^{page 499 note 1} Copies of the original recordings are deposited at the Universities of Leiden and Utreeht. The material investigated in 1–23 of this article is covered by the following references to the catalogue compiled by Mr. Staal (A catalogue of Veda recitation and Indian music; recorded in South India, April—July 1957): VI (1–5, 11); XIV (2, 6); XVII (3). In the later sections (26 onwards), additional material has been taken into account (Nambudiri Taittirīyaka): VI (10, 12); X (1, 2); XIV (1–3); XVII (1). In addition (for 29): X (3); XIX (1); XXI (16).

The publication referred to is Nambudiri Veda recitation, to be published shortly in the series ‘Disputationes Rheno-Trajectinae’.

^{page 499 note 2} ‘An analysis of Ṛgvedic recitation’, BSOAS, XXII, 1, 1959, 86–94, which had already appeared before the present analysis was undertaken.Google Scholar

^{page 499 note 3} The majority chanted a minor third below mid-tone (with occasional drops to a major third below) and a minor to a small major second above mid-tone. One reciter extended the compass to a fifth throughout.

^{page 500 note 1} Where a short vowel is followed by a nasal/(liquid ? see p. 504, n.3) plus consonant/semivowel, it is accounted a long vowel, but not where a consonant group follows the nasal, etc.

^{page 500 note 2} The combination -rh- does not count as CC after a long vowel, though it has the effect of lengthening the syllable (e.g. bārh-is fractionally short of 3 units, see 19), as also -hn-through the insertion of svarabhakti. Such incidental features will not be indicated in the signs.

The equation: CC = hiatus, could not be argued from the Ṛkprātiśākhya on the basis of time, where, in I, 33–4, drāghīyasī (svarabhaktiḥ) sārdhamātrā/ itare ca ‘the longer (svarabhakti) is half a mātrā, as also the others (i.e. consonants)‘. By II, 3–4, svarāntaram tu vivrttih / sā vā svarabhaktikālā ‘ the interval between the vowels is called vivṚtti (hiatus) and it has optionally the length of a svarabhakti’. If Uvata is correct, this should be interpreted to mean a quarter mātrā (the shorter svarabhakti length) between two short vowels, half a mātrā between a long and a short vowel, and three-quarters of a mātrā between two long vowels. Nothing, however, is stated about hiatus in the chapter on metre (XVIII), which might more readily solve the present question. In the previous article on Ṛgveda recitation, pp. 80–94, it was noted (p. 89, 6) that high-to-low-tone anudāttas were particularly clear in contexts involving a long vowel followed by two plosives. In fact, both informants (B and C) illustrated the tendency to dwell on a long vowel followed by a consonant group. (The references in the article referred to which were given for the Ṛkpr.related to verse numbers. The edition used for the present is that of M. D. Sāstri: The Rgveda Prātiākhya with the commentary of Uvaṭa, Vol. II, Allahabad, 1931. The previous references of III, 2–3, 11–13, 18, and VI, 15, would then read III, 4–5, 19–22, 32, and VI, 55, respectively.)

^{page 501 note 1} Where a lower tone might be expected on a vowel introduced by (C)Ch-, the alteration did not always take place if the consonant was palatal, e.g.: cch-, or Chy-.

^{page 501 note 2} For the sake of brevity, any tonal pattern (Nambudiri Ṛgvedic) which begins with a hightone will be referred to as high-tone. The two passages referred to in this paragraph are found in: Staal, XIV (6) and VI (2).

^{page 502 note 1} Where the anudātta consists of a long vowel plus visarga, four rapid swings between low- and mid-tone, the last reaching high-tone, are used instead (4 units). Only one example occurred in the passages recited.

^{page 502 note 2} The available contexts do not relate what happens in the case of a sarvānudātta (or pracaya initial in the line, see BSOAS, XXII, 1, 1959, p. 87, n. 2) occurring before a nasal plus consonant/semivowel. Presumably, more swings are heard.Google Scholar

^{page 504 note 1} The following paragraph deals with syllables in terms of metric length (while the signs in 3–16 attempt to indicate tonal patterns on the vowels as well as indicating metrio length). It would be pointless to attempt to compare these time divisions with what is said in Ṛkprātiákhya, I, on the relative lengths of vowels and consonants (whieh is also a reason for not taking what is said concerning hiatus too literally). The portions relative to metre (Ṛkpr., XVIII, 37–44, in particular) are of little help in the present context.

^{page 504 note 2} At least, the sign has been tapered! The impression is that the hiatus between short vowels is longer (1½ units) than that between long and short/long vowels (1 unit), but this feature is not represented in the illustrations at the end of the article (examples 1–3) where the signs are evenly spaced. In this respect, Nambudiri practice is more closely allied to what is said in the Tribhāṣyaratna on Taittirīyaprātiákhya, XXII, 13, than with Uvata's remarks (see p. 500, n. 2).

Account has not been taken of kampa, but examples of ekakampashow two intense swings from low-to-mid-tone (4 units). The dvikampa is a prolongation of sign 9 (12 units) with a swing touching low-tone in the middle. There was no example of trikampa (where CC follows) which, presumably, would be interminable.

^{page 504 note 3} Normally, the combination of a short vowel with a nasal plus consonant funotions as a long vowel (cf. p. 499, n. 3), while a short svarita vowel with -r- plus consonant functions as any other CVC-C syllable before anudātta (i.e. 5, second sign). Nevertheless, an example occurred where the syllable salm- in … was treated as before anudātta. No other examples of -I-in this position were found. Possibly there is a palatal element in the -l- in such positions. Where this occurs (e.g. in a context such as yVt-C), there is invariably an abrupt glide upwards in pitch towards implosion. This feature may have caused the present anomaly by making the syllable behave as (sign 10) instead of (second sign 5). Since the evidence of -r- in such contexts is also uncertain, the problem must remain open.

^{page 505 note 1} cf. also the preceding footnote. Mr. Staal has pointed out the chief features, namely, the pronunciation of -t- as -l- before a voiceless plosive (other than -t- itself) and finally in the line, as well as the difference between the Nambudiri visarga final in the line (akin to the ach-laut) and the -aḥ^a type current further north. The latter is an important, though well-known, feature. In connexion with ‘breathings’, it is worth emphasizing that the Nambudiris are meticulous in awarding Ch- its proper value.

^{page 506 note 1} It may be presumed, therefore, that this difference runs through the long udātta and svarita as well. The passages in Ex. 4 are the only examples available. The two most important Nambudiri Vedic centres are at Trichur and Tirunāvāya.

^{page 507 note 1} The Nambudiris treat the final syllable of the prior member of a compound occurring after the non-Vedic itì (the parigrahas in the vikrtis) as if it were a final syllable in the line (as also a compound occurring in pada recitation).

^{page 507 note 2} Indebtedness to Dr. A. A. Bake is acknowledged for this recording (as well as for those in Exx. 10 and 12) which was made in 1956 in Kathmandu at the time of the coronation. The recording given in Ex. 12 was also made from Brahmans who had come to Nepal to officiate at the ceremony. The reciter of Ex. 10 was a native of Nepal.

^{page 507 note 3} A Brahman must put considerable effort into reciting. The result of such concentration upon what he himself is doing renders him tolerably insensitive to the performances of his fellow reciters, and the requirements of the Sikṣā are not always evident: cf. ‘sweetness, clarity of the syllables, divisions of the words, proper accenting, precision, and aptitude for tempo—such are the qualities that pertain to recitation’ (Pāṇinīyasikṣā, 33; Sikṣāsamuecaya, Benares, 1893, p. 381).

^{page 508 note 1} An example of a performance (poor ?) will illustrate the point. One informant had particular difficulty with the anudātta. In the first verse of ā nό bhqdrāḥ…, where anudātta occurs twelve times, only three were below mid-tone. Noticeably high was and, on the long vowels, failed to drop below mid-tone, although (before hiatus) and (beginning of line) did strike the low-tone.

On the other hand, such recitations as sādhāraṇa mātrā, śālā mātrā, and cantādikkal are very measured performances. For descriptions of these, as well as of koṭṭ and ghoṣam, the reader is once again referred to Mr. Staal, op. cit. The Nambudiri Taittirīyaka has found numerous ways of deriving dignified entertainment from the fruits of his sacred duty.

^{page 510 note 1} Most sikṣas include some details on accompanying gestures, but none so comprehensively as Yājnavalkyaśikṣā.Some forty verses are devoted to a detailed description of hand movements, besides enjoining their use in verses such as (47): yathā dhanuṣy āvitate sare kṣipte punar guṇaḥ/ svasthānaṃ pratipadyeta tadvad dhastagalaḥ svaraḥ ‘as the bow-string, once the arrow stretched on the bow has been released, regains its rightful place, so does the accent in relation to the hand’.

^{page 511 note 1} Verses 22, 23, 51, 56, and 52 respectively (Śikṣāsamuccaya, pp. 3, 4, 7).

^{page 512 note 1} Pāṇinīyaśikṣā, verse 52 (Śikṣāsamuccaya, p. 383). ‘A mantra deficient in svara or sound, or falsely used, does not convey its meaning; a spoken thunderbolt, it strikes the worshipper like the foe of Indra through a fault in accent’. A reference to the well-known tale of the luckless TvaṣṭṚ who described a demon as ‘conqueror of Indra’ instead of indràśatruḥ ‘he who had Indra as his conqueror’ (bahuvrīhi). See Śatapathabrāhmaṇa, I, 6, 3,10 (and variant: Taitt. S, II, 4, 12, 1).

^{page 512 note 2} Nirukta, IV, iv, 25, ed. Mukund Jha Bakshi, Bombay, 1930, p. 202.

^{page 512 note 3} Ṛkprātiśākhya, III, 1, and Uvaṭa's commentary thereon.

^{page 513 note 1} The sentence: vyākaraṇam nāmeyam uttarā vidyā, is generally translated as ‘grammar is a (historieally) later science‘, but both interpretations are possible. The passage referred to is:Mahābhāṣya(Kielhorn's edition), I, p. 208,1. 19 and following.

^{page 514 note 1} As the performance was conducted by Vājasaneyīs, the characteristic [x] for -s- was heard in these two words, but not even the Vājasaneyīs were able to conceal the endemic tonal pattern of svarita.

^{page 514 note 2} For the musical intorpretation of the terms, see Varma, Siddheshwar, ‘The Vedic accent and the interpreters of Pāṇini’, JBBBAS, NS, XXVI, 1, 1950Google Scholar. For the musical interpretation of the accent, i.e. that the Vedic accent was a (predominantly) musical accent, see Allen, W. S., Phonetics in ancient India, London, 1953, 87–93Google Scholar. Also Sankaran, C. R., ‘Contributions on the study of Indo-European accent’, BDCRI, II, 1940–1941, 185–202Google Scholar (especially p. 192) as well as the sections dealing with Vedic accent in any descriptive grammar of Sanskrit. The authority of Wackernagel, , Altindische Grammatik, I, Göttingen, 1896Google Scholar, may be added to the group (p. 284, paragraph 244). The present pages offer a different point of view, since the question of pitch is disturbed by the fact that the anudātta can be both high and low relatively in musical pitch, and so can the udātta. It has also been suggested that ‘loud’ / ‘soft’ are the appropriate oppositions for uccaiḥ/nicaiḥ. Nica does not usually mean ‘soft’ (Patañjali has used the terms uccaih/śanaiḥ to express the opposition ‘loud’/‘soft’ or ‘gentle’; Mahābhāsya Bombay, 1935, Vol. II, p. 26, but Kielhorn's edition has nīcaiḥ(!), Vol. i, p. 207, 1. 2); if nīca did mean ‘soft’, then sannatara (I, ii, 40) would have to mean ‘softer still’ which is extremely unlikely. The term adopted here for nīcaiḥ, ‘downwards’, refers to an articulatory position, as does the term ‘more constricted’, used for sannatara. Relatively, udātta is ‘upwards’ from anudātta. In spite of the reasons given for equating uccaistarām with ‘more upwards‘, it could reasonably be held that Pāṇini did not intend this term to be understood relatively to uccaiḥ, in which case it could be understood as ‘louder’, i.e. even louder than when calling someone from a distance. Such, certainly, is the aim of the priest who wishes the gods to attend the sacrifice.

^{page 514 note 3} Kāśikā, on this sūtra, suggests that the permission to abandon svara is given to the student while he is still engaged in the effort of committing the hymns to memory. As an alternative to this unlikely explanation, Kāśikā proposes that, in reciting Brāhmanas, etc., it is not necessary to use svarawhen coming upon quotations from hymns, but both explanations are weak. A further conjecture might be that the licence is extended to certain types of Vājasaneyī chanting.

^{page 515 note 1} Staal, XIX (1) and XXI (16).

^{page 515 note 2} cf. also Patañjali's comment on this sūtra and Śānkhāyanaśrautasūtra, I, 1, 19-21.

^{page 515 note 3} See ŚŚŚ I, 4, 17.

^{page 516 note 1} Staal, X (3).

^{page 517 note 1} B (informant in the previous analysis) did show an occasional high-tone in a context such as where the first syllable of the second word was high-tone. But this was not a regular feature after hiatus.

^{page 517 note 2} The rise does not occur on open long svaritas before a long final.

^{page 518 note 1} The highly important feature of palatals, for instance. Compare p. 90, 7 (d), with end of 1 and footnote thereto, which gives a further illustration of the ‘reversal’ of tonal features.

^{page 518 note 2} An object lesson was provided by the unfortunate Nambudiri who could not, least of all in aspirate contexts, keep his anudāttas low-tone (see p. 509, n. 1).

^{page 518 note 3} To decide which features are ‘stationary’ and which involve ‘movement’ is a matter belonging to historical analysis. Both this matter as well as that of the relationship of the Vedic accent to that of other Indo-European languages demand separate studies. With reference to the historical aspect, it is interesting to compare the tonal patterns illustrated in the previous pages (pp. 91–4) with those given by Professor Firth, J. R. in ‘Phonetic observations on Gujarati’, BSOAS, XX, y, 231–41.Google Scholar

^{page 518 note 4} There is no evidence for assuming that the process of constriction results in the articulatory context required for the uttering of the important syllable. Different terms are therefore used. The suggestion that less breath is expended on udātta than on unimportant syllables might be opposed. The evidence points to a particular voice quality for udātta, and the likelihood is that less breath is released.

^{page 519 note 1} This latter conflict caused difficulties in ancient times, see BSOAS, XXII, 1, 1959, p. 90, n. 2.Google Scholar

^{page 519 note 2} Mahābhāsya, I, p. 61, II. 16 ff.Google Scholar

^{page 520 note 1} The last sentence of the above passage, yad dhy alpaprāṇasya sarvoccais tad dhi mahāprāṇasya sarvanīcaih, is generally taken to mean: ‘for what is very loud for a weakling is very soft for a vigorous person’. This is certainly Kaiyata's view. But Patanjali uses the terms alpa-and mahāprāṇa in only one other place in the Mahābhāṣya (a debt is owed to Pandit Pathak and Pandit Citrao for enabling such statements to be made), i.e. when discussing the distinctions between consonants, he equates alpaprāṇa/non-aspirate consonants and mahāprāna/aspirates (I, 61, 11. 16 ff.). This is the reason for suggesting the translation given, taking alpaprānasya as a bahuvrīhi, i.e. ‘in the case of that which has a preceding non-aspirate consonant’ (the Sanskrit grammarians always held that a vowel related to the preceding consonant). Further, the translation ‘intrabuccal space’ may be misunderstood; it is intended to convey the whole cavity as far as the back of the throat, i.e. as much as can be seen when looking into a person's mouth. The passage occurs in Mahābhāṣya, I, p. 207, 11. 1–14.Google Scholar

^{page 521 note 1} Pāṇinīyaśikṣā, verses 6–10 (Śikṣāsamuccaya, pp. 378–9).

'The mind allied with innate knowledge joins meaning with the desire to speak. The mind strikes up the bodily fire, the fire impels the air.

'The air as it moves in the chest softly gives rise to that inner sound wherein lies Gāyatrī in morning prayer.

'In the throat then Trisṭubh, middling for the noon-day prayer, and shrill in the head Jāgatī, for the evening prayer.

'The air risen up, struck baek in the head, reaches the mouth and gives birth to sounds; five kinds are known.

‘“But according to accent, to time, to place, and to movements greater and lesser”, so say those who are skilled in sound. Take careful heed of this.’