The Prakrit and Pali languages developed from
Old Indo-Aryan (OIA, e.g.
Vedic Sanskrit, ca. 600 BCE) by a set of regular
phonological transformations. This section summarizes the changes occurring between Vedic Sanskrit (ca. 600 BCE), Early Middle-Indo-Aryan (MIA) in
Pali or
Ashokan Prakrit (ca. 280 BCE), the dramatic regional
Prakrits like
Maharashtri Prakrit (ca. 200 AD), and finally the late Prakrit or
Apabhramsha stage (ca. 900 AD).
Conservative features lost in Vedic Prakrit and Pali languages partially-preserve some
conservative features of
Proto-Indo-Aryan (PIA) lost in Vedic Sanskrit. Specifically: • PIA
*kṣ, *gẓʰ merge to Vedic
kṣ, but remain distinguished later as
kh-, jh- initially and
-kkh-, -jjh- medially. Compare PIA
Hákṣi > Sanskrit
akṣi > Prakrit
akkhi "
eye" with PIA
gẓʰáranam > Vedic
kṣaraṇam, later Sanskrit
jharaṇam > Prakrit
jharaṇa "falling". • The distinction between
PIE *r and
*l survived in the "
l-dialect" of Indo-Aryan, which went on to form the basis of Classical Sanskrit and Central MIA. The Northwestern Vedic dialect, perhaps under areal influence of
Iranian, underwent an
l > r shift alongside the same shift in Iranian. Thus, Vedic is part of the "
r-dialect" of Indo-Aryan, though /l/ continues to exist as a relatively rarer phoneme in Vedic. Due to the
prestige of the Vedic dialect, the
r-dialect forms were often accepted into the dialect of Classical Sanskrit and later into Central MIA. More rarely,
l is encountered in the later language for PIE
*r. Compare PIE
*rikh₂-é-ti > Sanskrit
likhati > Prakrit
lihadi, lihaï. • The
l-dialect (and cases of /l/ in Vedic) then possibly underwent
Fortunatov's law, wherein pre-Sanskrit dentals
*t,
*tʰ,
*d,
*dʱ,
*s, and
*n underwent
cerebralization after PIA
*l, after which the
*l was deleted. — PIA
*ȷ́l̥tʰáram > Sanskrit
jaṭhára "stomach, womb". • Proponents of Fortunatov's law will separate this from the Middle Indo-Aryan rule in which a dental is cerebralized by a preceding rhotic. That later rule only affects plosives (and even then, it operates sporadically), while Fortunatov's law reliably operated in Old Indo-Aryan. It shifts
s > ṣ (Proto-Indo-Iranian
bʰā̆ls- > Sanskrit
bhāṣ- "to speak"), while the distinction between
s ~ ṣ ~ ś had already been lost by Middle Indo-Aryan. • In Vedic and Pali dialects, PIA
*ẓḍ >
ḷ (ळ) /ɭ/ and
*ẓḍʰ >
ḷh (ळ्ह) /ɭʱ/. Elsewhere (and regularly in Classical Sanskrit), PIA
*ẓḍ >
ḍ (ड) /ɖ/ and
*ẓḍʰ >
ḍh (ढ) /ɖʱ/.
Early changes common to Dardic The following changes are common to Middle Indo-Aryan (MIA) and
Dardic. • The
unpredictable pitch accent of Vedic was lost, resulting in the
mora-timed isochrony of Pali. The pitch accent is sometimes used to explain some irregular consonant doubling (Vedic
jitáḥ > Prakrit
jitto "won") or differences between
Marathi and neighboring languages, indicating that the Vedic accent may have converted to a stress accent (as with
Ancient Greek) and persisted in the
Maharashtri region. However, this topic is debated and in the case of the irregular consonant doubling there may be alternative explanations. • Before a consonant, Vedic
e /ɐj/ (ए) and
o /ɐw/ (ओ)
monophthongize to /eː/ and /oː/, respectively. In the same position, Vedic
ai /ɑːj/ (ऐ) > /ɑj/ and
au /ɑːw/ (औ) > /ɑw/. This pronunciation is already used in Classical Sanskrit, distinguishing it from Vedic. • Loss of word-final consonants in a few ways: • Sanskrit word-final plosives (only /k ʈ t̪ p/) are lost with
compensatory lengthening of the preceding vowel. • Based on the initial consonant of the following word, Vedic word-final
visarga /h/ has the
allophones , [ɸ], [ɾ], [x], [s], [ɕ], [ʂ], or [w]. Already in Classical Sanskrit, [ɸ] and [x] are not found. Later, Vedic -अः /ɐh/ becomes MIA -ओ /oː/ everywhere. Remaining word-final /h/ was entirely lost without a trace. • Final nasals are lost with compensatory nasalization of the preceding vowel, indicated by the
anusvara. Mishra and Bloch argue that word-final anusvara was realized as nasalization, but elsewhere anusvara was still pronounced as a nasal consonant. • MIA
monophthongization of /ɑj ɐjɐ ɐji ɐjoː ɐvi/ > /eː/. Similarly, /ɑw ɐʋɐ/ > /oː/, though occasionally /ɐʋɐ/ fails to reduce. — Sanskrit
avarodhanam /ɐʋɐɾoːd̪ʱɐn̪ɐm/ > Ashokan
olodhanaṃ /oːloːd̪ʱɐn̪ɐ̃/ "
harem" • A
dental spontaneously
cerebralizes to a
retroflex stop in the environment of a rhotic. This rule originated in the
east, and later to the
north and
northwest; it was less common in the
west. Some scholars like
Wackernagel argue that the original cases (or borrowings from eastern dialects) with a retroflex stop in the environment of a rhotic, like
prati- >
paṭi- and
mēḍhra "ram, penis" (already retroflex in Proto-Indo-Aryan
*Hmáyẓḍʰram) influence later analogical formation. Due to later dialectal mixture and adaptation to Sanskrit, this sound change ends up occurring fairly sporadically in MIA. For example, as early as Pali we see Sanskrit
ardhaḥ /ɐɾd̪ʱɐh/ > Pali
addho /ɐd̪ːʱoː/ or
aḍḍho /ɐɖːʱoː/ "half". Generally, cerebralization did not happen across a perceived
morpheme boundary. Hence, Sanskrit
nirdhanaḥ /n̪iɾd̪ʱɐn̪ɐh/ (morphologically
nir- "without, -less" +
dhana- "wealth") > Prakrit
ṇiddhaṇo /nid̪ːʱɐnoː/ "not wealthy". Many cerebralized words were old enough to be borrowed back into Classical Sanskrit, like
paṭh- "to read" (from older
pṛth- "to spread") with a specialized meaning. See also Fortunatov's Law, a similar rule mentioned above. • '
Loss of ṛ'''
is common to Dardic, Pali, and Prakrit. The exact sound change depends on the context and region/dialect. In the Central dialect, initial ṛ-
> ri
and medial -ṛ
> i
(but cases of ṛ
> a
and ṛ
> u
—especially after labial—are occasionally found). Elsewhere (especially in Pali, Western, and Southern dialects), ṛ
> a
is more common. — Sanskrit ṛṇa
> Prakrit riṇa
"debt" and Sanskrit kṛta
> Prakrit kida
, kia'' "done".
Sound changes common to Pali and Prakrit After the split of Dardic languages, these changes are common to Pali and Prakrit. These changes may be also grouped together as the "MIA assimilation rules", and are the most important rules from this period. Regarding the assimilations of Old Indo-Aryan consonant conjuncts, the
Jayadhavalā (ca. ninth century AD) writes
Dīsaṁti doṇṇi vaṇṇā saṁjuttā aha va tiṇṇi cattāri / Tāṇaṁ duvvala-lōvaṁ kāūṇa kamō pajuttavvō "When two, or three or four, consonants appear in combination,
elide the weakest one, and continue the process." Here, "weakest" refers to sounds of higher
sonority, and "elide" refers to either true elision/loss or
total assimilation of the weaker sound to the stronger sound. Specifically, the sonority scale of Prakrit is (weakest)
h Prakrit
dasa /d̪ɐsɐ/ "ten". • After a non-nasal dental or retroflex stop, /m/ and /ʋ/ assimilate to /p/ if the stop is unvoiced or /b/ is the stop is voiced. This stage is attested at
Girnar — Sanskrit
catvāraḥ /t͡ʃɐt̪ʋɑːɾɐh/ > Girnar Ashokan
catpāro /t͡ʃɐt̪pɑːɾoː/. • Vedic
jñ /dʑɲ/,
ny /n̪j/, and
ṇy /ɳj/ > Early MIA
ññ /ɲː/. •
Epenthetic b /b/ arises in -
mr- /mɾ/ >
-mbr- /mbɾ/ and
-ml- /ml/ >
-mbl- /mbl/. — Sanskrit
āmraḥ /ɑːmɾɐh/ > */ɑːmbɾɐh/ > Prakrit
aṃbo /ɐmboː/ "
mango". • A succeeding /j/
palatalizes a dental stop and a succeeding sibilant palatalizes /t̪/ and /p/. This tendency is represented scarcely in some "corruptions" in Sanskrit, like
jyotiḥ /d͡ʒjoːt̪ih/ "light" either Prakrit
tikkho /t̪ikːʰoː/ or
tiṇho /t̪inɦoː/ "sharp". In the first, /ɳ/ was deleted from the right and by a later sound change /kʂ/ > /kːʰ/. In the second, /k/ was deleted from the left and by a later sound change /ʂɳ/ > /ɳɦ/. In either case, the /iː/ shortens as it occurs before a consonant cluster. •
kṣ /kʂ/ >
kh- /kʰ/ initially or
-kkh- /kːʰ/ elsewhere. In the Central and Eastern regions, this outcome would naturally occur from the sound changes below. It is worth mentioning because in other regions, the special sequence
kṣ /kʂ/ >
ch- /t͡ʃʰ/ initially and
-cch- /t͡ːʃʰ/ elsewhere. There is some dialectal borrowing already seen in Pali (which usually attests the
-(k)kh- variant but sometimes has
-(c)ch-) and regional Ashokan Prakrit. • In a cluster with a
plosive or
affricate (always) or when
followed by a nasal consonant (usually, but not always),
s /s/ weakens to
h /ɦ/. Alternatively when followed by a nasal consonant, the
s can be retained (especially medially, where by the cluster simplification rule below the cluster becomes
-ss-) or
anaptyxis can break the cluster. • When
h /ɦ/ (either from older sibilant via the above rule or otherwise) precedes a consonant, the consonant and /ɦ/ metathesize. — Sanskrit
cihnam /t͡ʃiɦn̪ɐm/ > Pali
cinhaṃ /t͡ʃin̪ɦɐ̃/ "sign". • At the start of a word, the sequence of an
obstruent (i.e. non-sonorant) +
h /ɦ/ simply results in an aspirated plosive or affricate. Elsewhere, the
h /ɦ/ undergoes
regressive assimilation by transforming into the aspirated equivalent of the plosive or affricate. The last rule is already known in Sanskrit sandhi, where, for example,
tad +
hi →
taddhi, rather than
*tadhi so as to preserve the
syllable weight of the first syllable. — Sanskrit
vatsaḥ ~ vatso /ʋɐt̪soː/ > */ʋɐt͡ʃsoː/ (palatalization) > */ʋɐt͡ʃɦoː/ (sibilant weakening) > /ʋɐt͡ːʃoː/ (sandhi of
-c- +
-h- →
-cch-) > Pali and Prakrit
vaccho /ʋɐt͡ːʃoː/ "calf". •
Cluster simplification: The main rule being referred to in the above Jayadhavalā quote. All consonant clusters are simplified to achieve the MIA phonotactic situation where syllables are restricted to at most a CVC
structure (where the
coda C is further restricted to either the
anusvara ṃ, a nasal/plosive
homorganic to the following syllable onset C, or a sonorant if the following syllable onset is
h /ɦ/). Word-initially, only the strongest sound in a cluster survives — Sanskrit
grāmaḥ /gɾɑːmɐh/ > Ashokan
gāmo /gɑːmoː/ "village". Medially, the weaker sound totally assimilates to the stronger sound, and sequences like
-CʰC- or
-CʰCʰ- are immediately repaired to
-CCʰ- — Sanskrit
vyāghraḥ /ʋjɑːgʰɾɐh/ > */ʋjɐgʰɾoː/ (prior sound changes) > /ʋɐgːʰoː/ (initial /ʋj/ > /ʋ/, medial /gʰɾ/ > /gːʰ/, not */gʰg/) > Pali and Prakrit
vaggho /ʋɐgːʰoː/ "
tiger". • The treatment of a sonorant +
h /ɦ/ is complicated. Generally, aspirated sonorants are not phonemic in Pali or Prakrit, meaning
yh /jɦ/,
rh /ɾɦ/,
lh /lɦ/,
vh /ʋɦ/,
ñh /ɲɦ/,
ṇh /ɳɦ/,
nh /n̪ɦ/, and
mh /mɦ/ are all treated as a sequence/cluster of two consonants. However, these sequences are treated specially among the variety of consonant clusters, particularly because they
are permitted word-medially. Word-initially, the result is varied. In Pali, epenthetic
-a- breaks the sequence — Sanskrit
hradaḥ /ɦɾɐd̪ɐh/ > */ɾɦɐd̪oː/ (prior sound changes) > Pali
rahado /ɾɐɦɐd̪oː/ "lake". In Prakrit, this sequence is either permitted (especially in Maharashtri Prakrit), epenthetic
-a- appears to break the sequence, or the
h /ɦ/ is dropped. — Sanskrit
snānam /sn̪ɑːn̪ɐm/ > Prakrit
ṇhāṇa /n(ɐ)ɦɑːnɐ/,
siṇāṇa /sinɑːnɐ/, and
saṇāṇa /sɐnɑːnɐ/ "bathing". • The sequence
-sr- /sɾ/ can sometimes yield
-ṃs- /ns/ rather than expected
-ss- /sː/. — Sanskrit
aśru /ɐɕɾu/ > Prakrit
assu /ɐsːu/
, aṃsu /ɐnsu/ "
tear". • The sequence
-mh- /mɦ/ can sometimes fortify to
-ṃbh- /mbʱ/. • In general,
anaptyxis, rather than assimilation, can be applied to break a heterogeneous cluster. This is most common in cases of a stop followed by sonorant (except
r), as in
-tn-, -dm-,
-kl-,
-gl-, etc. — Sanskrit
ratna > Pali
ratana, Sanskrit
padma > Pali
paduma, or Sanskrit
klēśa > Prakrit
kilēsa "grief".
Changes after the split of Pali and Prakrit The following changes are only seen in Prakrit and not in Pali (other Pali-specific changes do also occur beyond this point). •
yh /jɦ/ and
vh /ʋɦ/ become
jh- /d͡ʒʱ/ and
bh- /bʱ/ initially and
-jjh- /d͡ːʒʱ/ and
-bbh- /bːʱ/ medially. In Pali, only /ʋɦ/ changes. — Sanskrit
guhyaḥ /guɦjɐh/ > /gujɦoː/ (prior sound changes) > Pali
guyho /gujɦoː/, but develops further into Prakrit
gujjho /gud͡ːʒʱoː/. • Relatedly, initial
y /j/ fortifies to
j /d͡ʒ/ and medial
yy /jː/ >
jj /d͡ːʒ/.
— Sanskrit
yaḥ /jɐh/ > Pali
yo /joː/, Prakrit
jo /d͡ʒoː/. • In Pali, geminate
-vv- /ʋː/ >
-bb- /bː/, but this never occurred in Prakrit generally. However, in the Central and Eastern region, initial /ʋ/ > /b/ and medial geminate /ʋː/ > /bː/ has occurred before New Indo-Aryan. It can be argued that this fortification occurs earlier alongside the fortification of /j/ > /d͡ʒ/, and orthographic व् /ʋ/ beyond this point is merely conservative. • Cases of
-ēy- /eːj ~ ejː/ and
-ī̆y- /iːj ~ ijː/ are re-analyzed as having a geminate glide and undergo the above rule as well. — Sanskrit
kālēyam /kɑːleːjɐm/ > Prakrit
kālēyaṃ /kɑːleːɐ̃/
, kālijjaṃ ~ *kālĕjjaṃ /kɑːlid͡ːʒɐ̃ ~ kɑːled͡ːʒɐ̃/. • In Pali, Sanskrit
jñ /dʑɲ/ becomes
ñ- /ɲ/ initially and
-ññ- /ɲː/ medially. In Prakrit, the result is usually
j- /d͡ʒ/ initially and
-jj- /d͡ːʒ/ medially. Sometimes,
ñ- /ɲ/ initially and
-ññ- /ɲː/ medially are found too. — Sanskrit jñāna > Prakrit jāna; Sanskrit rājñī > Prakrit rāṇī. • As noted before, the reflex of Sanskrit
ṛ is different in Pali, Prakrit, and Dardic (e.g. initial
ṛ > Prakrit
ri- always, but Pali and Dardic
a-, i-, u-). Also, the reflex of Sanskrit clusters involving a sibilant and sonorant is unstable between Pali and Prakrit.
Changes up to Dramatic Prakrits These changes occur after Pali and Early Prakrit, but before the development of the dramatic regional
Prakrits like
Maharashtri Prakrit and
Shauraseni Prakrit (ca. 200 AD). • Before a consonant,
ṅ,
ñ,
ṇ,
n,
m, and the
anusvara ṃ are in
complementary distribution in MIA. In Sanskrit, each different nasal consonant is typically written out. In later languages, all pre-consonant nasals are written as the anusvara
ṃ. • Merging of nasals
ṇ, n >
ṇ (n̪ ɳ > n), represented as a retroflex nasal. Whether the actual place of articulation of this sound was truly retroflex or was dental (and just orthographically represented as a retroflex nasal) is debated. • Lenition of intervocalic stops over time, through various attested stages. First, all single intervocalic unvoiced stops become voiced. Then, non-retroflex stops spirantize (one possibility is g, gʱ, dʒ, d, dʱ, b, bʱ > ɣ, ɣʱ, ʑ, ð, ðʱ, β, βʱ / V_V). Per
Chatterji, this stage is represented by vacillation between writing a voiced stop, semivowel, or nothing. The retroflex voiced stops
ḍ, ḍh likely become flaps intervocalically, but this distinction is not represented orthographically. Finally, aspirated spirants
debuccalize (ɣʱ, ðʱ, βʱ > ɦ), the spirant β > ʋ (romanized as
v), and remaining spirants ɣ, ʑ, ð are lost, leaving the surrounding two vowels in
hiatus. — Sanskrit
kathanam /kɐt̪ʰɐn̪ɐm/ > Prakrit
kahaṇaṃ /kɐɦɐnɐ̃/ "saying". • Between two
ā̆ vowels, hiatus is usually resolved by what
Hemachandra, in his grammar of Prakrit, calls a “lightly pronounced
y-sound” (
laghuprayatnatarayakāraśrutiḥ)
. As far as orthography/romanization is concerned, this results in the optional inclusion of epenthetic
-y- or less likely
-v- between the
ā̆ vowels. This orthographic choice should not be confused with the older genuine /j/ phoneme. Similarly, after a front vowel, euphonic/orthographic
-y-appears. Elsewhere, hiatus is fully tolerated. After
ā̆, the
diaeresis is often used in romanization (e.g.
aï, aü) to differentiate this sound from the older overlong vowels. • Sanskrit
śoka- /ɕoːkɐ/ > Pali/Ashokan
soka- /soːkɐ/ > Early Dramatic Prakrit
soga- /soːgɐ ~ soːɣɐ/ > Prakrit
sōa- /soːɐ/ "sorrow". • Sanskrit
caturtha- /tɕɐt̪uɾt̪ʰɐ/ > Pali/Ashokan
catuttha /t͡ʃɐt̪ut̪ːʰɐ/ > Early or
Shauraseni Dramatic Prakrit
caduttha- /t͡ʃɐd̪ut̪ːʰɐ ~ t͡ʃɐðut̪ːʰɐ/ > Prakrit
caüttha- /t͡ʃɐut̪ːʰɐ/ "fourth". • Occasionally, intervocalic
-d- /d̪/ became
-r- /ɾ/, as in the numbers from 11 to 18. • Sanskrit
ekādaśa /eː.kaː.d̪ɐ.ɕɐ/ > Prakrit
egārasa /eː.ɡaː.ɾɐ.sɐ ~ eː.ɣaː.ɾɐ.sɐ/. • Lenition of intervocalic
y /j/
, similarly to the above change. The optional inclusion of epenthetic
-y- sometimes makes this confusing, but at this point /j/ is no longer phonemic in Prakrit; it is merely an epenthetic hiatus-filler. — Sanskrit
nayanam /n̪ɐjɐn̪ɐm/ > Prakrit
ṇa(y)aṇaṃ /nɐɐnɐ̃/ "
eye". • Lenition of intervocalic
v /ʋ/ between
ā̆ and a high vowel. — Sanskrit
praviṣṭa- /pɾɐʋiʂʈɐ/ > Prakrit
païṭṭha- /pɐiʈːʰɐ/ "entered", but Sanskrit
nava /n̪ɐʋɐ/ > Prakrit
ṇava /nɐʋɐ/ "nine" with retention of
-v-. • Occasionally, the sequences
aï and
aü contracted early on in Prakrit to
ē̆ and ō̆. This is a separate change than the later coalescence of vowels in hiatus. — Sanskrit
sthavira- /st̪ʰɐʋiɾɐ/ > Earlier Prakrit
ṭhavira- /ʈʰɐʋiɾɐ/ > *
ṭhaïra- /ʈʰɐiɾɐ/ > Later Prakrit
ṭhēra- /ʈʰeːɾɐ/ "old". • Prakrit
ḍ, ḷ, l, and
r often alternate with each other, particularly in words loaned from non Indo-Aryan sources. — PIA *
swaẓḍaśa > Sanskrit
ṣoḍaśa > Prakrit
solasa /soːlɐsɐ/ "sixteen". • Also worth noting here is the addition of pleonastic suffixes to older nominals and roots. This becomes more prevalent by late MIA and early
New Indo-Aryan period. The consensus, implied by the name, is that these innovative suffixes have little semantic purpose and mainly serve to distinguish homophones (created by the sweeping sound changes between Sanskrit and Prakrit). They are applied after nominal and verb stems, before inflecting suffixes. Some are recognizable as the reflexes of Old Indo-Aryan diminutive suffixes. The most important suffixes are feminine
-iā- () Pali/Prakrit
gāmo /gɑːmoː/ > Central Apa. /ˈgɑː.ʋ̃u/ "village". • Prakrit final long vowels shorten and change in quality if necessary to merge with the short vowels: /ɑː iː uː eː oː/ > /ɐ i u i u/. This tendency is known since at least Ashokan times, where originally-long final vowels are frequently shortened and short vowels are confused with the long form (in a manner of orthographic conservatism)—we find Ashokan
tada /t̪ɐd̪ɐ/ for Sanskrit
tadā /t̪ɐd̪ɑː/ "then", Ashokan
Ambika /ɐmbikɐ/ for Sanskrit
Ambikā /ɐmbikɑː/, etc. • Intervocalic
-v- /ʋ/ is lost after a high vowel. • Long
ū is shortened before another vowel — (Sanskrit
kūpa- /kuːpɐ/ >) Prakrit
kūva- /kuːʋɐ/ > Apa. /ˈku.ɐ.u/ "
well". • Development of a
Latin-like
positional stress system. Stress falls on the penultimate syllable if it is
heavy, failing which it falls on the antepenultimate syllable if it is heavy, failing which it falls on the fourth syllable from the end. This system retroactively came to characterize
Classical Sanskrit, but it can be considered a MIA development that was only fully completed around the Apabhramsha stage. == Grammar ==