Page:Vedic Grammar.djvu/23

 L I. PHONOLOGY. CONTRACTION. HIATUS. SVARABHAKTI. 13 3. in the simple word titau- 'sieve' (probably from tams- 'shake'), by a loss of s, due most likely to borrowing from an Iranian dialect (where medial s would have become h, which then disappeared). b. 1. Though not written, hiatus is common elsewhere also in the Samhitãs. The evidence of metre shows that y and v must often be pronounced as i and u, and that a long vowel or a diphthong has frequently the value of two vowels. When the long vowel or diphthong is the result of contraction, the two original vowels must often be restored, within a word as well as in Sandhi. Thus pánti 'they protecť, may have to be read as på-anti (— pá-anti)³, anjan 'they anointed' as á-anjan; jyéstha- 'mightiest' as jyá-iștha- (– jyá-iṣṭha- from jya- 'be mighty'); áicchas as á-icchas 'thou didst wish'; aurmos as a-urnos thou didst open'4. 2. Hiatus is further produced by distraction of long vowels 5 which, as the metre shows, are in the Rgveda often to be pronounced as two short vowels. This distraction was doubtless originally due partly to a slurred accentuation which practically divided a syllable into two halves, and partly to the resolution of etymological contraction. From such instances distraction spread to long vowels in which it was not historically justified. It appears most often in ā, especially in the gen. pl. in -ām, also in the abl. sing. in --āt, the nom. acc. pl. in -as, -āsas of a-stems, in the acc. sing. in -am of such words as abjám 'born in the water'; and in many individual words 6. Distraction is further found in the diphthongs of words in which it is not etymologically justified; as in the genitives vés 'of a bird', gós 'of a cow', in tredhá 'threefold', néty leader', réknas- 'property', śréni- 'row'; and in other words7. 21. Svarabhakti.- When a consonant is in conjunction with or a nasal, a very short vocalic sound tends to be developed between them, and the evidence of metre shows that a vowel must often be pronounced between them. It is the general view of the Prātiśākhyas that when an precedes another consonant a vowel is sounded after it; according to some of them this also takes place after / or even after any voiced consonant. They call it svarabhakti or 'vowel-part', which they describe as equal to ¹/8, ¹/4, or ¹/2 mora in length and generally as equivalent to a or e (probably = e) in sound. I a. The metre of the RV. shows that an additional syllable is frequently required in words in which either precedes or follows another consonant; e. g. darśatá- 'worthy to be seen' (quadrisyllabic); indra-¹0, name of a god (very often trisyllabic); prá 'forth' (dissyllabic) ". ¹ Cp. WACKERNAGEL I, 37 b, note. appearance of slurred accentuation: WACKER- NAGEL I, 47. 2 See OLDENBERG, Prolegomena 434ff.: 'Hiatus und Contraction'; ARNOLD, Vedic Metre, chapter IV, p.70 ff. (Sandhi), chapter v, 'p. 81 ff. (Syllabic Restoration). 8 RPr. VI. 13 f., VPr. Iv. 16; TPr. XXI. 15f.; APr. I. 101 ff. 3 As a rule, one vowel (including e and o)| is shortened before another: see OLDEN- -BERG, op. cit., 465 ff.; 447 ff. 9 The vowel which has to be restored in the gen. loc. du. termination -tros, which must always be read as a dissyllable, is not to be explained as Svarabhakti, since -taros is the original ending. 4 WACKERNAGEL 1, 46 b. 5 See OLDENBERG, op. cit., 163 ff. (Vocale mit zweisilbiger Geltung). 10 Cp. OLDENBERG, ZDMG. 60, 711-745 (Die Messung von índra, rudrá u. a.). 11 There seem to be a few instances of a 6 See WACKERNAGEL I, 44. This is a very old phenomenon, as it is found in the Avesta Svarabhakti vowel being actually written: in the gen. pl. and in other forms: OLDEN | tarásantī, beside tras- 'tremble'; the secon- dary derivative śvaitárīm, beside śvitrá-(AV.) BERG 181; WACKERNAGEL I, p. 50. 7 WACKERNAGEL I, 46. This distraction 'white'; purusa- and pirusa- 'man', probably of diphthongs is also pre-Vedic, parallels for purşa- (WACKERNAGEL I, 51, cp. 52). being found in the Avesta. Its use gradually The initial vowel of uloká-, which is commoner decreases in the RV. and is lacking in the than loká- 'world', has not yet been satis- later Samhitãs, doubtless owing to the dis-factorily explained; cp. op. cit. 1, 52 d.