Professional Documents
Culture Documents
Levman, Bryan. 2010. "Is Pāli closest to the Western Aśokan dialect of Girnār?" Sri Lankan
International Journal of Buddhist Studies (SIJBS). 79-108.
Introduction
Despite Buddhaghoṣa’s belief that Pāli was in fact Māgadhī, the original language of the
Buddha (quoted in Norman 1990B, 128-30), we know that it is a mixed language containing
mainly western elements with some Eastern features, plus Sanskritizations normalized for
ecclesiastical purposes (Lüders 1954; Norman 1983, 4; Lamotte 1988, 563; von Hinüber
1994,180ff). Buddhist Hybrid Sanskrit (BHS) is another Middle-Indic (MI) dialect, believed to
stem from the school of the Mahāsāṃghikalokottaravādins (von Hinüber 2001, ¶43), which has
undergone even more Sanskritizations than Pāli. Both dialects appear to have developed “side
by side at the times of their origin” (von Hinüber 1994, 192); they “flow from the same source”,
which von Hinüber calls “Buddhist Middle Indic” (von Hinüber 2001, ¶40). The nature of what
this source is has occupied researchers for decades. As early as 1916, Geiger surmised that it
was a lingua franca containing elements of all dialects but free of the most conspicuous
dialectal phenomena (p. 3). He variously termed it a Hoch- und Gebildetensprache (high and
scholarly language), a Verkehrssprache (a lingua franca or interlanguage) and a Kunstsprache
(artificial language). Helmer Smith called it a Koine Gangétique in 1952 (p. 178). Bechert (1980,
34) wondered if differences in Pāli prose and verse pointed to a middle Indic Dichtersprache
(poetic language) which was transregional in use. Przyluski believed that “The first canonical
texts were written in the Magadhan dialect. When the Sthavira and Sarvāstivādin sects
asserted themselves in the regions of Kauśāmbi and Mathurā, each of them drew upon
a literary language which was a source of prestige and an instrument of propaganda.
Thus, the Scriptures were eagerly translated into Sanskrit at Mathura and into Pāli at
Kauśāmbi” (quoted in Lamotte, 1988, 584).
The mystery of Pāli’s provenance remains to this day. Lamotte has provided a comprehensive
list of the similarities between Pāli and Girnār (1988, 565), which list is often cited as evidence
of Pāli’s “western connection” (see below). I list here a few of the principal differences, in no
particular order:
1) all the Aśokan Prakrits have ā or āva (Gir.) for (Skt.) yāvat, yet Pāli reverts to yāva (“as
much as”, “until”).
2) For the world “self”, Gir. has atpā < (Skt.) ātma, while Pāli always has atta. Yet Gir.
preserves some absolutives in –tpā which Pāli echoes in –tvā (e.g. Gir. dasayitpā, P.
dassetvā, “having seen”). 1 So, apparently the western version (and certainly the
northwestern version) was ātman > atpa > atva. The form atva is well attested in the north-
western dialect of Mānsehrā (M.) (RE 12 F2, passim) and although the Shāhbāzgaṛhī (Sh.)
RE adopts ata-, Mehendale maintains this is an eastern borrowing and the proper
northwestern treatment of ātman is > atpa > atva, (1956/57, 167). Nevertheless, in the case
of Skt. ātman, P. is closest to Sh. and the dialects of Kālsī (K.), Jaughaḍa (J.) and Eṛṛaguḍi
(Eṛṛ) which also have ata. The form atva- survives in the Gāndhārī version of the Dhp, verse
1 Although the –tvā structure apparently did change to –ttā, as von Hinüber has pointed out, and it was
later re-Sanskritised back to –tva to prevent confusion with the nomen agentis (1994 188f.)
2 Captial letters (F, etc.) refer to location of the text with the rock inscription as per the Hultzsch system of
notation, found in his Inscriptionsof Aśoka monograph (1969).
362 (ajātva-rado vs. P. ajjhatta-rato, “delighting in one’s self”). To complicate matters, appa
also survives as an ArdhaMāgadhī (AMg.) or eastern form. 3
3) In some cases Gir. is more phonologically advanced than Pāli (e. g. RE 13 O, (Skt.)
samacarya > (Sh. K.) samacariyaṃ/samacaliyaṃ, > (Gir.) samacairaṃ (“spiritual calm”),
yet Pāli reverts to the (phonologically) older form (samacariyā) as used in Sh. and K.
4) Gir. has mahiḍā (RE 9C, Skt. mahilā > mahiḍā) showing an –l- > -ḍ- change, 4 but Pāli
reverts to Skt. mahilā (“woman”).
3 The appa- form of ātman (P. atta-) is well attested in ArdhaMāgadhī and Māgadhī (Pischel ¶277). In the
REs, the Brāhmī script shows a conjunct consonant with a pa- on top of a ta-, which is normally read ātpa.
Pischel says that it should be read āpta- by way of transposition of stops , ātma- > *atva- > *ātpa- > āpta-
atta (P.), based on the rule of consonant assimilation that between equals (-p- and –t- being equal;
Woolner ¶33, Pischel ¶270) the second prevails. If the reading were ātpa as Hultzsch has interpreted it in
RE 12 Gir., then the normal derivation is appa, which is only found in AMg. – most reflexes (P. and the
other REs) are atta- or atva in M. and ātpa in Gir.. The change of –m- > –v- is fairly common in Middle-
Indo Aryan (MIA , Pischel ¶251) but the change –v- > -p- is uncommon, it usually being the other way
around, as a form of intervocalic lenition (Pischel ¶199). Munda characteristically has an interchange of –
m- and –p- (Kuiper 1991, 37). We find a similar change with aspiration, in RE 13 B (K.) tasmāt > taspāt >
tapphāt > tapphā (“therefore”) and in Separate Edict 2, I, L: (Skt.) asma > *aspe > (Jaughaḍa = J., Dhauli
= D..) apphe (“we”, written as aphe) and (Skt.) yuṣma > yuṣme > *tuṣme > *tuṣpe > tupphe (“you” pl.,
written as tuphe). The –v- > -p- phenomenon seems to be most prevalent in the west and northwest. See,
for example, RE 4 B: (Gir.) dassayitpā < (Skt.) darśayitvā; RE 9 H also attests to this change in Sh. and
M. where (Skt.) svāmika > (Sh. M.) spamika (“master”) and RE 6 L shows the same change in (Skt.)
svarga > (Sh. M.) spagra, > (Gir. K. Dh.) svagga (“heaven”). RE 10 A has (Gir.) tadātpano < (Skt.)
*tadātvanam > (K. Dh.) tadatvāye, and > (Sh. M.) tadatvaye (“present time”). In RE 12 F passim, the
Mānsehrā edict shows several versions of atva-, while Gir. has ātpa-, K Sh. and Eṛṛ ata (“self”). In the
minor rock edicts - MRE 1 G, - several locations, Brāhmagiri (Br.), Eṛṛ., Pānguḍāriyāṃ (Pān.), Rājula-
Maṇḍagir (Rāj.), and Uḍegoḷaṃ (Uḍe.) have mahātpa for (Skt.) mahātmā (“great soul”) which Mehendale
sees as a northwest influence on these southern rock edicts (1955, 90); in the same section there is also
a (common) –p- > -v- change: (Skt.) prāptum > (Sahasrām = Sah.) pāvatave with other versions showing
pāpotave (“to achieve”).
4 Pillar Edict (PE) 7 SS: (Skt.) dharmalipi > dhamṃalipi > dhamṃalibi (“religious edict”), also > RE I A, et
al. (Sh. M.) dhramadipi, with an unusual l- > d- change which Woolner (1924, 97) says is Iranian in origin.
See Pischel ¶226 (-ḍa- > -ḷa-, but not vice-versa; this only happens in Munda (see Kuiper 1948, 6 which
shows equivalence of d/r/l/ in proto-Mundan). For change -ḷ- > ḍ-, see also PE 5 B where (Skt.) ḍuli > daḷī
(Delhi-Toprā =Top.), but (Allāhābād-Kosam=All.) > duḍī (“turtle”). PE 5 C also has another example of
5) Gir. has yārisaṃ < (Skt.)yādṛśa (“which like”), but Pāli has yādisa (a NW form). Eastern,
southern and Mānsehrā forms are ādisā/ādise/adiśe (RE 4 C).
6) It appears that the normal form of Skt. guru (“heavy”, “teacher”) in Gir. is the same, i. e.
guru. In RE 9 G, Gir. and K. have gurūnaṃ < (Skt.) gurūṇām, while Sh. has garuna; This is
also the case in RE 13 G, where Sh. now shows guru and only K. (a northern dialect) has
galu. Later in the same rock edict Gir. has a single instance of garumat for Skt. gurumat
(“serious”), while the other REs show gurumat. The normal and oldest form attested in P. is
garu, similar to the northern and northwestern form. Although guru is used in P., it is a
younger form dating from the commentaries.
7) In RE 9 C, Gir. has chudaṃ (< Skt. kṣudram), M. has khuda, K. has khudā and Dh. has
khudaṃ (Sh. putika; but khudra in RE 10 E), yet Pāli reverts to khuda (“futile”). Note also in
RE 13 L where Gir. has chamitave and Sh. has kshamitaviya (dat infinitive, “to bear”) and
Pāli does not use the western form, but the northwestern form khanti < Skt. kṣanti.
8) How does Pāli end up with the gerundivium ending – tabba? Gir. Sh. and the eastern
dialects usually have –tavya and -tavya should go to -tavva, by regular assimilation (Geiger
¶51-3) and –v- > -b- is not a normal change (–v- > -p-, although unusual, is well-attested as
noted above). Kuiper calls the interchange of -v- and –b- “a well-known crux” (1991, 33)
and notes the -b- > -v- change was an attempt to Sanskritize a foreign word with a foreign
phoneme, –b-. The change –v- > -b- is “inconceivable” (for the RigVeda; ibid, 34) and one
does not observe it in the Prakrits; however it does happen later in Pāli (Geiger ¶54-6) and
von Hinüber (2001, ¶255) notes the use of –tavvo in Pāli inscriptions and oldest
manuscripts. There are also several uses of –bb- for –vv- in the Dhammapada (Dhp), e. g.
verse 53 (P. kattabbaṃ vs. PDhp 5 kātavvaṃ, “to be done”) and verse 113 P. udayabbayaṃ
vs. PDhp udayavyayaṃ (“rising and falling”). Sn verse 537 has the word paribbajayitvā,
which, as Norman points out is a pun on pari + √vraj (to wander) and pari + √vṛj (to reject,
abandon) which “only works in a dialect where –bb- changes to –vv- ” (Norman 2006A, 263).
However it is still not clear phonologically why –v- > -b- unless through the process noted
change -ḍ- > -ḷ-, i. e. (Skt.) eḍaka > (Top.) eḷakā (“ram”). Also RE 2 A: (Skt.) kerala > (Sh.) keraḍa (Proper
Name = PN) and RE 9 C: (Skt. mahilā > (Gir.) mahiḍā (“woman”).
5 Patna Dhammapada = PDhp
above (with ātman) of –v- > -p- > -b- (i. e. –tavya > -tavva > -tappa > -tabba ). The latter
lenition (-p- > -b-) is very common and the former fortition (-v- > -p-) is well attested, if
uncommon, in Gir. (Skt. darśayitvā > Gir. dassayitpā, “having seen” ), Sh. (Skt. svāmika >
Sh. spāminka, “master”) and P. (Skt. prāvaraṇa > P. pāpurana, “cloak”; Skt. hāvayati > P.
hāpeti; “to sacrifice”, see Woolner 1928, ¶39-6). So in this case the P. –tabba ending
seems to relate both to the western and the north/north-western dialects.
9) in RE 14 E, we find: (Skt.) saṃkṣaya > (Gir.) sacchāya, > (K.) ṣaṃkheye, > (Sh.) samkhay-,
(M.) saṃkhay-, (Eṛṛ.) saṃkhāyāyā, but Pāli goes back, not to the western, but to the
northern version sankhaya (“loss” as a noun, or “having omitted” as a gerund).
10) Where the northern form and the eastern form are different, Gir. goes with the northern form,
as does Pāli: RE 6 F: (Skt.) ātyayika, > (K. Dh.) atiyāyike, > (Gir.) āccāyika, > (Sh.) acayika,
(M.) acayike, Pāli (accāyika; “urgent”).
11) While Gir. often preserves the -sṭ- conjunct (e. g. RE 4 C: (Skt.) anusiṣṭaye > (Gir.)
anusasṭiyā “instruction”; RE 4 G: (Skt.) śreṣṭha > (Gir.) sesṭe, “best” and RE 6 D: (Skt.)
sthita > (Gir.) sṭitā, hapax legomenon, “standing”, “being”, it is not preserved in Pāli
(anusatthiyā, seṭṭha, ṭhita). In this case P. is the similar to K. Dh. and Eṛṛ.
(anusatthiye/anusatthiyā, seṭhe) and to Sh. and M. which, however, retain the –r-
(sreṭhe/sreṭhaṃ).
12) Many consonantal clusters are retained in Gir. ( kr, tr, pr, vy) but few of these are kept in
Pāli which often adopts the northern/eastern from: e. g.
a) RE 1 G: (Skt.) prāṇa > (Gir.) prāṇa > (K. J. Eṛṛ.) pāna, (P.) pāṇa (“living being”);
b) RE 1 F: (Skt.) priya > (Gir.) priya, > (K. J. Eṛṛ.) piya, > (Sh.) pria, >(M.) priya, (P.) piya
(“dear”);
c) RE 5 I: (Skt) trayodaśa > (Gir.) traidasa, > (K. Dh. Eṛṛ.) tedasa, (Sh.) todaśa, > (M.)
treḍaśa, > (P.) terasa/telasa ( “thirteen”). The Pāli word shows a –d- > -r- change which is
characteristic of Gāndhārī, the north-western dialect (Brough 1962, ¶43-b);
d) RE 10 C: (Skt.) –tika, > (Gir.) –trika, > (K. Dh. J.) –tika, > (Sh. M.) –trika, (P.) –tika
(“threeforld”). In the Aśokan edicts this word occurs in the compound pāra-trika (MW:
“advantageous in another world” or “with a view to the next world” (Woolner 1924, s. v.).
Strangely, the compound does not exist in Pāli which uses the northern/eastern form or -tika,
not the western;
e) RE 4 F: (Skt.) putrāḥ > (Gir.) putrā > (K. Dh.) putā, > (Sh. M.) putra, > (P.) puttā (“sons”)
f) RE 6 B: (Skt.) atikrāntam, > (Gir) atikrātaṃ, > (Sh. M.) atikrataṃ, > (K. Dh. J) atikaṃtaṃ,
(P.) atikkanta (“passed”). Pāli follows the northern/eastern form. The same situation exists
for the common verb parā√ kram;
g) with the common Skt. word brāhmaṇa, P. is closest to the Sh. and M. versions (bramaṇa),
although P. is clearly a back-formation to Skt. Girnār uses the br- form of brāhmāṇa in two
out of 7 cases. 6
6 The different reflexes of the word brāhmaṇa in the Aśokan edicts, whether in compound or as a single
Girnār: bāmhaṇa (3D), brāmhaṇa (4A), bramhaṇa (4C), bāmhaṇa (8E), bamhaṇa (9G), bāmhaṇa (11C),
bāmhaṇā (13G),
Kālsī: baṃbhānaṃ (3D), baṃbhānaṃ (4A), baṃbhana (4C), baṃbhanibbhesu (5K), baṃbhanānaṃ (8E),
bābhanānaṃ (9G),
Shāhbāzgarhī: bramaṇa (3D), bramaṇanaṃ (4A), bramaṇa (4C), bramaṇibheṣu (5K), bramaṇanaṃ (8E),
baṃbhanānaṃ (8E), [baṃbha]nanaṃ (9G), baṃbhanāna[ṃ] (11C), bā[bha]nā (13G), [bā]bhane (13J).
Delhi-Topra: bābhanesu (7th Pillar Edict Z), bābhana (7th PE HH)
We note the following phonological changes. All locales except Sh. and M. (and five out of seven times in
Gir.), change brāh- > bā/ba. This is a normal MIA conjunct simplification at the beginning of a word
(Pischel ¶268). brāhmaṇa is derived from the Vedic root √ bṛh or √ bṛṃh (to grow great or strong, to
increase). The –h- is usually not retained, but is often progressively assimilated to the initial consonant
with the formation of the bilabial aspirated stop –bh-, so brāh- > bābh-. Sometimes, as in the northwest Sh.
and M. it is simply dropped (-h- > Ø). In the latter case, the second syllable – ma- is preserved (brāhma- >
bamaṇa or bramaṇa); however with the change to bābh-, the second syllable loses its initial m-: brāhma- >
bābha-, sometimes with an anusvāra after the –ā-: brāhma- > bāṃbha-, which is presumably a retention
from the Vedic root √ bṛṃh or a metathesis from the second syllable. Peculiar to Girnār is the reversal of –
hm- > -mh-, brāhma- > b(r)āmha-, and there is one case in K 13J where (brāhmaṇa > baṃhmane)
h) A case where Pāli preserves the conjunct vy-, from Gir. is found in RE 5 J: (Skt.) vyāpṛta >
(Gir.) vyāpatā, >(K. Dh.) viyāpaṭā, > (Sh.) vapaṭa, > (M.) vapuṭa > (P.) vyāvaṭa (“busy”,
“engaged”). Note the change of –p- to –v- and –t- > -ṭ-. The preservation of –vy- does not
appear to be invariably the case as in RE 8 B, Gir. has magavyā (< Skt. mṛgavya, “hunting”),
while P. has migavā, unless the latter is derived from alt. Skt. word mṛgayā (also meaning
“hunting”) with a change–y- > -v-. Other examples where the vy- is not retained is Skt.
vyāḍa > P. vāḷa (“snake”) and Skt. vrata > P. vata (“religious observance”).
13) In RE 4 F (Gir.) uses the term prapotrā to mean “great-grandson” (< Skt. prapautra), with
potrā meaning “grandson”, and putrā, “son”; the other REs use paṇatika for “great-grandson”
(< Skt. pranaptṛ) and natāle/nataro to mean grandsons (< Skt. naptāraḥ, pl. of naptṛ). Pali
has the word paputta, but it means “grandson”, not “great-grandson” and has no equivalent
to Girnār’s potra. For “great-grandson” Pāil has panattu (not in PED, but in Buddhadatta,
232) which, once again, corresponds with K. (panātikyā), Dh. (panatti), Sh. (pranatika) M.
(paṇatika) Eṛṛ. (panātikā), but not Gir. Pāli also has the word nattar for “grandson” which
corresponds to all the REs but Gir.
What we are witnessing in the above collection is that Pāli is not as closely related to the
Aśokan western dialect as has commonly been believed. In fact, in all but a few cases above, P.
is much more closely related to the northern (K.) or northwestern (Sh. M.) dialects.
where–hm- is retained but the preceding vowel nasalized. The normal Prakrit reflex of brāhmaṇa is
bambhaṇa (Pischel ¶250 in ArdhaMāgadhī and Jaina-Māhārāṣṭrī; for other dialects the reflex is bamhaṇa,
in Māgadhī and Śaurasenī, per Pischel ¶250, 287). The question then arises, since the “normal” Prakrit
form for brāhmaṇa is either ba(m)bhaṇa (prevalent at Kālsī in the north central area, Dhauli in the east,
Eṛṛaguḍi in the south and Delhi in the centre) or bamhaṇa (used at Girnār in the west) or b(r)amaṇa (in
the two north west sites, Sh. and M.): why did the authors of the Pāli recension chose the Sanskrit version,
using a form which is closest to the Sh. and M. reflex? This seems to corroborate the composite nature of
the Pāli language, formed by monks in “conscious interference with the natural development of this
What about the other resemblances between Pāli and the western dialect? Let us re-examine
Lamotte’s list of similarities (1988, 565), which are based mainly on morphological rather than
lexemic or phonological considerations:
1) P. has a nom. sing. in –o for a-stems, as does Gir. But so does Shāhbāzgaṛhī. M. has
nom. sing in –e (Brough ¶76) which might also account for such P. forms traditionally
xc), although whether this was long of short –a is impossible to tell, since vowel length
was not noted in Karoṣṭhī script (Salomon 1998, 75). K. has abl. in -ā.
3) P. has a locative of a-stems in –e, -asmiṃ and –amhi; Gir. has loc. in –amhi and –e; Sh.
has loc. in –aspi and –e. As we have seen above, -m- and –p- are closely related.
4) P. has the accusative plural in –e as does Girnār. Sh. has it in –ani. Neuter plurals are
5) P. has instr. of –an stem form as raññā and gen. as rañño; in Gir. the instr. is rāñā and
the gen. is rāño. Sh. has instr. as raña and gen. as raño.
6) In Gir. and P. the demonstrative ayaṃ functions as nom. sing. in both masculine and
feminine. Sh. has ayaṃ for masc. and aya for fem.
7) Gir. and P. has the 3rd. person sing. ending in –e and -ey(y)a and the middle opt. in –tha.
Sh. apparently has no dedicated opt. form except for the verb √ as (siya and siyati
compared to P. siyā for 3rd pers. sing). Kālsī has –eyā as 3rd opt.
8) P. and Gir. both have the development –u- > -a- < Skt. guru. But, as noted above, on two
out of three occasions, Gir. retains guru while it is Sh. and K. which have garu.
9) The change of Skt. khalu > kho is noted by Lamotte as an isogloss between P. and Gir.,
between P. and Gir. However this is also common to K. and Sh. Dh. and J.
11) The retention of -ṇ- in the stems is common to both P. and Gir., but this is also the case
in Sh. and M. (e. g. gaṇa- in RE 3 E or samaṇa in RE 4 C or caraṇa in RE 4 D).
12) Changing –bh- > -h- in instrumental plural (common to all REs) and in the verb hoti
differentiation of the sibilants, the latter two following Skt., and the former
indiscriminately (Hultzsch, lxxvi). The eastern dialects also collapse the sibilants,
although the palatal s > c in the root √sak (Separate RE 2 G: Dh.).
14) the shortening of long nasalized vowels esp. in the gen. pl. is done in Gir. and in P. But it
is also prevalent in Kālsī (Hultzsch lxxiv), Sh. and the other REs (e. g. RE 4 A: Skt.
bhūtānāṃ > Gir., Eṛṛ. bhūtānaṃ, K. bhutānaṃ, Dh. bhūtānaṃ, Sh. M. bhutanaṃ ).
In Summary, considering all 14 points of convergence outlined by Lamotte, there is only one
or two points (masc. accus. pl. in –e of –a stems and the levelling of the sibilants to the
dental s) which are peculiar to P. and Gir., with the others being common to the northern
and north western dialects, or sometimes to all. Plus there are many points of dissimilarity
between the two dialects, leading one to believe, with Norman that “there is probably no
connection between Pāli and the Girnār dialect of the Aśokan inscriptions” (2006B, 128). Yet
there does appear to be a strong connection with the north and northwest.
Conclusions
1) Pāli is a composite language, younger and more phonologically evolved than the Aśokan
Prakrits. It has many and varied correspondences to the dialects, but the strongest by
far is to Sh. and K., northwestern and northern.
2) We know that Pāli was not written down until the first century B. C. in Śri Lanka (Bechert
1992, 45); however writing existed in India from at least Aśokan times, and probably
earlier. Scholars are generally agreed that the earliest writing in India was in the Karoṣṭhī,
Aramaic script and was extant from the fourth century or possibly the fifth century
(Salomon 1998, 46). Karoṣṭhī was older than Brāhmī and quite possibly its precursor and
model (ibid, 54).
3) It would not be an unreasonable hypothesis therefore to suggest that the first Buddhist
teachings were written down in Karoṣṭhī in the north west to which Buddhism spread
from its earliest times through the existing trade routes; Buddhism had a special appeal
to merchants, for the new philosophy validated and encouraged their profession
(Thapar 2000, 926).
4) Assuming the Pāli teachings had a Karosṭḥī source or influence, this would account for
the significant number of northwestern influences in the Pāli “dialect”. Although the only
way to prove this statistically is to translate the entire Aśokan corpus into Pāli, my
intuition, from the little preliminary work done above, is that much more of the lexemic
and phonological inventory of Sh. M. and K. will be present in Pāli than will the forms
from Gir. This in itself does not prove anything more than the fact that Pāli is not closest
to Gir. as is commonly opined, but closer to the northern and northwestern dialects. It
does not of course prove derivation. The morphological correspondences between P.
and Gir and P. and Sh./K./M. are equally strong.
5) In this spirit, Appendix 1 shows a comparison of all forms of (randomly chosen) RE 4
which supports this hypothesis of a north/north-western origin of Pāli, with 43% of the
words in K. Sh. and M. being closest to Pāli, while only 19% are closest to Gir. This short
compilation treats data on a unitary basis and does not weight for important common
words like Pāli brāhmaṇa which is closest to the NW form - bramaṇa, but its change to
brāhmaṇa in Pāli may also be a later back-Sanskritization as Norman suggests (2006B,
134).
6) The Pāli- Shāhbāzgaṛhī/Mansehrā/Kālsī connection also helps to explain some of the
strange phonological anomalies in the edicts, like the use of the common word ātman.
The eastern and Pāli form is atta, but Gir. has atpā and M. has atva. Sh., although it
preserve elsewhere the –m- > –v change (e. g. asmin > aspi in loc. sing.) has atta. Yet
AMg., also an eastern dialect, preserves the form appa. This leads to the conclusion
that atpa/atva is a northwestern form and that the AMg. and Māgadhī form appa was
borrowed from there, while the other eastern dialects (Dh. J.) chose the atta form. A
derivation from the east to the northwest is precluded by the Dh. J. preservation of the
atta form (since –t- never changes to –p-). Alternately we are simply witnessing the
confusion noted by Pischel (¶277) of atman > atpa > appa or by metathesis atman >
interpreted to derive form Skt. pratyekabuddha (“a private Buddha”) but probably
comes from pratyayabuddha (“a Buddha awakened by specific conditions; i. e. not by a
Buddha’s teachings”); 7 or the ambiguity between viraja (“free from passion”), virata
(“stopped”, “ceased”) in corresponding versions of the Sabhiyasutta (in Pāli Sutta Nipāta
and BHS Mahāvastu) which goes back to an underlying word viraẏa; 8 there are dozens
of these ambiguities in the Pāli writings, some of which have been isolated in Levman
2009. As the most advanced of the Aśokan dialects, Shāhbāzgaṛhī – which is already
extensively softening consonants, dropping them or replacing them with glides - could
well be the source of, or at least a significant influence on these linguistic anomalies.
Abbreviations
7 pratyayabuddha (Skt.) > paccayabuddha > *pacceyabuddha > (palatalization –a- > -e- in the presence
of [-ANT] cons. –cc-), P. paccekabuddha. In this case the –y- form encountered by the translator was
interpreted as an intervocalic glide replacing a consonant (often represented as a –y- with a dot over it, –
ẏ-, as per Pischel ¶187), common in the north-west Prakrits, and the missing consonant (thought to be –
k-) was mistakenly replaced.
8 See Norman 1990B, 151. The consonant -ẏ- represents a weakly articulated intervocalic glide which
replaced consonants in intervocalic position, first in Gāndhārī and then in the other Prakrits.
BHSD = Buddhist Hybrid Sanskrit Dictionary
Br.= Brahmagiri
Dh. = Dhauli (Rock Edict)
Dhp = Dhammapada
P. = Pāli
Pān. = Pānguḍāriyāṃ
PDhp = Patna Dhammapada
PE = Pillar Edicts (capital letter following refers to Hultzsch’s section designations, 1969)
PED = Pali Text Society Pali-English Dictionary
PN = Proper Name
RE = Rock Edicts (capital letter following refers to Hultzsch’s section designations, 1969)
Sah. = Sahasrām
Pāli closest to K. Dh. Eṛṛ all Gir. Dh. Sh. K. Dh. Eṛṛ Gir. Dh. Sh.
Eṛṛ
Pāli closest to Gir. Sh. K. Dh. Sh. Eṛṛ. for brāhmaṇa – Sh. M.
9 Generally, I considered consonantal similarity to be more important than vowel, esp. at an ending.
Therefore, for example, I consider K. piyadassine to be closer to P. piyadassino, than Gir. priyadassino.
When the only difference between forms is in a long or short vowel (as in bahūni/bahuni) I treat them all
as equal.
Eṛṛ. asaṃpaṭipati se aja devānaṃpiyasa
Pāli closest to none Gir. K. Dh. M. K. Dh. Eṛṛ hatthi = K. Dh. Eṛṛ
Eṛṛ. dassanāni = Gir.
Pāli closest to Gir. K. Dh. M. Eṛṛ. Gir. Sh. M. Gir. K. (based on Gir. Sh. M.
the verse form)
10 divya is the verse form, dibba, the prose form, per PED.
Eṛṛ. dasayitpā janasa ādise bahuhi vasasatehi
Pāli closest to K. Dh. Eṛṛ none Gir. Dh. Eṛṛ. Gir. Sh. all
K. baṃbhanasamanānaṃ saṃpaṭipati
M. bamaṇaśramaṇana saṃpaṭipati
P. brāhmaṇasamanānaṃ saṃpaṭipatti
Pāli closest to K. Sh. M. K. Dh. Eṛṛ. vuḍha Dh. Sh. Err. Sh.
Pāli closest to Gir. Sh. M. all Gir. Gir. Dh. Sh. Eṛṛ
Pāli closest to Sh. for c’eva: Gir. K. K. Dh. Eṛṛ. K. Dh. Eṛṛ
Dh. , for yeva, M.
Eṛṛ.
Pāli closest Gir. M. Gir. M. all K. Dh. Eṛṛ all but Dh.
to
Pāli closest to Gir. K. Dh. Eṛṛ. Gir. Gir. Sh. M. all but Sh.
Pāli closest to all all but Sh. for Gir. K. Eṛṛ. Dh. Sh. none
imasa/imamh;
none for sā
Pāli closest to all but Sh. Gir. K. Dh. Sh. Eṛṛ. all
Pāli closest K. Dh. Eṛṛ. K. Dh. Eṛṛ. Gir. Sh. Gir. Dh. Sh. Dh.
to
Gir. lekhāpitam
K. lekhitā
Dh. lekhite
Sh. nipesitam
M. likhapite
Eṛṛ. likhite
Skt. lekhitaṃ/likhāpitaṃ
P. lekhitaṃ/likhāpitaṃ
Gir. 40 19%
K. 40 19%
Dh. 42 20%
Sh. 31 15%
M. 19 9%
Eṛṛ. 39 18%
Total 211
11 percentages total more than 100% because of multiple answers (Pāli is similar to more than one rock
edict form). “All” and “None” were omitted from count.
Works Cited
Bechert, Heinz. 1980. Allgemeine Bemerkungen zum Thema "Die Sprache der ältesten
buddhistischen Überlieferung." In Die Sprache der ältesten buddhistischen Überlieferung,
Abhandlungen der Akademie der Wissenschaften in Göttingen, Philologisch-Historische Klasse,
Dritte Folge, Nr. 117. English version in Buddhist Studies Review, 8 1-2 (1991): 24-34.
Bechert, Heinz. 1992. The Writing Down of the Tripitaka in Pāli. Wiener Zeitschrift für die Kunde
Südasiens 36 (1992): 45-53.
Bloch, Jules. 1950. Les Inscriptions d'Aśoka. Paris: Société d'Édition "Les Belles Lettres".
Brough, John. 1962. The Gāndhārī Dharmapada. London: Oxford University Press.
Geiger, Wilhelm. 2005. A Pāli Grammar. translated into English by Batakrishna Ghosh, revised
and edited by K. R. Norman. Oxford: The Pali Text Society.
Hinüber, Oskar von 1994. The Oldest Literary Language of Buddhism. Selected Papers on Pāli
Studies. Oxford: The Pali Text Society: pp. 177-194.
Hinüber, Oskar von 2001. Das Ältere Mittelindisch im Überblick. Wien: Verlag er
Österreichischen Akademie der Wissenschaften.
Lamotte, Étienne. 1958, 1988. History of Indian Buddhism from the origins to the Śaka Era,
translated from the French by Sara Webb-Boin. Louvain-la-Neuve: Université Catholique de
Louvain Institut Orientaliste.
Levman, Bryan. 2009. Vedhamissakena: Perils of the Transmission of the Buddhadhamma.
Canadian Journal of Buddhist Studies 5: 21-38.
Lüders, Heinrich. 1954. Beobachtungen über die Sprache des Buddhistischen Urkanons. Berlin:
Akademie - Verlag.
Norman, K. R. 1990A. Four etymologies from the Sabhiya-sutta. In Collected Papers II. Oxford:
Pali Text Society: 148-161.
Norman, K. R. 1990B. The dialects in which the Buddha preached.In Collected Papers II, .
Oxford, : Pali Text Society.
Norman, K. R. 2006A. The Group of Discourses (Sutta-Nipāta). Oxford: The Pali Text Society.
Pischel, R. 1965. Comparative Grammar of the Prākṛit Languages. Delhi: Motilal Banarsidass.
Salomon, Richard. 1998. Indian Epigraphy, A Guide to the Study of Inscriptions in Sanskrit,
Prakrit, and the other Indo-Aryan Languages. New Delhi: Mushiram Manoharlal Publishers Pvt.
Ltd.
Thapar, Romila. 2000. The Householder and the Renouncer in the Brahmanical and Buddhist
Traditions. In Cultural Pasts Essays in Early Indian History. New Delhi: Oxford University Press:
914-45.
Woolner, Alfred C. 1924. Asoka Text and Glossary, Part I. Introduction, Text. Part II. Glossary.
Lahore: Oxford University Press (Printed at the Baptist Mission Press).
Woolner, Alfred C. 1928, 1996. Introduction to Prakrit. Delhi: Motilal Banarsidass Publishers.