I. Phonology. Semivowkls.
43
b. r appears in place of phonetic d (= z, Ilr. i)' as final of stems in -is and -us, before endings which begin with bk-; e. g. havir-bhis and vdpur-bhis. This substitution is due to the influence of Sandhi, where is, us would become ir, ur. r also takes the place of d in ira- beside ida- 'refreshing draught'. In urubjd- 'wide open' r apparently takes the place of dental d (= *ud-ubjd-), perhaps under the influence of the numerous compounds beginning with uru- 'wide', e. g. uru-jri- 'wide-striding' ^
c. Metathesis of r takes place when ar would be followed by s or h + consonant. Under these conditions it appears to be phonetic, being due to the Svarabhakti after r being twice as great before h and sibilants as before other consonants 3; when a vowel followed the sibilant or h (e. g. darsatd-)^, the Svarabhakti was pronounced; but if a consonant followed, rS took the place of ar + double Svarabhakti. This metathesis appears in forms of drs- 'see' and srj- 'send forth': drdstum (AV.), drastr- (AV.) 'one who sees'; sdm-srastr- 'one who engages in battle', 2. sing. aor. sras (== *sra/e)i; also va.prds-ti- 'side-horse', beside /«V/-z/- and prsti- 'rib'; moreover in brahmdn- 'priest', brdhman- 'devotion', beside barhis- 'sacrificial litter' (from brh- or barh- 'make big'); perhaps also drahydi 'strongly' {drh- 'be firm'). The same metathesis occurs, being, however, very rare and fluctuating, before ks-: thus tuvi-mraksd- 'injuring greatly', mraksa-kftvan- 'rubbing to pieces' (from mrj- 'wipe', or mrs- 'stroke'), but tdrksya-, N.*.
52. The semivowel I. — The liquid sound / is the semivowel corre- sponding to the vowel / (which however occurs only in some half dozen perfect and aorist forms of the root kip- 'be adapted'). It is pronounced at the present day in India as an interdental; but it must have had a post- dental sound at the time of the Pratisakhyas 7, by which it is described as being pronounced in the same position as the dentals*.
a. It represents IE. /9 and, in a few instances, IE. r. It is rarer in Vedic than in any cognate language except old Iranian (where it does not occur at all)". It is much rarer than r, which is seven times as frequent". The gradual increase of /, chiefly at the cost of r, but partly also owing to the appearance of new words, is unmistakable. Thus in the tenth Book of the RV. appear the verbs mhic- and labh-, and the nouns loman-, lohitd-, but in the earlier books only mruc- 'sink', rabh- 'seize', roman- 'hair', rohitd- 'red'; similarly dasanguld- 'length of ten fingers', hlddaka- and hlddika-vant- 'refreshing', beside sv-anguri- 'fair-fingered', hradd- 'pond'. Moreover, while in the oldest parts of the RV. / occurs" in a few words only, it is eight times as common in the latest parts. Again, in the AV. it is seven times as common as in the RV.'3; thus for rap- 'chatter', rikh- 'scratch', a-srird- 'ugly', appear in the AV. lap-, likh- (also VS.), aslTld-. The various texts
sraj- 'wreath', ra- does not stand for ar, Wackernagel I, 190 d, note.
7 Cp. PiSCHEL, BB. 3, 264. An indication that it was not a cerebral is the fact that Is never occurs (while rs is common).
8 See 44.
9 Which, however, is largely represented by r also : cp. 51 a.
10 Loc. cit.
11 See WmTNEY, JAOS. il, p. XLff.
12 See Arnold, 'L in the Rigveda', in Festgruss an Rudolf von Roth, 1893, p. 145 — 148; Historical Vedic Grammar, JAOS. 18, 2, p. 258f.; Vedic Metre p. 37, 3.
13 Cp. Wackernagel i, 191 c.
1 Cp. above p. 35, note 3.
2 On a supposed parasitic r in chardis- 'fence', yajatra- 'adorable', vibhrtra- 'to be borne hither and thither', see Wackernagel I, 189, note 7.
3 According to the APr. I. 1 01.
4 Cp. above 21 a.
5 Cp. V. Negelein, Zur Sprachgeschichte des Veda 83, note 7. ^
6 On the interchange of ra and ar in raj- 'colour', rajata- 'silvery', raj- 'shine', and arjuna- 'bright' ; bhraj- 'shine', and bhdrgas- 'brilliance'; bhratr- 'brother', and bhartr- 'hus- band'; see Wackernagel i, 190 e, note (end). In vraja- 'fold', vrati.- 'ordinance'.
