In the previous post we met the two great changes: nine consonants dropping intervocalically, and five aspirates collapsing into h. Those two rules explain most of what looks unfamiliar about a Prakrit word. But between vowels a few other consonants also shift — not by dropping, but by becoming something else. This post rounds up those changes.
The theme is the same: the middle of a Prakrit word wants to be softer, easier, more open. When a consonant cannot vanish, it will often melt a little.
A note on the trailing “a”. In Devanāgarī every consonant sign carries a built-in a — क is ka, so the cluster त्य is written tya and its Prakrit result च्च is written cca. In a change such as त्य → च्च (tya → cca) or ट → ड (ṭa → ḍa), only the consonants actually change; the a is there just so the letters can be pronounced.
Retroflex softenings
Sanskrit has a pair of retroflex stops, टṭ and डḍ, made by curling the tongue back to the roof of the mouth. Prakrit keeps them in initial position and in clusters, but between vowels it voices them.
- टṭ → डḍ (unaspirated retroflex softens to its voiced form)
- ठṭh → ढḍh (aspirated retroflex softens the same way)
| Sanskrit | Prakrit | Meaning |
|---|---|---|
| नटnaṭa | नडnaḍa | actor |
| भटbhaṭa | भडbhaḍa | soldier |
| घटghaṭa | घडghaḍa | pot |
| मठmaṭha | मढmaḍha | monastery |
| पठतिpaṭhati | पढइpaḍhaï | reads |
| कुठारkuṭhāra | कुढारkuḍhāra | axe |
The rule is quiet but pervasive. Any Sanskrit word with an intervocalic ṭ or ṭh will show up in Prakrit with ḍ or ḍh.
ḍ takes one further step
And डḍ itself — the voiced retroflex — softens further, all the way to लl.
| Sanskrit | Prakrit | Meaning |
|---|---|---|
| गरुडgaruḍa | गरुलgarula | Garuḍa (the eagle) |
| तडागtaḍāga | तलाअtalāa | pond |
| क्रीडतिkrīḍati | कीलइkīlaï | he plays |
So the chain runs टṭ → डḍ and डḍ → लl. A Sanskrit word with intervocalic टṭ often ends up with डḍ in Prakrit; a Sanskrit word with intervocalic डḍ often ends up with लl. One important thing to keep in mind: once a rule has transformed a syllable, a second rule targeting that new syllable is not then applied to it. Each change is a resting point, not a stepping stone.
Every n becomes ṇ
This one is almost universal. Between vowels, Sanskrit dental नn becomes retroflex णṇ.
| Sanskrit | Prakrit | Meaning |
|---|---|---|
| कनकkanaka | कणयkaṇaya | gold |
| वचनvacana | वयणvayaṇa | speech |
| नयनnayana | णयणṇayaṇa | eye |
| मानयतिmānayati | माणइmāṇaï | he honours |
Notice that in नयनnayana even the word-initial n is often written ṇ. Hemacandra makes this optional at the beginning of a word but obligatory in the middle. The result is that Prakrit texts are peppered with ṇ where Sanskrit had n, and this is one of the easiest visual cues that you are looking at Prakrit rather than Sanskrit.
The sibilant merger
Sanskrit has three sibilants: palatal शś, retroflex षṣ, and dental सs. Prakrit merges all three into सs.
| Sanskrit | Prakrit | Meaning |
|---|---|---|
| शब्दśabda | सद्दsadda | word |
| कुशkuśa | कुसkusa | Kuśa grass |
| दशdaśa | दसdasa | ten |
| शुद्धśuddha | सुद्धsuddha | pure |
| षष्टिṣaṣṭi | सट्ठिsaṭṭhi | sixty |
| शेषśeṣa | सेसsesa | remainder |
One sibilant to rule them all. This means the same Prakrit सs can trace back to any of three different Sanskrit sounds — so when you reconstruct backwards, you have to guess. सेसsesa could in principle come from a Sanskrit word with ś or with ṣ; context tells you it is शेषśeṣa.
b joins the softening club
Just as intervocalic पp becomes वv (which we saw last time), intervocalic बb also becomes वv.
| Sanskrit | Prakrit | Meaning |
|---|---|---|
| अलाबुalābu | अलावुalāvu | gourd |
| शबलśabala | सवलsavala | dappled |
This shift is less common than the पp → वv one, simply because Sanskrit has fewer intervocalic b's to begin with. But when they appear, they too go the v way.
What about m?
Sanskrit m between vowels usually stays m — this consonant is stable. But in a small handful of specific words it gets nasalised: the m dissolves into a nasal quality on the surrounding vowel. So yamunā becomes ja:um̐ṇā (with a nasal echo on the ü), cāmuṇḍā becomes cāum̐ṇḍā. These are named exceptions rather than a general rule, and they are worth flagging only because they look strange when you first meet them.
Putting it together
The middle of a Prakrit word tends to be one of three things:
- A vowel where a Sanskrit consonant used to be (the great lenition rule).
- An हh where a Sanskrit aspirate used to be.
- A softer version of a Sanskrit consonant — डḍ for टṭ, लl for डḍ, णṇ for नn, सs for शś or षṣ, वv for पp or बb.
Once you internalise those three tendencies, most single-consonant Prakrit words become readable. What remains is the trickier terrain of clusters — two or three consonants stacked together, which Prakrit finds uncomfortable and resolves in inventive ways. That is what the next posts are about.