[K.R. Norman, in his study of the variations between the OIA (Old Indo-Aryan: Vedic and Classical Sanskrit) and MIA (Middle Indo-Aryan: Prakrits), finds MIA dialects contain many forms] which are clearly of IA, or even IE, origin, but have no attested Skt equivalent, e.g. suffixes not, or only rarely, found in Skt, or those words which show a different grade of root from that found in Skt, but can be shown not to be MIA innovations, because the formation could only have evolved in a pre-MIA phonetic form, or because a direct equivalent is found in an IE language other than Skt... the forms in that category go back to 'lost' OIA dialects... I know of no attempt to make a complete and comprehensive collection of the evidence for this interesting category of forms in MIA, and it remains scattered through the pages of Indological writings. I believe that, until such a collection is made, the amount of material available will be underestimated.
British philologist (1925–2020)