The following phonological changes distinguish typical MIA languages from their OIA ancestors: • The replacement of vocalic liquids
ṛ and
ḷ by
a,
i or
u • The OIA diphthongs
ai and
au became the monophthongs
e and
o which were long in open syllables and short in closed syllables. • Long vowels become short in overweight and later pre/post-tonic heavy syllables. • The three sibilants of OIA are reduced to one, either
ś (Magadhi) or
s (elsewhere). • OIA clusters either became geminates through assimilation (deletion if the output would violate phonotactics) or were split by vowel epenthesis. • Initially, intervocalic aspirated stops spirantised. Later, all other intervocalic stops were deleted, weakened, or voiced. • Dentals (and sometimes retroflexes) are palatalised if directly preceding /j/. • Most final consonants delete except in
sandhi junctions. Final
m became
ṃ instead, which was preserved. Note that not all of these changes happened in all MIA languages. Archaisms persisted in northwestern Ashokan Prakrits like the retention of all 3 OIA sibilants, for example, retentions that would remain in the later Dardic languages. The following morphological changes distinguish typical MIA languages from their OIA ancestors: • The dual number in nominal declensions was lost. • Consonantal stems were thematicised. • The
i-/u- and
ī-/ū- declensions were merged into one
ī-/ū- declension. • The dative was eliminated and the genitive took on its former functions. • Many different case-endings could be used for one verbal paradigm. • The middle voice eventually disappeared. •
mahyaṃ and
tuhyaṃ became used for genitives and
me and
te for instrumentals. • New verbal forms based on the present stem coexisted with fossilised forms from OIA. • Active endings replaced passive endings for the passive voice. A Middle Indo-Aryan innovation are the
serial verb constructions that have evolved into
complex predicates in modern north
Indian languages such as
Hindi and
Bengali. For example, भाग जा (bhāg jā) 'go run' means run away, पका ले (pakā le) 'take cook' means to cook for oneself, and पका दे (pakā de) 'give cook' means to cook for someone. The second verb restricts the meaning of the main verb or adds a shade of meaning to it. Subsequently, the second verb was
grammaticalised further into what is known as a
light verb, mainly used to convey
lexical aspect distinctions for the main verb. The innovation is based on Sanskrit atmanepadi (fruit of the action accrues to the doer) and parasmaipadi verbs (fruit of the action accrues to some other than the doer). For example, पका दे (pakā de) 'give cook' has the result of the action (cooked food) going to someone else, and पका ले (pakā le) 'take cook' to the one who is doing the cooking. ==Attested languages==