On the Origin of the Perfect Participles in / in the
Neo-Indian Vernaculars.
By L. p. Teggltorl.
The last discussion made on this subject, is that by Professor
Sten Konow in his Note on the Past Tense in Maräthi, inserted
in JBAS., 1902, pp. 417—421. As may be gathered from the
title, the above Note treats the question from the point of view of
Maräthi only, and p. 417 the hope is expressed that a discussion 5
of the whole matter may "soon proceed from a more competent
authority". Being absolutely far from being such an authority, I
ought to refrain from any attempt to interfere in the matter, but for
my having struck upon some new evidence, which, if made known,
will greatly advance our knowledge of the subject. lo
That the theory formerly resorted to by scholars for the ex¬
planation of the Neo-Indian perfect participles in I, namely that
they are derived from Sanskrit -ita, through Prakrit -ida ]> -ida,
whence -ira > -ila, was not satisfactory, had long been recognized,
even before Dr. Hoernle published his Gaudian Grammar (1880), is
where the above derivation is defended for Ihe last time. The
first serious doubt about the correctness of it was raised, I think,
by Mr. Kellogg in the Chapter on the Origin of the Verbal Forms
in the first edition of his Hindi Grammar (1875), where he
suggested that "in this participle in I we have a form as ancient io
as the Sanskrit participle in t, and having no connection with it".
His suggestion was not disregarded by Mr. Beames, who in the
third volume (1879) of his Comparative Grammar advanced the
opinion that the Neo-Indian participle in I might be somehow con¬
nected with the Slavonic preterites in I, and possibly represent the 25
survival of an ancient form not preserved in classical Sanskrit nor
in the written Prakrits, which was in existence before the sepa¬
ration of the various members of the Indo-European family. But
Dr. Hoernle resorted again to the customary derivation, and tried
to remove the difficulty involved in the passing of a dental into a so
cerebral and then back again into a dental, by omitting the two
intermediate steps d > r and deriving I directly from d. This
87*
572 Tessitori, Origin of the Perfect Part, in I.
was ingenious , but the explanation was not yet satisfactory on
account of there being no sufficient evidence on which to rely for
the change > Z. The three Mägadhi perfect participles ^1%,
1%, {Mrcch), quoted by Dr. Hoernle in support of his view,
5 proved, if anything, that this was wrong, for the Mägadhi, besides
'F^, possesses also •s. (Var. ii, 15; Mrcch),
which point out that the set «B^ etc. is probably to be corrected
into ^i§fe, T§fe, and to be explained as a modification of
etc.^). Besides, the change (i > Z is very rare in Prakrit and
10 in some of the cases where it apparently occurs (eight or nine in
all), it is really doubtful whether I represents the pure dental, or
the cerebral I, derived from d, through d^).
The clue to the right derivation was first given by Sir Charles
Lyall in his Sketch of the Hindustani Language, published in
15 the same year (1880) as Dr. Hoernle's Gaudian Grammar, where
he suggested that the I was a diminutive \ suffix. To do to the
latter scholar full justice, however, it must be mentioned that he
also had independently come to a similar conclusion in regard to
the Gujaräti perfect participle in -("Zö, which he explained as merely
20 a pleonastic form of the participle in -yo. The first definitely to
connect the modern perfect participle termination with Prakrit -ilia
was Sir R. G. Bhandarkar in his Wilson Lectures^), and his theory
was subsequently discussed and proved by Professor Sten Konow
in his note mentioned above. From the latter, we learn also that
26 Sir George Grierson too had previously come to the same opinion.
Before proceeding to exhibit tbe new evidence which I have
discovered on the subject, I shall give a summary prospect of the
forms under which the perfect participle in I is. met with in the
different vernaculars. Taking the root TTJ "to strike" as the
30 standard, we have in Neo-Indian the range of forms following:
Maräthi: »ITft^T CTTT^T);
Bihärl : «RTT^ ;
Oriyä: «if*.«!!!, TTf^^ (only used in the oblique form •%);
Bengali: ^ifK^ (ditto);
S6 Guiaräti: TT^Wt, TT^^.
F'rom the above it will be seen that, with the exception of
Gujaräti, which has e, the vowel before the I wavers from i to a.
That the latter is but a weakening of the former vowel, and there -
1) See Pischel's Grammatik der Prakrit-Sprachen, % 244.
2) Ibid., § 24S.
3) I owe this information to Sir George Grierson. Professor Konow does not make any mention of Sir R. 6. Bhandarkar's view, in his Mote.
fore both -ila and -ala are referable to the same prototype, namely
Prakrit -illa, it may well be taken for granted, though in Maräthi
some forms in -ala occur, like ^^T, 'Wl'si I and a few others,
which apparently seem to point out -alia as their origin.
Now, in Prakrit the taddhita suffix -illa, a form traceable to s
Sanskrit -ild^), is very largely used, mostly indeed to derive posses¬
sive adjectives corresponding in meaning to the Sanskrit adjectives
in -mat, -vat, or adverbial adjectives indicative of place or time.
Sometimes, however, the same suffix is also appended to nouns or
adjectives pleonastically, i. e. without altering the original meaning lo
of the word (^"^, Hc. ii, 164)''). It is in the latter mode of
employment of -illa, that we find the reason why it came to be
appended to perfect participles too , these evidently partaking of
the character of adjectives. The only instance of the suffix -illa
being added to perfect pai-ticiples in Prakrit, Professor Konow seems ib
to know of, is ^fifff^W (= Sanskrit ^'H'TT) occurring in the
ArdhamägadhI of Vivähapannatti, 961 , and quoted by Professor
Pischel § 595 of his Prakrit Grammar. Prom the way in which
Professor Pischel quotes it, it is clear that he considers it as an
exceptional form. This does not seem to be the case, however, as 20
the evidence of perfect participles in -illa in Prakrit does not end
hei*e. How the very learned and accurate German Professor could
have overlooked the many instances of perfect participles in -elliya,
-ellaya « -illiya, -illaya, cf. Pischel, Op. cit. § 119), which occur
in the Äva6yaka tales edited by Professor Leumann (1897), I am 25
not able to understand, the omission appearing still more .strange
if we consider that § 595 of his Grammar he quotes from the
Avaiyakas the forms 'ftfflRl, Tf^^lT and ^fWlf. The fact is
that the Jaina Mähärästri of the Avaäyaka tales abounds in perfect
participles in -elliya, e. g.: ^TTlflraT f. »Come" (p. 27) ^tflPn f. so
"Betrothed" (p. 29), Wff Hint "Spilt" (p. 44, note), etc. Two such forms, to wit f^^lWT f. "Seen" and »TftrfiraTS) f. "Spoken to",
are likewise found in a Prakrit version of the Cakravedha, quoted
by Gunavinaya in his Sanskrit commentary {tlka) on the Bhava-
veraggasayaya, whereof a MS. in the Library of the India Office*). 35
This Prakrit Cakravedha is quoted under Stanza 92 of the Bhava-
veraggasayaya, and the language in which it is written is so much
akin to that of the Ava^yakas that I should not be surprised at
1) Pischel, Op. cit., g 194.
2) Op. cit., § 595.
3) MS. HfWif^.
4) S 1564, a. It is through Dr. F. W. Thomas' kindness that I was able to collate tbis HS. at the Biblioteca Comunale of Udine (Italy).
4 2 *
574 Tessitori, Origin of the Perfect Part, in I.
all if it was taken from the latter. Sporadic instances of perfect
participles in -illiya are certainly found in other Jaina Mähärästri
texts. Stanza 292 of Dharmadäsa's Uvaesamala I have remarked
the form WfwflRI, used in the acc. f. sing, as a substitute for
8^ (Sanskrit "Obtained".
In the above instances , however scanty they may seem , we
have a continuous chain of evidence from the ArdhamägadhI of
the Jain Arigas to the uncultivated Jaina Mähärästri of the
Ävaäyaka tales, and thence to the literary Jaina Mähärästri of
10 the Uvaesamölä. Through a careful consideration of this evidence,
we may easily arrive to conclude that the custom of appending
the pleonastic suffix -illa, -ella to perfect participles is a feature
of the popular speech, whereof the existence is to be traced at
least as back as the redaction of the Jain Canon. The popular
16 character of these participles in -illa, -ella is amply evidenced by
the fact of their being very common in the language of the
AvaJyakas, which represents for us the most popular form of
Jaina Mähärästri known to the present day, and their being quite
exceptional in tbe literary Jaina Mähärästri, where they are, no
«0 doubt, to be explained as a borrowing from the language current
on the mouth of the people.
If we now come to the Apabhraipäa stage, we find that we
are here very far from being so fortunate , as we were in the
Prakrit. No instances of perfect participles in ll occur in the
«6 literary texts collected by Professor Pischel in his Materialien zur
Kenntnis des ApabhrarnSa (1902), nor can we find any sure
traces of such forms in the later Apabhramia of the Prakfta-
pairigala. The almost exceptional instances of participles in -ala,
which are met with in the latter work, cannot be relied upon at
so least till a better edition of it is available. Even the most evident
of such forms, to wit ^^«ü (= Sanskrit ^fT) is subject to suspicion,
for it cannot be absolutely excluded that it must be read T^R35
and explained as a derivation from ♦J'f« (< Sanskrit *^^TZ).
Instances of d having passed into l^) are very common in the
ib Prokrtapaingala. Cf. fiU"*!* < f«U'W« (< Sanskrit f'^ISZ),
^mfSB < (< Sanskrit 1»^), < (< Sanskrit
Hflfn), etc. In the case of ^^IW, however, I think it more
likely to derive it from Prakrit (< Sanskrit *^f7Wl),
whence, by simplification of the double consonant, *J1,»S , and
40 thence W^W. That the process of simplification of double con¬
sonants had already begun in the later Apabhraipäa, it is testified
1) In common editions, of course, no distinction is made between I and I.
by the passive forms in -ije, for -ijjai, which are frequent enough
in the Prähtapaingala. But *«J'^»!| might as well be taken as
staying for (<! Sanskrit "^nn), the -ala suffix occurring
in other words in Apabhramäa, e. g.: {Prabandhac., 121, 10).
Coming now to the early stage in the development of the 5
Neo-Indian Vernaculars, I am glad I can quote a very important
evidence , which throws the most clear light on what was the
darkiest phase in the whole process of development of the modern
participles in /. I discovered it in an Old Western Räjasthäni^)
MS. in the Regia Biblioteca Nazionale Centrale of Florence, bearing 10
the title Nemijinavaraprabandha , and dated in the year Saipvat
1641"). Here is the passage where the evidence in question occurs:
^ WT ^^^UTT ^f^WJ I
fTTfT 'fPNr ^fwr « «i, «
•[Lord Neminätha] heard the cries of beasts, [and] out of pity 15
shook [his] head".
The two forms ^fuiQll and ^ftUTT are obviously two perfect
participles in the masc. pl. {pluralis majestatis) from ^««1. and
Win[, respectively. The retention of the double consonant ll, which
in Old Western Räjasthänl ought regularly to be simplified to /, »0
is a feature of the Apabhraqisa, which in old vernacular poetry
is often allowed at the end of a verse. It is in this sort of
archaism that we have the most positive proof that modem I is
derived from ll, and at the same time a sure testimony to the
participle in -illaa being not unknown to the Apabhrarpia*). It K
must be noted, however, that the two forms quoted above are the
only ones I came across in the Old Western Räjasthänl MSS. I
have examined to this day, and are therefore to be considered as
exceptional forms from the point of view of the Old Western
EäjasthänI. This remark will prove of use further on. so
The chasm between ArdhamägadhI -iUa and Neo-Indian -üa
being thus bridged over, there is no more discontinuity in the
derivation, and there can be no possible doubt about the correctness
of our conclusions. We have begun from the Prakrit and gradually
proceeded to the Neo-Indian stage, but we might also follow the ss
reverse course with as much success. The modem perfect participles
1) Under the term 'Old Western Räjasthänl" 1 understand the common parent of Modern Qi^arätl and MärwSrI (cf. JRAS., 1913, p. 554, n. 1). For
further particulars see my Notes on the Grammar of the Old Western
Räjasthänl etc. in Ind. Ant., XLIII (1914), p. 21 ff.
2) The HS. is registered in Professor Pavolini's catalogue under No. 715.
3) In Old Western Räjasthänl a consonant is never doubled at the end of a Tene, if originally single in Apabhramsa.
576 Tetaitori, Origin of the Perfect Part, in I.
in I contain sufficient evidence to show that they are derived from
ll, and not from {r <^ d ■<) d <^ t, as was formerly maintained.
As far as Maräthi is concerned, Professor Konow has shown that
every intervocalic I in this language is derived from Prakrit ll
6 whereas I is from I, and therefore "the suffix {-ila) of the past
tense probably goes back to an older -illaa". The same argument
may be applied to the case of Gujaräti, where the same law is in
existence. Prom the prospect given above, it appears that Gujaräti
has for the participle in I two forms, to wit 'RT^^ and TT^'Sr
10 of which the former is inflected and the latter uninfiected. Now
to account for dental I in these forms, there is no other possible
explanation than trace it to ll. Had ♦llX'^ derived from Apabhramia
♦flTfl,^^ < Sauraseni *TrfX;^^, it is clear that we ought to
find I in the place of /, just as much as we have ?T§fe from
15 Apabhramsa (•I'Slf..
The criterion adopted for Maräthi and Gujaräti fails to be
applicable to the remaining vernaculars, which have no cerebral
In the case of these, we must look for some evidence of a different
kind. We find it in the so-called irregular perfect participles such
20 as Bihärl ftsf^^ connected with the verbal root ^iT "to do",
thi, iNf connected in meaning with ^ "to go", *J<'!I,
connected with "to die" etc. ^). In these forms the I could never
be explained as a modification of Sanskrit t. Even granting that
Sanskrit ITT: might result in *1% ('!*>) in Mägadhi Prakrit, and
25 Sanskrit in *^ (^[^) , this would not account for the i
which appears before the I suffix in Bihärl and *JThese
forms, on the contrary, are easily explained if we trace them to
Prakrit *iy9 or IXm [cf. the Jaina Mähärästri form Wnjt%?I
cited above] « Sanskrit *lf7ra) and gi[ir or TTflT «
so Sanskrit *^f7!5I), respectively. The form ^ is obviously but a
contraction from the ♦J'«<!l which is found in the Präkrtapaingala.
Similarly fl^ and f«lift?^ are from *^5Tir, *^ni^I and *ft!1[5r,
*f«<m«S respectively, in the latter case h being probably but an
euphonic element inserted to obviate the concurrence of the two
35 IS. Cf. also the forms f^f?^ from^, and fwf^^ from where
the same insertion of h has taken place.
It remains to explain the Gujaräti forms, which, as seen above,
have e instead of i before the I suffix. Here one might be inclined
to think that e is the regular result of a contraction, and to explain
1) See Hoernle's Gaudian Grammar, § 304, and Grierson's LSI.,
Vol. V, Pt. ii, pp. 1, 39, 52. Cf. also the Marafhl forms ifWT. ^WT.
TT^^ as derived from Apabhramia "■TTf'CT'^j possibly through
*«11 ^t,*!!^, by dissimilation of ii to at. I do not think this
would be correct. The Old Western Räjasthänl evidence produced
above tends to show that in this language the regular termination
for the pleonastic perfect participle was -ilaii, not *-ailaii, *-elaU, t
or, what comes to be the same, that in the case of ordinary verbs.
Old Western Räjasthänl too , like the other modern vernaculars,
formed this participle by adding -Hail directly to the root. We
are , therefore , to take as the Old Western Räjasthänl
original of Modern Gujaräti *t I'^.'sH , and to explain e as derived lo
from i. The passing of i to e was probably effected through the
former vowel being first amplified to at, whence e. Cf. the case
of Modem Gujaräti "«ifllcto^^ "Forty-two", which is from Old
Western Räjasthänl f^fTTW^^f, through W^TTT^f^, which last
form is evidenced by the Navatattvabälävabodha and Adinätha- 15
caritra, two MSS. pertaining to the Regia Biblioteca Nazionale
Centrale of Florence^), and also by Somasundara's UpadeSamcdä-
vahodha (Samvat 1567), whereof a MS. was kindly procured to me
by Sri Vijaya Dharma Süri. Other illustrations of the same change
in Old Western Räjasthänl are A<\y (<C Sanskrit flfll), occurring so
in Daäadrstänta, 1, and 1^!^ < f*!^ <; Apabhramsa « Sanskrit
1<T^), occurring in Sädhuhamsa's Bälibhadracaüpal, 10"^). The
same tendency to amplify i to aü is found in Märwäri, e. g. ^TTT
(<; Sanskrit f'T?!!), and ^§ < f^^, occurring in the Nasaketa-rl
kathä^). Cf. the analogous case of Old Western Räjasthänl a > di, ss
whence Gujaräti e and Märwäri ai, whereof illustrations will be
found § 2, (3) of my Notes on the Grammar of the Old Western
Räjasthänl.
Lastly, I wish to make a remark concerning the diffusion of
the perfect participle in /. From the prospect given above , it so
appears that this participle is now-a-days found in Maräthi, Bihärl,
Oriyä and Bengali, four languages belonging to what Sir George
Grierson has termed the Outer Circle , and also in Gujaräti , a
language belonging to the Central Group. If we take to consider
the Prakrit and Apabhramsa stage , we find that we have sure S8
evidence of the use of the participle in -illa only in the Ardha¬
mägadhI and Jaina Mähärästri, two languages of the Outer Circle.
The observation that this_ participle is especially common in the
Jaina Mähärästri of the Avadyaka tales, leads us to conclude that
it must have been a feature of the vulgar speech , i. e. of the 40
1) No. 602 and No. 700.
2) Both these MSS. are in the above mentioned Library of Florence, and are classed under No. 756 and No. 781, respectively.
^) R. S. O., vi (1913), pp. 113—130.
578 Tessäori, Origin of the Perfect Part, in I.
Mähärästra and Magadha Apabhramia, whence it was introduced
into the literary language. We have seen that no instances of
participles in -ilJa are met with in the literary Öaurasena Apa¬
bhrarpia of Hemacandra, and this is quite consistent with the fact
i of the -illa diminutive suffix being utterly foreign to the genius
of the äaurasenl^). The first sporadic and dubious instances of
participles in 1 we find in the languages of the Öaurasena or Central
Group, are only in the later Apabhramia of the Präkftapairigala,
and in the Old Western Räjasthänl, where they are, no doubt,
JO quite exceptional. It is clear that the perfect participles in 1 were
from the very beginning a feature of the Outer Circle, and there¬
fore absolutely strange to the languages of the Central Group, and
only in comparatively recent times were introduced into the Old
Western Räjasthänl language of the latter group, and had their
15 utmost development in the Modern Gujaräti. This is in perfect
agreement with Sir George Grierson's theory, according to which
the original language of Gujarat was a member of the Outer Circle
of Indo-Aryan speeches
1) See Pischel, Op. cit., § 595.
2) LSI., Vol. ix, Pt ii, p. 326—27.
Textkritische Bemerkungen
zur Käthaka- und Prasna-Upanisad.
Von Alfred Hillebrandt.
F. 0. Schräder gebührt das Verdienst, den ersten Grund zu
einer textkritischen Ausgabe der üpanisads gelegt zu haben, und
sein erster Band der „Minor üpanisads' wird allen willkommen
gewesen sein," die den Versuch machen, diese trotz ihrer Berühmt¬
heit oder vielleicht gerade wegen ihrer Berühmtheit in einzelnen 5
Partien recht unvollkommen überlieferten Werke zu übersetzen.
Erst wenn ein Codex üpanisadum vorhanden ist, der die Lesarten
der wichtigeren Handschriften verzeichnet, wird es möglich sein,
dem vollen Verständnis dieser alten Traktate, die an der Spitze
der indischen Philosophie stehen, näher zu kommen. Einige Kon- lO
iekturen zu Texten, die Schräder noch nicht gibt, seien hier vor¬
geschlagen.
Käthaka-Upanisad 1,28:
ajiryatam amrtänäm upetya jiryan martyah kvadhahsthah prajänan
abhidhyäyan varnaratipramodän atidirghe jivite ko rameta || is
In diesem Verse bereiten die Worte kvadhahsthah und varna¬
ratipramodän dem Verständnis Hindernisse. Whitney^) deutet
kvadhahsthah als , standing below', ,with the ku of depreciation
prefixed"; Böhtlingk (SKSGW., 14. Nov. 1890) sagt, mit kvadhah¬
sthah oder der v. 1. kva tadästhah sei nichts anzufangen und ver- 20
mutet darin einen von prajänan abhängigen Akkusativ; Kern liest
sadhastham. Unseren Ansprüchen genügt nur die Lesart kvava-
sthak (d. i. ku-avasthah), die ich in der Fußnote der Anandäirama-
ausgabe (Heft 7, Jahr 1897) als aus einem in K&Si gedruckten
Text entnommen finde, ,in übler Lage': das sind die Gebrechen 25
des jiryat. Aber sie reicht nicht aus. Zu upetya muß ein Objekt
gesucht werden, das allein in prajänan stecken kann. Wie wir
cintäm upeyivän im Mbh. (PW) finden, so wird man auch upetya
prajflänam lesen und annehmen können , daß prajflänam leicht
und um so leichter verlesen werden konnte , als sonst prajüänam 30
mit einem Gen. obj. nicht verbunden zu werden scheint.
Die zweite Schwierigkeit finden wir in varnaratipramoda.
1) Transactions of tlie American Pliiiologicai Assoc. 21, S. 97 (1890).