suri'u for Skt. sDrya (Brough
1962:311)
The wJ rime group set up by Ting agrees fairly well with the values we
have posited for ONWC. Final (2) is entirely absent from the ONWC data
used in our recently published study of early northwest Chinese (Coblin
1991a) and may have had a value which was inappropriate for transcribing
Indic sounds. Our form -o is simply a backwar-d projection of the reconstruc-
tion set up for northwest dialects of ca. 600 A.D. But it is possible that the
ONWC value was -io [io ?], allowing for perfect phonemic interriming during
the early Six Dynasties period. Some corroborative evidence for this has since
come to light in texts not used in the 1991 study and is included in a more
recent paper (Coblin Ms. 2)
The Han material suggests that final (1) was o-like in second-century
-899-
W. South Coblin
Luoyang. The TK material, stemming mainly from Wu area (i.e. T 206, 362,
474, 79O) points to an a-like value for the Jiankang area. This feature of the
Wu pronunciation has already been alluded to by Pulleyblank (1962: [Pt. II]
214). Several more such examples from the Wu area are given by Yu (1984:
300):
tt tho: Skt. ta E do Skt. dha
And he also cites one northern example, attributed to Zhu Foshuo e (#FA:
H do- Skt. dha
As we have noted, final (Z) is absent. from the Han-time data. Yu Min's
TK form ffi , transcribing $ya, points to an element -i-, followed by an a-like
vowel, for the Wu area.
In the Han data, final (3), when stemming from the earlier hou catego-
ry,transcribes u and o. But when this final derives from the yu group, it. can
also transcribe foreign syllables in a. Pulteyblank (1983:83) questions the
validity of these examples of T , but later he seems to reverse himself on
this matter without further comment (1984:175).
Beginning with final (g) in the Luoyang dialect, it seems clear that we
must divide it into yu and hou types, the former having the qualities of
both a and o and the latter having no a-like timbre at all. The two varieties
could be restored as *ue and *uo respectively. Final (1) was primarily o-like
in Luoyang and a-like in Jiankang. In poetry of the late Han and WeiJin
periods it rimed freely with finals (2) and (3). Historically it is widely
thought to have had an a-like vowel in mid-Han and earlier periods. As a
compromise, we can perhaps restore it. as *-c. In a similar compromise, final
(2) can be reconstructed as *-Ic. In the north it may have been phonetically
something like [ii] or [io].
The set of forms suggested here leaves a gap in the reconstruction, for
-900-
BTD Revisited--A Reconsideration of the Han Buddhist T:
\A,e now no longer have a plain *o in the system. The reason for this, we
may suspect., is that an earlier *o, originally belonging to the Hou group'
had broken to -ou (= QYS -au). Syllables of this type, which we have placed
under final (1) of our You category above, transcribe Indic u-syllables in
TbVo and, o-syllables in 25Vo of the cases in the Han data. Their final clearly
was rather unlike our *-uo (derived from the Hou group artd corresponding
to QYS -j") and did not rime with it in Ting's WJ data' The system
proposed here would thus be as follows:
ONWC
> -ou
) -----> -uo
-o
-io > -o (?)
Pre-BTD
*-o
*-uo
*-uc
:1. -f,
*-ic
BTD
) *-ou
> *-uo
) *-uc
It will be noted that the finals postulated here for the "pre-BTD" period
are among those which are widely recognized as having rimed together fairly
freely in Han times (Luo and Zhou 1958), and ;t could be objected with
reference to our reconstruction that such interriming between -o and -e
would have been unnatural. However, as suggested elsewhere (Shao 1983;
Coblin 1986:127-8), this rime pattern seems to have been based on standards
derived from the works of the great Western Han fu poets of Shu f it'
whose dialect there may have been a true merger of the various finals
involved here. It is therefore unnecessary for us to assume such mergers for
the progenitors of BTD, ONWC, etc. Speakers of such dialects as these might
have fashioned their rimes according to tradition rather than on the basis of
their own sound sYstems.
Sample reconstructions:
-901-
W. South Coblin
I. ffi *?r> ?o-E *dc) do- fi *kc> ko: fi *Ur> no: ffixkc) ko
ffJ *t, > to 4[ *nc ) no tE *tsc ) tso: fr *so ) so- ;fr *p, > po-
7I *k > ko {iI tgc > 'Io i! *?c ) ?o H *U" ) Uo-
III. ffi *kic > ko- # *hic ) ho: ffi *sc ) so- # *s, > sr- iE *flc ) flo
ft *j, > isk *nic ) no: ffi *dic > dsffi,, *sric ) so E *gu, > guo
fE *gu, > guo- E *Ut, > Uuo f *w, > uo N *wc ) uo: ft *puc ) puo
Q *buc) buo: ffi *mu:) muo @ *khuo ) khuoffi *tshuo > tshuo
ffi *suo ) Suo ft' *iuo ) iuo fj *duo ) duo: {n *tshuo ) tshuo:
ffi *dZuo ) dZuo- tl *phuo > phuo-fff *muo ) muo- @ *tsh-ruo > tghuo
ffi *sruo ) suo- ffi *flun > fluo ffi *suo > suo
3.7 The Ge ffi Group. This group comprises the following ONWC finals:
(1) [-a] -" (2) [-u6] -ua
(3) [-a] -a (4) [-w.] -ua
(5) [ja] -a, -ia
Finals of this group are very common in the transcriptional materials
and are used to render Indic a and a. The graphs 1!! ka, f* kh., and {f[ ga
are widely thought to have been specially invented for use in transcriptions
of foreign words. They do not occur in Ting's rimed texts.
Sample Reconstructions:
I. * *to > to,fEI *go ) ro ffi *go >'Io- fr *tso ) tso: iE *po > po
4*bo > bo B *mo ) mo ffi *ko ) ko { *kho ) kho: fl *kuo ) kuo:
$ *dzuo ) dzuo: ffi *duo ) duo:
II. fg *kro > ka H *k-ro> ke- i4 *s.ro > sA m *mro ) ry12i
* *kro > ke T *gro ) ta: * *Uro ) 9?i fp *tsro > t$zi- tft *phro > phe-
ffi *mro ) mii:
[ *gruo > !uii; {f, *hruo > huii-
[l *k:uo > kuri S *k:uo > kuii: $ *wro > ?uA
-902-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
III. fE *2o (or *jo ?) > 2a W. *tsio > tsia
{ff *tsio ) tsia- ffi *sio > sia: ffi *zio ) zia- 1l *d1o > dila
$ *tsho > tsha
it! *ka > ka f* *kha ) kha fifl *gu > gu
3.8 The Zhi { Group. This group comprises the following oNWC finals:
(1) [-ai] -ei e) [-wai] -uei
(3) [-je'] -e (4) [-jwea1 -ue
(5) [-jien] -ie (6) [-jwiea] -uie
{7) [-iei] -ci (8) [-iwei] -uci
In this group we have made two changes from our earlier oNwc recoR-
structed forms. Final (5), earlier written as -iC, is now simplified as -ie' Final
(6) has been changed from -ue to -uie. These finals correspond for the most
part to Skt. i, r, e, and ai (= Pkt' e) vowel syllables in the data' e'g'
(3)
T 602.170.2 H+{ Piek tse Skt. pratyeka; P. Pacceka; Gd.
prace'a; Maharaggri Padi6kka;
Ardhamagadhi Patteya
^I 626.404.2 ffi[fiffiffiffi lo do no ke dou Skt' ratnaketu
(4)
Skt. uruvilva; P. uruvela
Skt. asankhya - asankhYeYa
Skt. maghi
Skt. sumeru
Skt. amiEbha
Skt. prasenajit
Skt. Varafasi
T 196.149.3 IHffiffi ?u ue lo
(5)
T 224.427.3 PEIfgtE?o ses gie
T 224.431.2 @+fi mo gie
T 224.467.1 ,Hffi suo mie
T 418.905.1 P,lffi[t ?o mie do
T 224.434.1 i&ffiffi po sie nik
T 418.903.1 i&ffiff po la sie
-903-
W. South Coblin
(6)
T 280.446.1 trt$tf huie lou ian Skt. vairocana
(7)
T 224.431.1 fflf[ffiffi dei ruo gat lo Skt. dipamkara
T 184.461.2 ifttg'ffi ddi ?ua uei- Skt. devavatara
For this group the ONWC forms can be retained with two modifications,
i.e., 1) we shall assume that no finals of the group ended in high front final
*-i and, and 2) we shall restore Jt tSe as BTD *kie. Compare also fl& rnd tlfr ,
both ONWC gr., which, with Schuessler (Ms.), \Me speculate did not undergo l
palatalization in the presence of final *-ie. Finally, as pointed out in section
3.5 above, we suppose that final *-6 was phonetically perhaps something
like [e"].
Also to be included here is i0 dii-, which rimes in this group and is
found by Yu Min in one transcription (1984:314): ,U dii- Skt. dhi
Sample Reconstructions:
II. FE *bre > bei: ffi *k. > kdi: ffi *mre ) mEi: ffi *tsre > t$i-
fi*dzrc > dz6i ffi *s"re > s€i-
*f. *krue > kuci- E *wre > ?uei-
III. 6 *ke, ge ) ke, ge ffi nr", je > se, ie- & *je > ie Sff *lie > lie
H *thie > thie E *b. > b. W *me ) me
cf. W *de (?) > dii-
ilE *gtr > gre A *khie > khie:,- f[ *tie > tie f *je > ie-
Itr *tshie > tshie:
FD *sie ) sie- ft *dze > dfue; H *tse ) t6e- 51 *ir" > ire 4*pie > pie
)t* *bie > bie- dH *mie ) mie:
5f *kie > tse fi *9" ) ge: q *9. ) ge- Ifr *gie > gie
H *we > ue ffi *khue ) khue ffi *gue > gue: ft *Uue > lue
-904-
BTD Revisited--A Reconsideration of the Han Buddhist TranscriPtional Dialect
ffi *wie > zue ( *thue ) thue
E *huie > huie ffi *kui. ) kuie
IV.# *k6, gd > kci-, fti- ffi *kc > kei fffl *5c > lei- F *te > tei-
fE *ac > dci EE *lc > lei-
* *kue > kuei fH *w" > luci
3.g The Jie tsf, Grou.p. This group contains the following oNWC finals:
(1) [-ai] -ei 1- -ai) (2) [-uai] -uei (- -oi)
(3) [-ni] -ei (4) [-wei] -uei
(5) [-iei] -ci (6) [-iwei] -uei
A qusheng example of final (1) occurs in the following case:
'I 224.438.1 = m som mei- Skt. samadhi
This transcription of samadhi also occurs in longer compounds in
the data.
Ting (1975:106-7; 212) found a single qu/ru rime contact involving final
(2) and a few cases of interriming with the Ji and Tai categories (sections
3.12 and 3.11). These point to the existence of a final coronal consonant in
qusheng syllables having this final. We shall represent the said consonant as
*-C here.
In the Han data, final (5) most commonly transcribes Indic syllables
having the vowel i. It is also represented in the following examples, where ai
and ay are rendered:
T 224.440.2 lB1s nei lei Skt. niraYa
T 196.149.3 ffiffiffi n6i lan dZan Skt' nairafljana
And in the TK data we find:
T 198.180.3 ffi*+w4 kei sa- kim pho li skt. [ajital kesakambala
In the following examples it is uncertain whether it is final (5) or final (1)
which is represented:
-905-
W. South Coblin
T 224.494.2 ftr$ iui dei- - ddi- Skt. virya
T 280.446.1 #+E put uo dei- - dei- Skt. purvavideha
But with Yu (1984:292) I suspect that ONWC dci- is the syllable in play
here. The transcription of virya is curious, and it seems possible that ifl there
is an error'for some other word, such as fift *tei-.
Finals (f) ana (2) can be reconstructed with the main vowel *a. Final
(5), which renders a range of different sounds, such as i, e, ai (= Pkt. e ?),
can perhaps be restored as *-6i. The entire group can then be reconstructed
as follows:
(1) *-ai, -aC (2) *-uei, -uaC
( 3) *-rai (4) *-ruai
(5) *-ci (6) *-uei
Sample Reconstructions:
I. Htai >?si$H*khai > khei: E*daC (?) > dei-ffi*meC) mei-
ffi *maC ) mei- fifl *phaC > phei-
ffi *weC ) tuei- ffi *guai > guei ff *dzuai ) dzuei: ffi *lrrai > luei
ffi *duaC > duei- ffi *tshuai > tshuei fr *nrraC ) nuei-
II. E*kri >kei=#*g:ai >'Iei m*tsrai ) tsdi ffi*wrai > ?uei
ffi *wrei > ?uei f/F *brai > bei
IV. A *dzei ) dzei ffi *kci > kei ;-f *kei > kei- #' *dci > dei: ffi *lei > lei:
E *sei > sei l{ *mdi ) mei i/fl *nei ) nci E *dci (?) > dei-
ffi *khuei > khuei H *wei > ru6i-
3.10 The Zhi flff Group. This group contains the following ONWC finals:
(1) [-jeiJ -i (2) [-jwei] -ui
(3) [-ji, -ji3] -i (4) [-jwi, -.1wi'1 -ri
(5) [-i1] -ii (6) [-wia] -uii
To begin, we should note that finals (1) (3) and (2) (4) form two
-906-
BTD Revisited--A Reconsideration of the Han Buddhist TranscriPtional Dialect
common entities, -i and -ui, in ONWC, whereas in the QYS they are
analyzed as four distinct finals. I do not find final (1) in the data at all.
Yu (1984:318) lists two occurrences of final (2), but I have not seen the
examples on which they are based:
H ui Skt. ve E ui Skt. ve
The following qusheng examples of final (2) occur in my data:
T 186.147.3 ffiEE dei ui- Skt. trapuga; cf. P tapussa, Khotanese ttriiviiysa-
T 362.317.2 lfrffiEEIJ pui- pii- do ia Skt. puqpadhvaja
lt seems probable that this final had final *-C in qusheng words'
QYS finals -ji (non-chongniu) and -ia are the preferred renderings of
foreign i in these materials, e.g.
Skt. sila
Skt. updsika
Skt. bhik-su; cf. Gd. bhikhu; P. bhikkhu
Skt. isana
Final (a) is used to transcribe foreign vi, or occasionally ve (< vai):
T 224.4?4.2 F si
'r 224.431.t IH.*H ?u bo i
T 73.241.1 LY,E bii- khu
T 224.43r.1 FiD ?ii $e
T 626.394.2 IEWW iui mo lo
T 626.404.2 ffits'FE iui su do
T 196.161.2 frffWW iui ia lie
T 196.163.2 ttr'ffi iui uei-
Skt. vimala
Skt. viSuddha
Skt. vais6li
Skt. vipasyin
-I 624.363.3 €EffiiE iui ma la net Skt. vimalanetra
The following qusheng example of this final perhaps attests to the existence
of finzrl *-C:
T 196.155.2 trlAffi ni guo lui- Skt. nyagrodha; P. nigrodha
Pulleyblank (1983:100, n. 12) has remarked that in my earlier treatment of
BTD (1981) I neglected this example and "passed over it in silence." In fact
however, my doubts about its interpretation \Mere mentioned in a footnote to
-907-
W. South Coblin
this example (1981:174, n.333), and it was these doubts which led to my
hesitation about including it in my discussion of the *-C problem. The fact
is that the same early transcriber (i.e. Kung Mengxiang) also renders
Nyagrodha as frww ni guo do (T 184.461.1; no. 292 in the data list). These
competing examples make it difficult to decide whether Indic -dh- in the
form Nyagrodha represents a stop, dh, or the Prakrit. >Pz suggested by
Pulleyblzrnk. A possibility is that both forms are variants or corruptions of an
original * tr+,offiFt. 11 is equzrlly possible, in my view, that T 196 represents
an original Prakritic text of the type envisioned by Pulleyblank, while T 184
is based on some other language, where intervocalic -dh- was a true stop' In
any case, I have no objection to the example as a possible representation of *-C.
QYS final -jis occurs in the following example:
T 224.425.3 effiffi gi dZa gut Skt. grdhpkuta; P. gijjhakuta
Since Skt. examples of this type are sometimes used to argue for a rhotacized
or r-like quality in QYS -ji3, it is worth noting that the Pkt. value for the
firsr syllable ffi here was probably gi. cf. Gandhari gihi (skt. g;hin-), kici
(skt. kltya), etc. compare also Yu (1984:311), who finds f; transcribing skt.
gl and Sih. The character fi occurs in a number of other transcriptional
forms, where it renders foreign ji or, in one case, ci. I suspect that there it
may be a scribal error for W dli-.
This group as a whole can be reconstructed as follows:
(1) - (g) *-i, (-iC ?) Q) - G) *-ui, -uC, -u(i)C (?)
(5) *-ii (6) *-uii
Sample Reconstructions:
III. ff *ki > ki ffi *khi(C) ? > khi- fr *ti > ?i EE *kui > kui ffi *w(u)i > ui
ft *pri > pui i# *puC > pui- FE *wuc > ui- X *muC ) mui-
H *si(C) ? > si- *$ *khi(C) ? > khi- ffi *ji(C) ? > i- 7,8 *pi > Pi
-908-
BTD Revisitedl--A Reconsideration of the Han Buddhist Transcriptional Dialect
^eil, *ti > ki ll3 *C, > C, E *C, > Ci E *t6i > tSi: zr *dZi > d2i-
(n.b. not 2i- in ONWC sources) * *si > Si: E *tsi > tsi- * *fri > fli-
[! *si > si- EF *sri ) oi iE *di > di fU {'li > li- H *ji > i
E *mi > mi W *tsi > tsi fi, *si ) si:
fiff *ui > iui Ettr *sru(i)C ) sui- {f *wiC ? > ui- S *tsu(i)C ) tsui-
g*1s(i)C > lui- ffi *Su(i)C > gui- ffi *wiic ? > zu(i)C ) zui-
& xkui ? > kui (This word rimes in the present group in w poetry,
in the you category in Han poetry, and in the zhi Z category in
pre-Han texts!)
ffi *mii > mii- ffi *khii- > khii- ih *pii, bii > pii:,bii-
X *kuii > kuii: 5 *kuii > kuii-
3.11 The Tai ffi Group. This group contains the following oNWC finals:
(2) [-uai] -ue,i - -uoC
(4) [-wai] -uEi
Most examples for this group involve finals (1) and (2), the ONWC forms
for which differed by sub-dialect. In the chang-an area they were apparently
open and "e-like." In the Gansu Corridor they had an "a-like" vowel and
ended in a consonant, which we transcribe as -C and whose basic phonet'ic
value may have been sibilant-like and palatal (Coblin 1991a). In compounds
this consonant may have undergone various assimilatory sandhi changes
depending on the phonetic nature of an immediately following syllable initial
consonant. A similar situation is also observable in the S dialect of sixth
century Jiankang (Coblin 1990). It is probable that the Corridor Dialect was
the more archaic here, and we consequently cite reconstructions for this
(1) [-ai] -ei - -oC
(3) [-ai] -di
dialect in the following examples.
T 784.472.2 F* poC to
T 196.147.3 1NWH po lo noC
Skt. pattra
Skt. vdrapasi
-909-
W. South Coblin
T 224.431.1 W@H-ttr ?o ?uoc suan su
T 280.446.1 [qlgtrH ?o ?uoC suan su
T 418.e13.3 *Hffi loC bii
T 458.437.2 ffivtfflffi. loC ta- 'vuo lo
Yu (1984:315) rfi bei-
Skr. dbhasvara + Su(bha)
Skt. abhasvar.a + Su(bha)
Skt. raSmi
Skt. r5gtrap-la
Skt. path(aka)
It is clear that final -C was present in BTD and that it could be used to
represent various foreign coronal consonants. We shall restore it as *-C. It
seems not unlikely that BTD *-C was subject to sandhi changes of the sort
found in the S dialect and the Gansu Corridor variety of ONWC and that its
behavior in the texts can be explained in similar ways. However, in order to
maintain this theory in the face of certain counter examples, it would be
necessary to ignore or "explain away" the offending cases. We could, for
example, begin by hypothesizing that in BTD *-C was phonetically [s], the
value we suspect it had in ONWC and the one we seem to obserye for it in
the rendering of raSmi above. For a form like pattra we can assume that *-C
assimilated to following syllabte-initial t-. But what is to be done with Yu's
transcription of pathaka? We might hypothesize that the form originally
contained more characters following [F and that they have since been
deleted. The first of these might., for example, have been a Chinese syllable
beginning with a coronal stop, which would have led t-s] to change by
assimilation to a similar stop. The BTD texts contain many transcriptions
such as {ffi "Buddha" and $ ffi "bodhisattva" which surely derive ultimately
from longer forms that were presumably shortened at a later date. Perhaps [F
is a case of this type. But in the end all this remains a speculation, because
we have no direct textual evidence for longer renderings of pafhaka in the
early texts. It would seem that for the moment we must leave matters as
they stand and continue writing *-C.
-910-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
The syllable ffi tshac - tsh6i- presents special problems in the BTD
materials. It has been discussed in sect'ion 2'3 above'
Sample Reconstructions:
I. # * goC > - roC - tei-t *aoc > doC - dei- [H *tthoc >
tshoC - tshei-l
€ *woC ) tuoC - tuei- ff *5uoC > SueC - nuei- ft *duoc >
duoC - duei- fi *PoC > PoC - Pei-
U. H *throC > thei- ffi *proC, broC > p6i-, b6i- [F *broc > bei-
'R *khruoc > khudi- Ffr *wroC > "IuEi-
?.L2TheJiSGroup.ThisgroupcomprisesthefollowingONWCfinals:
(1) [-di] -6i
(3) [-j.i', -juil -ei
(5) [-iix'1 -iei
(6) [-iei] -:i Q) [-iwei] -u:i
Finals (3), (4), (5), and (6) are
transcribe foreign syllables in a, e.g'
T 13.233.2 6ffii sa- uei-
T 152.12.1 f ffi Pun uei-
T 184.461.1 UftrffiGf ka iui lo uei-
T 184.461.2 ffitrn'fu dei ruo uei-
T 196.163.2 itr'ffi iui uei-
T 224.440.1 HE'df Su: do uei-
T 224.440.2 tr&ffi sei- ta la
T 418.903.1 lJlffirffi kii la uei-
T 418.903.1 IEIHE ?a d1a lei-
T 418.906.1 {B sei-
T 152.9.1 ffi€ru kei- nii ian
(2) [-wni] -udi
(4) [-3wliiu, -jwuiJ -uei
attested in the data. TheY usuallY
Skt. Srdvasti
Skt. pilCaPdta
Skt. kaPilavastu
Skt. dev6vatara
Skt. vipasYin
Skt. suddhavasa
Skt. sdstf
Skt. kaPilavastu
Skt. ajdtasatru
Skt. gatha
Skt. kpq4ajina
-911-
W. South Coblin
T 362.300.1 EM lei- uat Skt. revata
^t 224.435.1 'rF.H-+$,ffi suo tei- tSe nou- Skt. sudarsana; P. sudassi; cf.
Gd. daSayadi (Skt. darsanti)
T 224.435.t ',,8H suo tei- Skt. sudrsa; cf. P' sudassa
With one exception, these examples point to the existence of final *-C in
this group. The use of E to transcribe foreign re is atypical and is more
reminiscent of examples found in the ONWC data. I cannot explain it, unless
it is a late intrusion of some sort.
The first five finals of this group are reconstructed with *-ac. Finals (6)
and (7) are restored as *-6C and *-udC , which, as outlined in section 3'5
above, are assumed to have been phonetically [e'C] and [ue"C] respectively.
Sample Reconstructions:
il. /i' *k .C > kei- ffi *tsraC > tsEi- fi *sraC > $ei- f$ *praC > pei-
ffi *wraC > ?u6i-
m. lt gaC > Wi- +E *khaC > khei- {E *gaC > gei- 4 *UnaC > U"i-
E *s.C > sei ff[ *tsac > tsei- fr *tsaC ) tsei- ffi *laC (?) > lei-
ffk *biaC > biei-
ffi, ?uaC > ?uei- ffi *phaC > phei > phuei &t *b.C > bei > buei-
ffi*p.C > pei > puei- ff{ *kuaC > kuei- ffi *waC ) uei- ffi *suaC ) suei-
ffi *juaC > iuei- ffi *SuaC ) Suei-
lV.4 *kheQ > khci- H *tec > t6i-
ffi *weC ) 1u:i-
3.13 The Deng ff Group. There is a single final in this category:
[-engJ -ec
This final is attested several times in the BTD data:
T 607.230.3 M{Il sa3 ga Skt. samgha
T 224.433.1 ffi rs5 Skt. ganga
-972-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
The ONWC value can perhaps be retained here'
Sample Reconstructions:
I. ff *ta5 > tr! ffi *ba5 > brl tE *gr5 > ?aC EEI *ba3 > btU fB *sa5 ) san
3.74 The Zheng ffi Group. This group contains one final:
[-jang1 -ip
This final does not occur in our BTD data. Yu (1984:296) finds one
exarnple of it:
[* lil Skt. lavim(k)
I., w poetry it has large-scale rime involvement with the Deng grouP, and
Ting (1975:214) is in fact hesitant about separating the two categories' We
can tentatively restore it as *-ie5.
Sample Reconstructions:
IIl. frfr *kie5 > kil ffi *2aC > ziU ffi *jr! > il ]Ft *Uran > $U iH *dia! > di1
ffi nsaU > sin- {h *fia! > fril jzk piaS > pin
3.15 The Dong fi Group. This group contains the following oNWC finals:
(l) [-ung] -oc
(2) [-angJ -ag
(3) [-jwongJ -uop
These hnals are not attested in the transcriptional data. Final (2) can be
restored as *-rog to account for the \4J rime evidence'
Sample Reconstruct ions:
I. I *ko! > koU H *ton > to5 E *son > sou- * *moU > moc
II. 6 *groU>'IaU-ffi*kro0>kiiu: ff *sro5> sel*[ *p.lou> pan
ffi drop > dii5-
III. * *kuo5 > kuoS x] *huo5 > huoS ft*?uon > ?uou fi *duo5 > duoS-
ffi *juo5 > iuog ffi *luog > luoS ffi *tshuog > tShuoS fffi *zuog > zuos-
f{ *puol > puo3 S *buo5 > buo3:
-913-
W. South Coblin
3.16 The Dong $ Group. This group contains the following ONWC finals:
(1) [-uongJ -ouu (?)
(2) [-ang] -au
(3) [-jungJ -us - -iug
These finals do not occur in the transcriptional data. Final ( 1) is
attested in the ONWC materials in only one example, where the character H
transcribes the Indic syllable nahu-. The vocalism of this final is therefore
problematic. The riming patterns observed by Ting can be provisionally
accounted for by reconstructing final (2) as {'-Ju!.
Sample Reconstructions:
I. f *toug > touu fi *tsoug > tsouU ffi *noug > nouU
II. [4 *gru!, krun > ?AU, kAU-
m. + *dz:iug> dzugBd *k,r!> k.Uffi *g,rU> gu5 + *tu5> tu3
f;!*tsug > tsuS fi *flug > fluU E *luc > lu! H *PhuU > phu5
E *ku! > k.r5 f,ff *wu5 (l) > ruU )E *bu5 > bu5
F *mu5 > mo3
3.77 The Geng ffi Group. This group comprises the following ONWC finztls:
(1) [-ong, -engJ -dg e) [-wong, -wengJ -u63
(3) [-jungl -e3 G) l-jwongJ -uel
(5) [-jang1 -ies (6) [-jwang] -u(i)et
(7) f-iengl -eU (8) [-iweng] -ueu i
From this group, final (7) is attested in a number of variant. BTD tran-
scriptions of the name Bimbisara (- Bimbasdra), where the characters ffi '
ffi, and iff, ull beg, render foreign bim..
At the outset, one can perhaps reconstruct final (7) as *-e5' (*-i5 would
of course also be a possibility.) Finals (1) and (2) can then be restored as
*reg and *-ruen respectively.
-9t4-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
Dong (7978:247-249) argues convincingly that finals (3) and (5) were
not distinguished by the glossist Yan Shigu in the early seventh century.
Final (5) does not occur in the ONWC data at all. Final (3) appears in one
northwest transcriptional form of Dharmakgema:
T 192.2.2 * (,HEAP.) Effi ?jung - ?jung (read: ?[uo + m]jlwlung) gji'ta
Skt. angiras
Here { renders foreign an-, and we are told by the appended fanqie gloss
that QYS ?jong is the appropriate reading of the graph. To account for this,
QYS -jung was restored as *-eg [enJ for the ONWC stage, while for QYS
-jiing we tentatively retained *-ieg as our ONWC form (Coblin 1991a). We
have no evidence at all about the corresponding hekou finals, (a) and (6).
The majority of words having frnal (3) rimed in early Han and pre-Flan
poetry with the yang category of those periods and are thought to have
undergone a major shift to the geng group at a later time. The word { in
the T 192 example above is in fact a syllable of this type. Ting (1975:218)
has found that in WJ texts these words are unique in that they rime fairly
freely with both the yang and geng groups, and he identifies this as "a
phenomenon of the period of transition." We can perhaps signal this
peculiarity by reconstructing final (3) in such cases as *-e3, as against *-ieg
for final (5).
Ting (1975:219) has noted that in the works of WJ poets of the north-
west area the geng group finzrls as a whole have a slight riming affinity for
the yang category. He recognizes this as a dialectal feature. This coincides
with the fact that as late as early Tang times these finals, as manifested in
northwest dialects of that period, yr'ere rendered in the earliest layer of
Tibeto-Chinese transcriptions as Tibetan -ang. For the late Han and WJ
periods it may have been the case that these finals had high, i-like vowels in
-915-
W. South Coblin
Central Plains dialects of the type underlying the BTD transcriptions, and
lower, a- or e-like vowels in the northwest.
Sample Reconstructions:
II. ffi*kreg>kEU#*greu> 165: $*tsre5> t-se!ff*pre!> pen-
S. *sreg > -se5 ffi *dreg > d6! ffi *pr.n > pen
![ *k:eg > kEU fJ *greu > ?eU ffi *mreg I mEU-
W *wreg > ?u6u ft, *wrcn > 'IueU
fft *wre5 > ruEg
III. gE *ki"U > kieg: ffi *tshieS > tshieS '[L *sien > sieS- ]f *dzie0 > dzie5-
fi *tie! > tieg E *tse5 > tseS- fr *dZe! > dlen S *mie'tr > mieg
[,8 *me5 > meU T *b.U > b.n
ffi *kheg > kheg- il *uel > nen 4 *?ec I ?el fi,fi *geu > geu
F *k.U > keg:
E *pe3 > pe3 lfi *b.5 > be!- 45 *me3 > me!
ld *hue5 > hueS
{tE *khu(i)el > khu(i)el IE *gu(i)e3 > su(i)e!
ff *we! > ue5
It *we5 > ue!:
IV. ffi xk6u > k.u ffU *g.u > r.u B *tsh65 > tsheS E' *seU > seg
T *teu > teu F *b.U > be0 iF *meU ) mec E *dec > deu-
ffi *kuel > kueg ft. *wE! > 'IuaU
3.18 The Yang ffi Group. This group comprises the following ONWC finals:
(1) [-angJ -al e) [-wang1 -uau
(3) [-.iang] -(i)au (4) [-jwangJ -uag, -a{l
These finals are not attested in the transcriptional data. The ONWC forms
are here tentatively projected backward to BTD. In words like fH siaS and dfi
giag it is difficult to know whether the BTD form should be reconstructed
-916-
BTD Revisited--A Reconsideration of tht H4t Eg44t'l!!-TE=!!4
with *-iag or *-an. Either form would "work" in our system, but we lack
BTD evidence to make a cogent choice. A hint for the syllable $fi is perhaps
to be found in the following transcription of Dharmaraksa 9*W, a north-
west translator of the third century:
T 77.886.1 HqlJdfi€ lo ia giang gi
We can suspect that ONWC gtT rg here should
*gang in the northwest dialects.
Words like tr are reconstruct,ed for t,he early Shazhou period as diarl
(Tibetan spelling: j*g). For the oNwc period, in the absence of textual
evidence, we reconstruct ONWC *dang' However, from the Central Plains
materials of Tang times we have evidence which points in another direction'
In the YJ data we find:
T e85.466.2 [Tk 23] ,HF QYS ?uo djang Skt. odiyanika
This suggests that after the so-called sheshang initials, such as d-, this QYS
finalbeganwithahighfront'elementinYJ'slanguage'Averysimilar
example occurs in the XZ data, where Shi (1983:43) finds Chinese fi (QYS
djang:) transcribing skt. dyan. It is possible that BTD, as a probable direct
ancestor of XZ's Tang-time Luoyang dialect, actually realized fi as *dia5' But
earlier *da0 yielding a later *diaU in the Luoyang area is also a possibility'
The matter remains uncertain.
Sample Reconstructions:
I. ffi[ *koU > kou H *kho5 > kho5 Tt *ga! > 'IoU E *dou > do3
ffi *dzon > dzoS * *soU > soU
)f, *kua5 > kuaS E *wan >'vua0 * *ba! > ba! ffi *huag > huan
m. * *t(i)al > kia! # *jaU > i.n tf *z(i)a5 > zian 5ft *g(i)a0 > sra0
E *d(i)at > d.n F. *l(i)au > liaS fE *s(i)as siaS ffi *flau > fla!-
ffi*Sal > Slffi *srag ) ga3 H *dflan> dia! ff *tsrag > t$aU
Skt. lomasakangiYa
be projected back to earlier
-917-
W. South Coblin
H *tsau > tsa! ffi *ts(i)a5 > tsiag-
E *ba! I ba! > buag t nma! 2 man > muau
3.19 The Zhen ffi Group. This group combines finals belonging to two large
rime categories of the pre-Han period. Ting treats it as a single group in the
Wei period but divides it into three separate classes for the Jin period:
A. Hun ffi Category
(1) [-en] -an (2) [-uan] -on
B. Wen { Category
(3) [-janJ -in (4) [-juanl -un
C. Zhen ffi Category
(5) [ (-jen),-jen') -in (6) [-juen,l -uin
(7) [-jien,] -iin (8) [-juient] -uiin
Ting (1975:279-22O) does not reconstruct different Wei and Jin values for
these finals and in fact does not claim that the temporal differences in rime
classes here represent phonological changes. It seems possible that varying
poetic rime conventions underlie these differences. The statistical table Ting
gives (1975:27?) leads me to doubt the validity of the distinction between
categories B and C. [t seems clear, however, that A was distinct from B and
C in some way.
Final (1) is absent from the data. Final (2) is represented by two
characters:
T 13.236.3 i4F5 *qa mon Skt. Srama4a
T 418.905. | ,Hf1*suo mon Skt. sumana
T 458.438.1 4ffiF5 *Uo lo mon Skt. brahmapa
T 152.44.3 IH# *?u pon Skt. utpalak
These examples suggest that fl and ff had unrounded vowels in the BTD
-918-
BTD Revisited--A Reconsideration of the Han Buddhist Transcri
period, and we can restore them as *men and *pen respectively' We can
assume that. in such examples earlier *e was later labialized by the preceding
initials. Examples for non-labial initial syllables are absent from the data' In
such cases one could perhaps posit *ue, as is done in a number of the
current QYS reconstructions. [n non-guttural initial syllables, plain *a could
be reconstructed.
Finat(3)isabsentfromt'hedata.Final(4)frequentlyrenderslndic
syllables having the vowel u, e'g'
T 224.454.1 ,)iqWfe# pun mon- do ni put
T 313.754.1 frWtr[ Pun do li-
T 224.431.1 ffid[X Sek ka mun
T 752.?4.3 *6n kuo mun
T 784.462.2 WE lo un
T 474.523.2 WT lo un
'f 280.445.3 fi+ff( gun no-
Skt. par4amaitrdYa4iPutra
Skt. puq{a"rika
Skt. Sakyamuni
Skt. kumuda
Skt. rdhula
Skt. rdhula
Skt. gu4a
Skt. pipdaPdta
seems unclear' For the nonce
(7) usually transcribe foreign
an a-syllable in one case'
However, there are also cases where Indic syllables in a are transcribed' e'g'
-l 224.425.3 XffiHmfU mun dZuo qi li- Skt' mafrjusri
T 224.426.2 W## sot un fla: Skt' sarvajha
In the first of these examples we may suspect that the underlying Prakrit
formhadmufr-;cf.Gd.mumjavamda=Skt'mafrjuv-da-'Gd'mumjukrita=
Skt. mafrjukirti- (Fussman 1989:466-67)'
And one case rendering Indic i is attested:
T 752.72.1 ,j'ffi Pun uei-
How such differing forms should be reconciled
it seems best to retain the ONWC value here'
As Yu (1984:295) has noted, finals (5) and
syllables in i. However, final (7) transcribes
-919-
W. South Cnblin
Compare:
T 196.163.2 WOH biin dou
T 362.300.? H.ffi. piin 3ie:
Skt. bandhumd
skr. -bhijna
I can offer no explanation for this as it stands. Could it be that the Prakrit
form underlying bandhumd is *bundhumd and the character ffi here is an
error for ffi bun? Somewhat similarly, the graph ffi lin is used in these
materials not only for syllables such as -lin- and -ndin- but also for -ran-.
The same behavior is observable for ffi len. I believe this is due to graphic
confusion, with ffi being the original or "correct" form for -in type transcrip-
tions and tfif for the -an type.
The graph fr[ pin is used to represent both -pin- and purq-. Another
graph which is used for pur4- is ffi bun. I believe that these [wo have been
scribally confused and that Jil should be used only for unrounded syllables
like -pirr-.
Final (6) appears in the syllables ffi luin and =# luin, lon-, both tran-
scribing foreign -rul-. ONWC contains no syllable lun to contrast with luin.
It is possible that luin derives from earlier *lun here. But it could also be
that an earlier *luin was the best possible choice and was used for this
reason. This is the solution we shall provisionally adopt.
In conclusion, I assurle that sub-category types B and C formed a
common rime group in which the high Yowel syllables *-in and *-un could
interrime. Typ" A would have constituted a final *-an class which
occasionally rimed with B and C.
Sample Reconstructions:
I. ffi *ken ) kan ,S, *?en ) ?an f *then > than (?)
ffi *wan ) ?on tr *ben ) bon # pan > pon 6 *prn > pon:
fl *men ) mon # *dz(u)an ) dzon ff *huan ) hon fE *s(u)an > son:
-920-
Revisited--A Reconsideratio, of t!" l3*-E*94t'l!tf
=6 *1(u)an ) lon- g *6(u)en > don
III. +fi *tsin > tsin- ffi *air^, dlin ) hin, dain 7J' *flin ) flin:
# *lin > lin-r[ *kin ) kin € *bin > bin H: *gin > gin
)[ *gin ) gin:,- ,fr *kin > kin ffi *trin > cin =F
*'srin ) sin
Bl *jin > in: E *tsin > tsin tF *Lir > lin A *nin > flin H *sin > sin
H *dzin > dzin: tF *dzin >dZin- E *tsin ) tsin- {fr *sin > sin-
EF *din > din- b *tin > tin
X *mun > mun E *kun ) kun S *wun > un f+ *gun ) gun
fr *guin > guin: fif; *win ) uin: $ *phun > phun ffi *mun > mun-
6 *tShuin > tShuin ff; *dZuin > diuin IIE *Zuin > Zuin-
! *uin or *juin (?) > iuin ffi *win > uin @ *wiin ) zuin
i6l *suin ) suin
* *kiin > kiin: E *riin > *?iin fi *piin > piin E *uiin > biin
.R *miin > miin
€ *kuiin > kuiin
S.2oTheHansGroup.ThisgroupcomprisesthefollowingoNWCfinals:
(1) [-an] -an (2) [-uin] -uon
(3) [-an] -an (4) [-uan] -uiin
Finals (1) and (2) are fairly common in the data' e'g'
Skt. ananda
Skt. nandanavana
I. f *kon > kon ft*?an ) ?on$ff *non > non @ *tson > tson-
iffi *mon > mon:
ffi *kuon ) kuon ffi *won > 'Iuon: f{ *duon ) duon-
T 14.24t.3 [El*ff ?o non
T 224.477.3 EtrtEE non don 'Yuon
Finals (3) and (4) are not attested'
Sample Reconstructions:
-92t-
W. South Coblin
II. # *kron ) k5n ffi *5ron > nrin- ffi *sron > *injS *mron ) miin-
ffi *pron > piin:
ffi *k-ruon > kuln nfi *lron > Suiin ffi *dzruon: ) dzuiin:
3.21 The Yuan 7U Group. This group comprises the following ONWC finals:
(1) [-an] -an e) [-win] -uiin
(3) [-jr.t, -jtin3J -an (4) [-jwun, -jwiin3l -uan
(5) [-jian'l -ian (6) [-jwiiinnJ -uian (?)
(7) [-ien] -en (8) [-iwen] -uen
Of these, (1), (3), (4), (5), and (7) are attested, as exemplified in the
following:
(1)
'I 224.434.2 tr18 tshiin:,- dei Skt. ksanti
(3)
T 224.438.3 WWW gan- do lo Skt. gandhdrva
T 224.435.1 ffiWW gan do lo Skt- gandhdrva
T 607.232.3 {FE tsan don Skt. candana
T 32.814.2 E ffis muk gan lian skt. maudgaly-ayana; P. moggallana
T 280.445.2 EZBtr kit lian "ruon Skt. *hiranyavarqa
T 185.476.2-3 ESffi ni lian dZa Skt. nairafijand; P. nerafljand
- nirafljand
(4)
T 224.475.2 fiHItr ku uan Skt. kumbhrpda
T 302.300.3 fEHrtrffi lou i suan lo Skt. lokeSvara
(5)
T 224.427.3 @;rjfii mo ho ian: Skt. mahaydna
-f 474.527.1 Hro iu ian Skt. Yojana
-s22-
T
T
752.33.1 WBfr mo ?iin sen
1e8.180.3 ftVftfiHEBffr s6n
T ?9O.729.L {fo$ Pie s6n nik
T 1e6.157.2 IEfiH ?u den
T 418.902.3 i[gf# ka len
T 418.903.3 Wl#tr da l6n ni
Yu (1984:314): IE den
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
(7)
T 224.465.1 tr t€jfl ni iui sen Skt. naivasar.n [j franasamj ff ayatanopaga] ;
P. nevasafr lRanasaffayatanupaga]
Skt. mahendrasena
gue ku huie lo ti
Skt. safljayi-vairattl - sdfljaya-vairati
This example is corrupt in some way, but the first portion seems clear'
T 474.522.2 fty],R++ sen bii lo di Skt' samjayin vairati
Skt. prasenajit
Skt. udayana
Skt. kalandaka
Skt. dhdrar.li
Skt. dyan, P. den
Finals (3) - (5) can be reconstructed with their oNWC values- Final (6) is
very rare, and there is no information on it in either the BTD or ONWC
data. Final (7) transcribes both e and a vowel syllables. As argued in section
3.5 above, I tentatively reconstruct this final as *-en [ean]'
The character ffi , which rimes in this group and appears in the TK
data, has been discussed in section 2'7 above'
Sample Reconstructions:
II. ffi *kran>kdnffi *gran>'viin [I] *sran> Xinffi,*dzren) dzdn:
ffi *bran > biin- ffi *kran > kiin fifl *gran ) "I5n: ffi *phran > phiian-
fl *wran ) ruiin-
m. e *kan ) kan- E *Uan ) rJanffi *gan ) gan lii *jan > ian:
E *khi.n > khian:,- ffi *?an ) ?an, *-an > -an (enclitic only)
:s *t(i).n > lian lE *tsh(i)an > tshian: # *j-, z(i)an ) ian-, zian-
ffi tan ) tan: ffi *tsan > tSan- ffi *dZtn > dlan ft' *man > man:
-923-
W. South Coblin
hf *ban ) ban: fF *ti.n > bian- ffi *mian ) mian-
E *puan > puum: ffi *muan > rnuan- ft Cuan > Suan ffi *wan ) uan
S *wan > uan: ffi *guan > guan:,- H *wan ) uan- fE *gr- > guan
ff *kuan ) kuan- ffi *guan > guan- ffi *kuian > kuian- (?)
*$ *tuan ) tuan:,- } *dzuan ) dzuan ffi *wian > zu,an
H *suan ) suan $ *tSuan ) tSuan
IV. _E *k6n > kdn-, *gen > r6n- ffi *dzen ) dzen E *p"n > pen
ff*phen > phen- ft *sen ) sen fr *tCn ) ten: ffi *den > ddn
H *g6n ) t6n E *den > ddn- ffi *den > den fi *nen > nen
fffi *l6n> len ffi*w6n>,yuen- @ *wen ) 1u6n;,- f,*wen> ?uen
ffi *?ueen ) ?uen
3.22 The Qin f{ Group. This group comprises the following finals:
(1) [-am] -om
(2) [-jeml -im
(3) [-jungl -u5
In both the Han and TK period data these finals transcribe Indic
syllables having the vowels a, d, or r , e.B.
T 13.233.2 [qJft?o rom Skt. dgama
T 418.903.1 ,EiR suo Sim Skt. susrma
T 474.522.2 HqF+ (EfiiE >) iEffi kie ia kim pa la Skt. kesakarnbala
T 280.445.3 ffiW pu5 mo Skt. brahml
From the TK data there 'dre also examples transcribing foreign rounded
vowel syllables. For example, from Yu Min's data we have (1984:298):
$ kim Skt. kum -- som Skt. sum
And in our own TK material we find:
T 362.300.3 *+6J]$. tshoi- kuo dzim Skt. -samkusum[itabhyudgata]
Examples of this type stern from the Wu area, and it seems possible that
-924-
BTDRevisited--AReconsiderationoftheHarrBuddhistTranscri
they represen[ a southern dialect which had a rounded vowel here.
A possibility in dealing with this group is to adopt a "traditionaf" recon-
struction in which final (1) is restored as *-em and (2) as *-iam. But the
latter form is not satisfactory, either for our Han or TK examples of, final
(2). Ttris leads us back to Ting's rime data (1975:168-172) for a review of
his delineation of this rime category. At the outset we find that examples of
final (1) are rather scant in the corpus as a whole while final (2) is very
common. Cases where finals (1) and (2) interchange in Ting's materials
occur in the poetic works of LuJi ffiffi, Lu Yun EE, ZherrLg Feng AIH'
and Zhang H.n ffiffi . Now the Lu',s and Zhang Han were all natives of the
Wu area. Zheng Feng was by ancestry a native of Peijun iffi E[ but was the
son of a prominent official of the wu state and probably grew up in that
area. For Guo Pu +]1 Ix , a native of the central Plains area, Ting lists the
following rime sequence: iH/H (fr*,*H.)' Ting considers 4 here to
belong to rime (2), but I believe that in this poem it should be read as QYS
tshQm, sense of "to intertwine, intermingle (s.. branches of a mulberry
[ree)." What emerges here is the realization that in this period interriming
between finals (1) and (2) was a feature of the'Wu area, which, as Ting
(1975:262-264) has shown, probably had special dialect features of its own'
Though these two finals interrime in poetry of Han and earlier times, there
is no real eviclence that this was so in WJ poetry of north China' In the
north, the two finals may also have already been distinguished in dialects
such as B',ID. Our solution for the present will be to adopt the ONWC value
for final (2) and to restore final (1) as *-am'
In Ting's data final (3) belongs firmly in the Dong $ group (see section
3.16). There is no question about its \4{f period assignment there. But in our
material the transcription of brzrhmd in the T 280 example suggests that it
-925-
W. South Coblin
belongs in the present category in BTD. [t can be reconstructed as *-uam
(< *-am), as suggested elsewhere (Coblin 1991c). In connection with this
final we also tentatively include here the word fift "bear." Modern northwest
dialects (which read 9y5 or gyu5 for this word) point to an earlier form *'Iug
at some pre-modern stage, but from material representing the Gansu
Corridor dialect of ONWC we have the following example:
T 664.387.1 iEffiFHB[EI{n po lo [QYS .lungJ mo ?a no Skt. trrzrhmanu
Herefifirepresents foreign -hm- and may reflect ONWC'Iuam (> later ru5 ?).
Perhaps this is derivable from earlier *wem.
Sample Reconstructions:
I. ffi *kam ) kom: ffi *nam ) nom = *sem > som
m. + *kim > kimE x?im> ?i-ff *lim > lim ffi *phim > phim:
|fr *Sim > Sim ,i. *sim ) sim ff *srim > Sim
E * (plem ? > ) puem > pu5(9)
ftft *wem (?) > *'yuam (?) > 'IuU
3.23 The Tan =ff Group. This group comprises the following ONWC finals:
(1) [-am] -om
(2) [am, -dm] -iim
(3) [-jom, -jtimr] -am (4) [-jiiima] -iam (5) [-jwom] -uam
(6) [-iem] -em
Finals (1), (3), (4), and (5) are attested in the data, all tremscribing foreign
syllables in a, e.g.
T 280.445.3 &W dom mo Skt. dharma
T 224.471.3 5H tsam bak Skt. campaka
T 224.439.3 lQ iam Skt. yama
T 22,1.432.1 ffii?flJ iam bu li- Skt. jambudvipa
T 13.236.2 ff buam- Skt. brzrhma
-926-
BTD Revisited--A Reconsideration of the Han Buddhist Transcri onal Dialect
Sample Reconstructions:
I. ff *kom ) komffi *lom ) Iom=ff *dom ) domff *dzom > dzom-
II. ffi *kram > kiim ffi*dztam> dzilm ffi *g:am > tiim ffi *kram > kiim:
iS *dram > diim:
III. m *ctm ) Ym X. *kham > kham- ffi *pam (?) > pam: (The develop-
mentofthiswordisirregular,sincewewouldexpectBTD*pamtoyield
later *puam.) fft *kam ) kam: ffi *nam > Cam- E *?am ) ?am lffi *?iam >
?iam- ffi *ju- > iam ft *ju- > iam ffi *s(i)am ) siam
p *pham > phuam- /il, *bam ) buam H (*bltm ? >) *bam ) buam-
IV. ffi *kem ) kem iE *dem ) dem ft xnsrn ) ndm-
3.24TheDeffiGroup'ThisgroupcontainsthefollowingONWCfinals:
(1) [-ek] -ak (2) [-wak] -uak
(3) [-ek] -ek (4) [-uak] -udk
Final (1) is attested in the BTD data, e'g'
T 196.150.3 [-r]BS, ?o mo lak Skt' amalaka
Yu (1984:294) gives an example for final (2): iii ruak Skt' vak
Sample Reconstructions:
I. fi *khak > khak ffi *tak > tak ffi xrn3[ ) mek
@l *kuak > kuak fi *wak ) 'Iuak
II. s *k.ragk > k6k fr *mrek > mdk
fiil *k:uek > kuEk
3.25TheZhiffiGroup'ThisgroupcomprisesthefollowingoNWCfinals:
(1) [-jak] -ik (2) [-jwek1 -uik
Final (1) is attested in examples such as the following:
T 280.446.1 iflqlJffi kuo ia nik
T 224.429.1 fEH kuo ik
T 224.434.1 illH ka ik
Skt. godaniya
Skt. kauSika
Skt. kayika
-927-
W. South Coblin
T 196.150.3 ffiiE iam pik Skt. campaka
In this example, l$ is possibly an error for H bak, which occurs in other
forms of the transcription of campaka.
A possitrle exarnple for final (2) is given by Yu (1984:317-318):
b} uik Skt. vik
We can perhaps retain the ONWC values for these finals.
Sample Reconstructions:
III. ffi *glk > grk H *dik > dik € *jik > ik {4 *65i1 > tsik ;Eh *sik > sik
@ *srik > sik ffi *pik > pik
1el *huik > huik bI *wit > uik
3.26 The Wo )f, Group. This group comprises the following finals:
(1) [-uok] -ouk (or -ouk ?)
(2) [-ak] -ak
(3) [-jukJ -(i)uk
(4) [-iet<] -et
Final (3) appeaxs in our BTD data:
T ?2.814.2 EffiS muk gan lian Skt. maudgalydyana; P. rnoggalldna
T 607.230.3 Xb then tuk Old lranizrn hinduka
T 224.458.1 pqln ?o tqhuk Skt. aksobhya
And Yu (1984:310, 376, 377) gives the following examples:
El muk Skt. muk, muc 6[ ?(i)uk Skt. ug, yug
lt seems probable that the main vowel of final (3) was an u-like sound,
though it may have had o-like qualities as well. For the nonce it seems best
to retain the ONWC value here. For the remaining finals we have only
Ting's rimes to guide us. Final (1) can then be retained as *-ouk, a value
which is in itself problematic, and final (2) can be derived from earlier
*-ruk. Final (4) is rare in Ting's materials, but where it occurs it has a
-928-
BTD Revisitedl--A Reconsideration of the Han Buddhist TranscriPtional Dialect
noriceable affinity for final (1); see Ting (1975:177-78). Perhaps it can be
restored as *-6uk.
Sample Reconstructions:
I. E *kouk > kouk E *douk > douk
II. E *bruk > bek $ *gruk >'Iak
ilI. ffii *prrk > puk El *muk ) muk ff *tut > tuk ffi *suk ) suk
fifi *sriuk ) ouk ffi xt6uk > tsuk fi *luk > luk fn *Sut > Suk
ffi *krrk > kuk ffi *huk, thuk > huk, thuk H *i"t > iuk
?E *prk > puk [P *buk > buk EI *wuk > uk
IV. E *dduk ? > dck m *tsheuk ? > tshck
g.27 The Wu ffi Group. This group contains the following finals:
(1) [-uk] -ok
(2) [-ak] -ak
(3) l-jwokl -uok
Final (l) is attested in three examples:
T 626.399.1 +qtr kuo sok
T 626.401.3 +qffiB kuo sok mo
T 362.300.3 fBFtX[fl{ t.an do sok iuo
These transcriptions point to an u-like vowel for final (1), and we are hard
put to differentiate this vowel from the *u we have posited for the Wo
group. But it seems worth noting that our examples themselves are not
without their peculiarities, for it is difficult to account for the use of final
-k in rendering the open Indic syllables here. Now, in the cases from T 626
it seems very probable that the form at T 626.399.1 is a shorter variant of
that found at 401.3, and it is thus possible to suppose that s in the former
has been simplified from ffi in the lat.ter. And from here we may hazard a
guess that ff in all these cases is really an error for ffi sou-, which lacks the
Skt. kusuma
Skt. kusuma
Skt. candra-surya-
-929-
W. South Coblin
problematic -k and would be a regular rendering for foreign su. If this were
true, it would leave us with no transcriptional examples at. all for this group.
In any case, for the present I propose to retain *-ok as the BTD final for
the entire category.
Sample Reconstructions:
I. ;6 *kok > kok E *?ok > ?okffi *dok > dokffi *lok > lok
ffi. *dzok > dzok ff *sok ) sok l. *pok > pok f *mok ) mok
{* *bok > bok
II. fr *kok > kakE *nok > Uekfi{ *prok> piikffi *trok > tiik
iffi *drok > dek tE *tsrok > ts{ik ff *srok > flak
III. ffi *khuok > khuok ffi *juok > iuok E *Uuok > Suok ffi *duok > duok
ffi *luok > luok [ *tsuok ) tsuok ffi *zuok ) zuok {6 *zuok ) zuok
$ *fluok ) fruok H Suok > Suok ffi *dZuok > dZuok
?.28 The Yao ffi Group. This group comprises the following ONWC Finals:
(1) [-ak] -ak (2) [-wak] -uak
(3) [-ok] -6k (4) [-wuk] -udk (5) [-at] -at
(6) t-jakl -ak (7) [-jwak] -uak (8) [-j:it, -;ut] -iet<
(9) [-iek] -ek (10) [-iwek] -u6k
Finals of this group transcribe Indic syllables having the vowels a or a, e.g.
T 280.445.3 'l'EffiHmfIJ flak no- qi li- Skt. jfranasri
'f 224.429.1 ffi sek Skt. Sa&ra
T 224.431.1 ffiillX Sek ka mun Skt. Sakyamuni
T 206.510.2 ffiffi sot bok Skt. sarthavaha
Sample Reconstructions:
I. ffi *?ok > ?ok6 *kok > kokE *lok > lok# *nok ) nokP *dok > dok
'fF *tsok > tsok ffi *bok > bok H *sok ) sok
ffi 'rok > rok * *tot > lok S *dzok > dzok
-930-
BTD Revisited--A Reconsideration o! !!9-flg-Eggqhi!!
FIJ *kuok > kuok ffi *wok > 'Yuok
II. B *kr,ruk > kh6k t6 *k,^k > k6k E. *brak > bek 4 *dJa-k > dek
H *srak > sek
E *wrak > luek
* *!rok > lak ffi *krok > kiik ff *prok > pak $ *trok > tak
III. fl *khak > khak # *Rat > frak € *tuk > tak E& *lak > lak ffi gak > gak
E *nak > n k ffi *g.k > gak # *juk > iak fF *tshak > t'Shak
! *dzak > d?ak f,,! *a* > ?ak ffi *juk > iak M *flak > flak
EU *g^k > gr"k frffi *t iat< > hiek iE *5rak > 5rtk H *ti^k > siek
9 *ziak > ziek ffi *dziak > dziek /i *ji"k > iek E *d7iak > d2ek
ffi *si.t > sek E. *tshiak > tshek # *pi"k > piek
ftS *U"t > buak B *kutk > kuak
Iv. E! *tek > ftk fA *nek ) nek iff *kck > kek
S.2gTneXiffiGroup.ThisgroupcontainsthefollowingoNWCfinals:(1) [-ek] -6k (2) [-wek] -6k
(3) [-.iak] -iet (4) [-jwzikl -uek
(5) [-iek] -ek (6) [-iwek] -uek
Final (3) occurs in a single example:
T 602.170.2 ffi{ Piek tSe Skt. pratyeka; P. pacceka; Gd' prace'a;
Maharaqtri; padiEkka; Ardhamdgadhi patteia
This example is difficult to interpret, but for what it is worth, it seems to
point to an "a-like,, vowel in final (3). This brings to mind the interesting
fact that this rime category, though taken as distinct by Ting, must have
been very similar to the Yao group. Indeed it has 32 rime interchanges with
that. group, as opposed to 40 internal or "regular" contacts (Ting 1975:224'
226).ThiscanbecomparedwiththeparallelYangandGenggroups.They
have, respectively, 711 and 774 inLernal contacts, as oPposed to 23 mutual or
-931-
W. South Coblin
intercategory contacts. The situation in these "parallel" nasal final categories
was therefore essentially different from that. we face in our checked final
groups.
By the ONWC period the Xi group finals had all merged with
corresponding Yao category finals. In the light of the rime data and the
admittedly scant transcriptional evidence, it would seem reasonable to
reconstruct finals in *-ak (and *-ek) for the Xi group and assume that they
formed, in effect, a single final category with the Yao group.
Sample Reconstructions:
II. ffi *t<-rat> kekfr *tsrak > tsEkm *fiak > tekg *pr.k > p6k
flffi *mrak > mdk
E *wrak > rudk
m. A *?iak > ?iek iffi *siak > sek # *tsiak > tsiek 5[ *jiak > iek
ffi *piuk > piek 1g *'phiak > phiek
fft *juat< or *uak (?) > iuek
IV. ry *kek > kck ffifl *Uek > 5ek ry *ph€k > ph6k ffi *lek > iek
ZII *ttrct > thek fD *rck > sek ffi *ts6k > tsek
ffi *kuek > kuek
3.30 The Zhi H Group. This group is the checked final analogue of the
Zben category. Ting (1975:226-227) recognizes two Jin period sub-groups
for it:
A. The Mo i9 Category
(1) [-et] -at (2) [-uat] -ot
B. T}:.e Zhi H Category
(3) [-jett, -jat1 -it (4) [-jueF] -uit
(5) [-just] -ut (6) [-jietr] -iit
(7) [-jiueta] -uiit
-932-
BTD Revisited--A Reconsid".uti"" "f tht H ional Dialect
Examples of attested finals:
(2)
T 362.300.3 ffittJYEfrt si li- kot kue: Skt' sri-kufa
The last character in this compound seems to be an error for something else'
-r 74?2.1043.1 X.EW dot kiit lo Skt' duskrta
(3)
T 280.445.2 -#SE kit lian ruon Skt'*hirapyavar{ra
T T 224.435.1 i&* &{6W po lit to suo ho Skt' parittaSubha
(4)
T 602.173.1 ,f,iffi iluit dza
T 624.363.3 f[ffiiEftJ ( l-f ' ) ffifi rwo do
T 313.758.3 ffirF,i no Zuit
T 458.435.2 FqW'e ?o non luit
T 196.153.3 ?\'EW kuo luit do
(5)
T 73.2?3.1 {# but
T 280.445.3 g (if, >) W ?ut tom
Skt. vidYa; dd. vija, P' vijja
po li Zuit dei-
Skt. vrataParisuddha
Skt. naYuta
Skt. aniruddha
Skt. kolita
Skt. Buddha
Skt.. uttama
(6)
T 224.425.3 WEIftW& po, fla: po lo miit Skt' prajfiapiramita
The forms to be reconstructed for this group parallel those posited for the
Zhen category.
Sample Reconstructions:
I. W *gaL > ?at i9 *mat ) mot
ff *kuat > kot ff. *huet > hot 4 *'15(u)et ) tsot
ilt. z *khir > kJ].it z *?it > ?it
ffi git > git 6 *mit > mit E *'jit > it ;p *11, > lit -b *tshiL > tshit
-933-
W. South Coblin
lF*dzit > dzit H *sit > sit B *f,it > nit ffi *dit > dit H *Lit > zit
ffi tsrit > tgit LT *srit > $it € *pit > pit
# *put > put ffu *lrlrut ) mut ffi *khut > khut fr *tshuit > tshuit
ff (Early BTD *uit ? >) *zuit > iuit ffi xluit > luit pfi *sruit > suit
ffi *suit ) suit $ xjuit > iuit
H *tiit > kiit i* *tt iit > khiir ,1, *piit > piir E *phiit > phiir
& *miit > miit * *?iit > ?iit
fffi *kuiit > kuiit
3.31 The He E Group. This group contains the following finals:
(1) [-at] -ot (2) [-uit] -uot
(3) [-at] -at (4) [-wat] -uiit
Finals (1), (2), and (3) are attested, e.g.
T 224.433.1 FEfn sot 'Iuo Skt. sattva
T 280.445.3 i6E 'vuot it Skt. vajra; P. and BHS vajira
T 418.913.3 +UfU tshiit li- Skt. kqatriya
This group is the checked final analogue of the Han category and can be
reconstructed in parallel with it.
Sample Reconstructions:
I. ffi*kot >kotiS *dot > dotE *got > lotX *mot )motEfi *bot > bot
iffi *wot > 'vuot S *duot > duot ffi *luot > luot
III. fg *grot ) 1At W *hrot > hdt
6( *kruot > kuiit fi( *gruot > guiit
3.32 The Yue fi Group. This group comprises the following ONWC finals:
(1) [-at] -at (2) [-wit] -uiit
(3) [-jot, -jzitt] -at (4) [-jwot, -jw;it3] -uat
(5) [-jiat41 -iat
(6) [-iet] -et (7) [-iwet] -uet
-934-
BTD Revisited--A Reconsideration of the Han Buddhist Transcriptional Dialect
Attested finals are illustrated in the following exarnples:
(3)
T 224.471.1 tr^&ffi dom muo gat Skt' dharmodgata
(4)
T 224.471.3 ffiffiru gan do uat Skt' gandhavati
T 196.148.1 1frEE{g iuat dou don Skt' Suddhodana
T 224.429.3 ffiXiuat t-sha Skt' yakqa
(6)
.f224.427.2-3fgffiMiE'sennosenn6tSkt.sannahasannaddha
T 280.445.3 iEffiHffifU net lo si li- Skt' netrasri
Yu (1984:283): iE# n6t bon Skt' nirvdpa
T 752.21.2 ffifr set mot Skt' sadamattam
This group is the checked final analogue of the Yuan group and can be
reconstructed in parallel with it.
Sample Reconstruct ions:
II. ffi *srat > $at * *tshrat > tshiit y'f *krat > kat #i *brat > biit
/\ *p-rat > pat HL *?:at > ?At ffi *grat > 'Yiit
ffi *wrat > ruiit
III. +E *kat ) kaL WR *hat ) hat El.J *bat, pat ) bat, pat {X *gat ) gat
'v& *gat > gat im *miat > miat ffi *biat > biat fr *iat > i'at
fJl *l(i)at > liat ffi *t(i)at > tat ffi *flat > flat
{Ji *buat > buat ff *puat > puat fi *0uat > nuat ffi *wat ) uat
ffi *guat ) guat ffi *Suat ) Suat ffi *juat > iuat '!) Luat ) luat
IV. ffi *dzet ) dzEL fl *khet ) khet E *met ) met' ffi *tset ) tset
f* *ket > ket
ft* *krret ) kuct ffi. *huet > huet f, *wet ) ?uet
-935-
W. South Coblin
3.33 The a tr+ Group. According to Ting this group contains the following
finals:
(1) [-+p] -op
(2) [-japJ -iP
But in his data (1975:192-193) I find no interriming at all between these
two finerls. From the standpoint of WJ riming practice they can be considered
separate.
The following examples occur in the BTD data:
T 313.753.3 ffi$ru gan dap rua Skt. gandharva
T 196.156.1 iffi ( & >) YE Sip pa Skt. Slvaka
The group can be tentatively reconstructed in parallel with the Qin group, .
its nasal frnal analogue. Yu (1984:312) includes the following example for
final (1):
A *rop Skt. gup
This example stems from the Wu area and may represent a southern dialect'
Sample Reconstructions:
I. ft *gap >'Iop ffr *nap > nop S *trp l toP ffi*dzap > dzop
III. & *grp > gip tr *?ip > kp R *dzip > dzip y'. *flip > flip
f *dzip > d2ip B *zip > zip ]7 't'lip > lip #t *tsiP > tsip ifr *sip > sip
3.34 The Ye ffi Group. This group comprises the following ONWC finzrls:
(1) [-ap] -op
(2) [-ap, -dpl -iip
(3) [-jop, -iaps] -ap (4) [-jiapa] -iap (5) [-jwupJ -uap i
(6) [-iepl -ep
Finals (1), (3) and (5) are attested in the data. e.g.
(1)
T 224.435.1 ffi ?op Skt. abha
-936-
BTD Revisited--A Reconsideration of the Han Buddhist T
(3)
T 150.880.2 fi1 kap
T 458.435.2 iEgH ka iap
(5)
T 169.411.1 +Effii* *tSan lo puaP
Skt. kalpa; Gd. kapa, P. kaPPa
Skt. kdSyapa
Skt. candraprabha
This group is the checked final analogue of the Tan group and can perhaps
be reconstructed in parallel with it.
Sample Reconstructions:
I. H *gop > ?op [H *lop > lop W dop > doP
II. ff *krap > k5pJF *grap >'Iepffi *?rap > ?iipS *srap > $ap
{ *krap > kiip ffi *tshrap > tshiip i6 *gr.p > r5p
m. X *Urp > Cap {* *khap > khap E *n(i)ap > niap ffi *srp > sap
+* *ts(i)up > tsiap $ *diap > d2ap ffi *jap > iap
i* (*pt.p ? >) *PaP > PuaP 2 *baP > buaP
IY. Ih *g6p > rep F& *thep > thep f* *kep > k6p E *dep > d.p
3.35 Summary of the Vowel System. The vowel system reconstructed for
BTD is as follows:
ie u
o
f,
Within this system, the vowel *c (Yu category) is defective in distribution. It
is not distinct from *o in Ting's WJ materials, and the BTD data on it are
rather scant. [t is on the contrary primarily the TK materials, stemming from
the -|iankang area, which have led us to reconstruct *f, as a distinct entity. It
is therefore entirely possible that some Central Plains dialects of the BTD
period lacked *o and had the following eight-vowel system instead:
a
o
-937-
W. South Coblin
The distribution and behavior of the BTD vowels with reference to
allophonic variation and subsequent development of certain other sounds
(such as */g/ [g, r] ) enables us to speculate about phonetic detail in the
vowel system.
In BTD, the vowel *u probably had front and back allophones, [u] and
tyl. The phone [u] would have occurred in diphthongs before non-front
vowels, whereas [yJ would have occurred before front vowels. The absolute
final *-u (= QYS -jeu) may have been phonetically [y"] or [yuJ in the BTD
period. The diphthong xuo was perhaps phonetically [yo) in all environments;
and *uo was perhaps [y51. In fact, the vowels *o and *c may have been
allophonically fronted whenever preceded by front vowels or by palaf al
consonants.
BTD *e was probably rather high, and it
followed by *y sound other than *-i or *-!,
quality, perhaps phonetically [eal or [e"].
BTD *e was higher when preceded by *i
like, elsewhere.
BTD *-ii is a direct backward projection of oNWC -ii. This ONWC final
was perhaps a diphthong or triphthong (phonetically [iei] or [ial] ??) in
ONWC. In the fanqie formulae of Yan Shigu, which supposedly represent the
northwest area of 600 A.D., this -ii and the corresponding labizrl final, -uii,
take "non-yodized" guttural initials and thus behave as if they began wit.h
non-front rather than front vowels. This peculiarity is not found in contem-
porary fanqie of other areas, so far as I know. The phonetic vzrlue of *-ii as
-938-
e
e
I
e
was definitely non-front. \4rhen
it was probztbly diphthongzrl in
or *u and lower, PerhaPs e-
BTD Revisited--A Reconsideration "f !}'q IIglEg44!S!-I
restored for BTD is of course very uncertain, especially since this language
was probably not a northwest dialect at all. In BTD, *-i and *-ii contrast
after labial and guttural initials (i.e. the famous chongniu distinction of the
Qys; here = QYS -ji3 vs -i+). In the BTD data, both finals can transcribe
foreign i. BTD *-i is quite rare, while *-ii is common. For the nonce' a
guess might be that *-i was something like [r] while *-ii was someholtr more
like [i]. I find no direct evidence in the data that *-i was rhotacized or
velarized, as has sometimes been suggested; but this absence of evidence does
not preclude such interpretations as theoretical possibilities' For there musf
have been something about *-i which made it a less appropriate match for
foreign i.
(Accepted for publication 7 May 1993)
Abbreviations and Signs
BHS
BTD
Gd.
P.
Pkt.
ONWC
QYSS
SZ
TTK
wJ
XZ
YJ
Buddhist Hybrid Sanskrit
Han Buddhist Transcriptional Dialect
Gdndhdri
Pali
Prakrit
Old Northwest Chinese
Qieyun !1ffi System
Safighabhara
Shazhou i4l'l'l
Taisho Tripi[ aka
Three Kingdoms Period (220-265)
Wei-.|in Period (264-479)
Xuanzang f ftYijing ff)$
-939-
W. South Coblin
References
Bailey, H.W. 1946. "Gandharl." BSOAS 77.764'797.
Brough, John. 1962. The Gdndharl Dharmapada. London.
Coblin, W. South. 1981. "Notes on the Dialect of the Han Buddhist Tran-
scriptions." Proceedings of the International Conference on Sinology,
section on Linguistics and Paleography. Academia sinica. Taipei. pp. 121-
183.
-----. 1983. A Handbook of Eastern Han Sound Glosses. Hong Kong.
-----. 1986. "The Rimes of Chang-an in Middle Han Times. Part I: The Late
Western Han Period." Acta Orientalia (Copenhagen) 47.93-131.
-----. 1987. "The Rimes of Chang-an in Middle Han Times. Part II: The Early
Eastern Han Period." Acta Orientalia (Copenhagen) 48.89-110.
-----. 1988. "Notes on the Finals of a Northwest Dialect of Tang Times."
BIHP LIX, Part III (1988), pp. 835-890.
-----. 1989. "Notes on the Initials of a Northwest Dialect of Tang Times,"
Proceedings of the Second International Conference on Sinology, Section
on Linguistics and Paleography, Taipei, 1989. Vol. I, pp. 725-744.
-----. 1990. "Notes on Sanghabhara's Mah-mdyuri Transcriptions." Cahiers de
Linguistique Asie Orientale XIX, pp. 195-251.
-----. 1991a. Studies in Old Northwest Chinese. Journal of Chinese Linguistics
Monograph Series Number 4.
-----. 1991b. "A Survey of Yijing's Transcriptional Corpus." Yuyan yaniiu ffif;
wn rc91.1.68-92.
-----. 1991c. "Dentilabialization in the Tang-time Dialects of Shazhou." T'oung
Pao LXXYII (1991). pp. 88-107.
-940-
BTD Revisitedl--A Reconsideration of the Han Buddhist TranscriPtional Dialect
-----. Ms. 1. "Comparative Studies on some Tang-time Dialects of Shazhou'"
To appear in MS.
-----. Ms. 2. "Remarks on Some Early Buddhist Transcriptional Data from
North-west China." Unpublished.
Dong, Zhongsi HtsE. 1978. Yan Shigu suozuo yinqie zhi yaniiu ffifrfrfrFfilF#
wzwfr. Doctoral dissertation, National zhengzhi university. Taipei.
Edgerton, Franklin. 795?. Bud.dhist Hybrid Sanskrit Grammar and Dictionary'
Two vols. New Haven.
Fussman, G. 1989. "GAndharl 6crite, Gandhari parl6e." Dialectes dans les
litt&atures Indo-aryennes. Paris. pp. 433-501.
Luo, ChangpeiffiHH, and Zhou Zumo E+EH' 1958' Han Wei lin
Nanbeichao yunbu yanbian yaniiu i$finffffiJt$f HHElSiEEffifr . Peking'
Matisoff, James A. 7g7?. The Grammar of Lahu. university of california
Publications in Linguistics 75. Berkeley.
Norman, Jerry. 7988. Chinese. Cambridge.
Norman, K. R. 1983. "The pratyeka-buddha in Buddhism and Jainism." in
Buddhist Studies Ancient and Modern. Collected Papers on South Asia'
London. pp. 92-106.
Pelliot, Paul. 1933. "Papiyan > 1fr.6J Po-siun." T'oung Pao 3O:85-99'
Pulleyblank, E.G. 1962. "The Consonantal System of Old Chinese"' Par[ I:
AM (n.s.) 9.58-144. Part II: AM (n.s.) 9.206-265'
-----. 1978. "The Nature of the Middle Chinese Tones and their Development
to Early Mandarin." JCL 6.773'203.
-----. 1g7g. "Some Examples fo Colloquial Pronunciation from the Southern
Liang Dynasty," in Studia Sino-Mongolica: Festschrift fiir Herbert Franke,
ed. by Wolfgang Bauer' Wiesbaden. pp. 315-28.
-941-
W. South Coblin
-----. 1983. "stages in the Transcription of Indian Words in Chinese from
Han to Tang," in Sprachen des Buddhismus in Zentralasien. Wiesbaden'
pp. 73'7O2.
-----. 1984. Middle Chinese. Vancouver.
-----. 1986. "The Locative Particles yii+, yiib!, and hu +." IAOS 106.1-12.
Schuessler, Axel. Ms. "Palatalization of Old Chinese Velars"' To appear in
JCL.
Shao Rongten H$#6. 1982. Qieyun yanjiu +{Jffiffifr ' Beijing'
-----. 1983. "Guyun yu-hou liang bu zai Qian-Han shiqi de fenhe" EEE +*{'f+
ffi E[ A ffi iH ffi tr fr! r] e' Zhongsuo vuvanxuebao + W;E E g' 1' 127- 138'
Shi, Xiangdong ffifif H. 1983. "Xuanzang yizhe zhong de Fan-Han duiyin he
Tangchu zhongyuan fangyin" y x## + wft iHsiEf[tru + IRfr E -
Yuyan yanjiu 1983.1.27-48.
Svantesson, Jan-Olof. 1988. "fJ." LTBA 11.1.64-133'
Takata, Tokio E E ffi ffi . 1988. Tonka shiryo ni yoru chugokugo shi no
kenkyu "?L''/E'RN!E l. A +tr;E-t affifr. Tokyo.
Ting, Pang-hsin T *ls ffi . 7975. Chinese Phonology of the l4/ei-Jin Period:
Reconstruction of the finats as Reflected in Poetry. Institute of History
and Philolog/, Academia Sinica, Special Publications No. 65, Taipei.
-----. 1986 . Danzhou cunhua {E ,'J\l FJ ff; . Institute of History and Philology,
Academia Sinica, Special Publications No' 84, Taipei'
wang, Xian T.ffi. 1961. "Qieyun de mingming he Qieyun de xingzhi" wwill
fr&friwwH!{tH . ZGYW 4.76-25.
Woon, Wee-lee ttrflj. 1987. Hainan fangyan trffif fr ' Hong Kong'
Yu, Minft{E{. 1984. "Hou-Han San-guo Fan-Han duiyin W" lk1{=ffiftiHYj
B#. Zhongguo yuwen lunwenii + B#X-# X-R' Tokyo' pp' 269-319'
-942-
VY
BTD Revisited--A Reconsideration of the Han Buddhist TranscriPtional Dialect
-----. 1989. "Han-Zang tongyuan zipu gao" iHffiEiffi+#ffi. MZYW 1989'1'56-
77.
Yuchi, Zhiping trf )Ei6+. 1985. "Lun Sui-Tang Chang-an-yin he Luoyang-yin
de shengmu xitong -- ji* da Liu Guanghe tongzhi" ffiffitrFfitrf[i&WEWWE*ffi- ffi$grjtrffflfi]ff.. Yuyan yanjiu 2.38-48.
Ziircher, E. 1959. The Buddhist Conquest of China- Leiden.
-----. 7977. "Late FIan Vernacular Elements of the Earliest Buddhist Transla-
fions." loarnal of the Chinese Language Teachers' Association 12.777-203.
-943-
Top Related