Notes :It is absolutely impossible to unite this root with PGn *kar- 'to stir, churn', as Burrow and Emeneau do, since the inlaut consonants are different; however, this actual root does have good SDR parallels. PK *-c- is the relict of an old causative suffix.
Notes :The suffixes do not coincide in PK and PGn, but the root is obviously the same in both languages. In PK *kaṛay- > *kṛay-, with *-y- > *-ʔ- before the suffix.
Notes :Bisyllabic root. Logically we would expect PK *kräm- > Kui krāmbu. However, bisyllabic nominal roots seem in general to behave unlike verbal roots. This may be due to accent problems; thus, if we accept the hypothesis that verbal bisyllabic roots had a barythonese and nominal ones - an oxytonese, no irregularities should be postulated (since only unaccented *-e- > PK *-ä-).
Notes :An ultrashort epenthetical -e- seems to be present already on the PGn level (cf. Gondi vs. Kui forms). However, since there is no development *-e- > *-ä- in PK, the root should be reconstructed as monosyllabic.
Notes :Inlaut consonant unclear. Gondi points to *-d_-, while PK points to either a simple *r or an alveolar *r_. The inlaut consonant was probably alveolar, but either *r_v > Gondi *Rv or *d_v > PK *rv.
