Notes :This deictic base had the following declineable forms in PGn: *aw-ant_[-u] (whence Gondi *ōn-R, Konda vānr_u with a metathese, etc.) 'that man', plural *aw-ar[-u] 'those men'; fem. *a-d-i (PK *ādi by analogy with masculine forms), plural *a-v-i. Also reconstructed for the PGn level is *a-c- 'that many, that much' (PG, PPM, PK); *a-tal 'that side' (Konda, PK, possibly Gondi); *a-mb-/*a-b- 'there' (Konda, PPM, PK). An enormous number of derivatives has been formed independently in different languages (which is characteristic of all deictic bases).
Notes :The form is rather local (in one Kuwi dialect and Konda) and with few external parallels (Telugu aḍḍa, which may have been the source for borrowing).
Notes :The Konda form unexpectedly changed conjugation type (aṭ- can only go back to *aḍ-C-). Vowel length in PK is irregular; as a matter of fact, Kuwi_Su has short a-, so this variant is probably more archaic.
Notes :The root was probably not used by itself in PGn. The form in Konda goes back to an old causative (hence the devoicing: aṭ- < *aḍ-p-), and there are derivatives with the suffix -mb- (in PK) and with the suffixes -ŋ- and -s- (in Konda).
Notes :The root was heavily used with posessive prefixes. The most frequent form is *yay- 'my mother'. In some dialects this form became generalized, with the following shift of meaning: 'my mother' > 'mother in general', 'mother' > 'woman'. A new form, based on the 1st person pl. pronoun (Gondi māy-, Kuwi māya) resulted from the loss of initial y- and the secondary merger of the simple and the prefixed forms.
Notes :A common PGn root with no external parallels. Vocalic correspondences are identical to the ones found in PGn *vajr̯- 'alone' q. v., resulting in the provisional reconstruction of an inlaut j-cluster.
Notes :Although the resonant here is most certainly a plain dental -l-, it becomes retroflex in Konda (cf. especially such derivative forms as aṛpa-). This is not quite clear; maybe it is a general feature of *-l- before the causative marker. Unfortunately, this is the only example of such a cluster (in a few other cases we have no BB forms with -ṛ-, which are diagnostic).