________________
VARIANT ENDINGS -U,-AÜ AND -A IN THE APABHRAṀSA VERSES 29
57 In the same way Alsdorf derived the plur. ending -a from -aya. Though this plural ending will not be discussed separately, all the questions raised in connection with the long ending of the singular equally apply to the one for the plural.
58
While the uneven padas of the regular Doha consist of a gana with six morae which is followed by one of four, and tails off with three short syllables (i.e. 6 + 4 + 3), Alsdorf (Apa-Studien, p. 40 ff.) has drawn attention to several instances in which the sixth short syllable of the first and the first short syllable of the second gana have coalesced into one long vowel. In nine out of the ten instances identified this involves the long ending -a. Of these nine eight concern the suffic -ḍā, the ninth instance is hua in 351: bhalla huă ju māriā. An exception to this pattern is appaṇem in 416d: aha appaṇem na bhanti. Alsdorf apparently overlooked an instance of a similar phenomenon, but this time between the second and the third gana, namely in 422, la: ekku kaïahā vi naā vahi.
Another metrical irregularity is found in the third gana of the uneven pādas, which, as already indicated, consists of three short syllables. As such it formed the ideal slot for the extended ending, as is shown by e.g. 341, 2: aggalau, 389: mumḍiaŭ, and 401, 1: lalasaü. In several instances, however, one may find in this positon the long ending, which means that the three morae are resolved in an iambic rhythm: 330, 2: vāria, 343, 1: siala, 370, 4: veggala, 379, 2: pārakkaḍā, 379, 2: amhahā taṇā, 409: hārāvia, 414, 3: sampesia, 418, 1: vinṇāsia, 423, 4: karāvia, 438, 3: vāria, and 439, 1: ghaṇā. For instances of yet another resolution of the three short syllables, namely by -, see Alsdorf, Apa-Studien, p. 43.
It should be noted, however, that these "irregularities" do not affect the total number of moric instances of the verse. In this respect the metre is respected. For the same reason -a is never found in the Doha at hemistich-final position, which should always be a short vowel.
59
See Alsdorf's discussion (Apa-Studien, p. 55) of pādas 3 and 5 of 422, 6:
to vi drehi jahim kahi vi Rāhī daddha-naana nehem palutta.
The metre of the verse, Radḍā, requires that the pādas end with two short syllables and that the two lines should rhyme. This leads Alsdorf to emend Rāhī, i.e. Radha or Radhika, into Rahiya, and, consequently, paluṭṭā into paluṭṭiya (paluṭṭita). This emendation was based on the argument that the long endings would go back to disyllabic ones. An alternative emendation would, of course, be Rāhā.
60
The same ending -a is also found in Old Gujarati (e.g. hatha sing.). This hatha is opposed to the adjective ghanau (nom.) or ghana (obl.); hatha (plur.) to ghaṇā (nom. and obl.); see Dave, A Study of the Gujarati Language, p. 21.
61 For the ending -o, see Prakṛtapaimgala, Part II, pp. 105 ff.
62 See in this connection also dayālā (-ū < -u) in Neminathacatuṣpadikā 10a, referred to above in note 43.
63
In this respect the case of fem. -iya is easier to understand. In another respect, however, it is different, for it does not preserve the element of rhyme, -iya standing in concord with words ending in -i as well as with those ending in -a.
64 See Edgerton, Buddhist Hybrid Sanskrit Grammar, § 22.21 ff.
65 See Turner, "Gujarātī Phonology" JRAS 3 (1921), pp. 329-365, id. 4, pp. 505-544, and "The e and o Vowels in Gujarātī" Orientalia ii, 2 (1925), pp. 337-347.
66 A similar phenomenon can be seen in Prakrit, e.g. in Jaina Mahārāṣṭrī. For the re-introduction of the anusvära in particular in the North Indian manuscripts of a Prakrit text, see Tieken, Hala's Sattasai, pp. 168-169.