¹ Thieme, Studies presented to Joshua Whatmough 263.

A version of this paper was presented at the LSA meeting at Indiana University, 31 July 1964.

The following abbreviations are used: Bh. (Pātañjalamahābhāṣya, ed. Kielhorn; references are given by a roman numeral followed by two arabic numerals, referring to volume, page, and line); Kāś. (Kāśikāvṛtti, Kāśi-Saṃskṛt-Granthamālā; Benares, 1952); Nyās. (Kāśikā Vivarāṇa Pañcika, ed. S. C. Chakravarti; 3 vols., Dacca, 1925); SK (Siddhāntakaumudī, Chaukhambā ed., 1958); Kāt. (Kātantra, ed. Liebich, Zur Einführung 1, 1919); AP (Atharvaprātiśākhya, ed. Whitney; JAOS 7, 1862); RP (Rikprātiśākhya, ed. M. D. Shastri; Allahabad, 1931); TP (Taittirīyaprātiśākhya, ed. Whitney; JAOS 9, 1871); VP (Vājasaneyīprātiśākhya, ed. Weber; IS 4, 1858). A reference made by means of three arabic numerals with no further specification is to a rule of Pāṇini's aṣṭādhyāyī.

² Pāṇini gives phonetic details concerning the pronunciation of anunāsika sounds and the svarita pitch (1.1.8, 1.2.32), and Thieme (op.cit. 265–7) believes this is to be explained as due to these sounds being ‘“made up” of two different, independently occurring sounds’. Another possibility, which I consider more plausible and plan to discuss in detail on another occasion, is that these descriptions were required because Pāṇini meant to use nasalized vowels and svarita pitch as part of his metalanguage.

³ RP 13.11 classes vowels, anusvāra, and spirants as aspṛṣṭa, while TP 2.45 specifies that the middle of the articulator (karaṇamadhya) is open (vivṛta) in the articulation of spirants. AP 1.31 reads ūṣmaṇāṃ vivṛtaṃ ca, in which ca refers to īṣat of 1.30, hence the spirants are classed as īṣadvivṛta. This is also the classification of the Madhya- and Laghusiddhāntakaumudī (ad 1.1.9), while SK 12 classes both spirants and vowels as open (vivṛtamūṣmaṇāṃ svarānāṃ ca). On Patañjali's discussion see note 7.

⁴ See Bh. I.61.14–9, 62.4; SK 12; AP 1.12–3; RP 13.1–5; TP 2.4–10.

⁵ This order of sounds appears in RP 1.1–10, TP 1.5–9, VP 8.2–22, Kāt. 1.1.2–15.

⁶ The finally accepted analysis of āsya in Bh. ad 1.1.9 is that it is a taddhita derivative with suffix -ya (5.1.6) like dantya. Therefore āsya, analyzed as meaning āsye bhavam ‘located in the mouth’ (Bh. I.61.25), includes a reference to sthāna and, concomitantly, to karaṇa; cf. Bh. I.61.25–6 kiṃ punarāsye bhavam / sthānaṃ karaṇaṃ ca. Bh. I.62.11–4 gives three alternative analyses of the compound tulyāsyaprayatnam, with āsya considered a taddhita derivative, but, in the last analysis, these interpretations are prompted by arguments which are not cogent, so that āsyaprayatna could be taken to mean ‘effort in the mouth’. Nevertheless, we would still have to interpret it as referring to a sthāna; otherwise the result of a sequence d-h by the application of 8.4.62 (§6) could be gh. Hence I do not think we can state, with Thieme (Pāṇini and the Veda 94 n.1), that Pāṇini did not know the doctrine of sthāna and karaṇa.

⁷ Bh. I.64.7–9 also considers qualifying ūṣman as īṣadvivṛta (see note 3), but this is only a tentative statement and Bh. goes on to place the concept of savarṇatva within the context of a five-member sequence of rules (see §4); within this context the spirants are classed as vivṛta.

⁸ For our purposes 1.3.3 is important: halantyam ‘(In the instruction,) a final hal (consonant is termed it)‘.

⁹ Bh. I.64.11–5: varṇānāmupadeśastāvat / upadeśottarakāletsaṃjñā / itsaṃjñottarakāla … pratyāhārah / pratyāhārottarakālā savarṇasaṃjñā savarṇasaṃjñottarakālam … savarṇagrahaṇam / etena sarveṇa samuditena vākyenānyatra savarṇānāṃ grahaṇaṃ bhavati ‘Now the instruction of sounds; after the instruction, the technical term it; after the term it (is introduced) … pratyāhāra; after pratyāhāra, the term savarṇa; after the term savarṇa (is introduced) … the inclusion of savarṇa (sounds). By this whole unit statement the inclusion of savarṇa (sounds) comes about.‘

¹⁰ The Pāṇinean procedure has the additional advantage of actually postulating, strictly on the basis of synchronic morphophonemics, a form which is to be reconstructed; PIE *g^when: Skt. hanti, pft. jaghāna.

¹¹ Cf. also Emeneau, Sanskrit sandhi §33.

¹² Similar formulations in AP 3.7, RP 4.5, TP 5.38, VP 4.121. In the Prātiśākhyas morphophonemic changes are viewed as sound modifications; cf. Renou, Terminologie grammaticale du sanskrit 3.490 s.v. vikāra. For Pāṇini these changes are substitutions. Since the vikāras of the former result in the formulation of rules pertaining to individual sounds, I have taken the liberty of referring to them as sound-for-sound substitutions as opposed to Pāṇini's system of feature substitution. Cf. Bh. I.31.18, where vikāra is equated with ādeśa.

¹³ Cf. AP 2.13, 31, 35; RP 4.10, 6, 7; TP 5.25–8. It is to be noted that although AP has a rule similar to Pāṇini's sthāne'ntaratamaḥ (AP 1.95 āntaryeṇa vṛttiḥ ‘The procedure [of change] is according to similarity‘), it still formulates the rules in question separately.

¹⁴ Other treatises accord independent status to anusvāra, as well as the spirants visarjanīya, jihvamūlīya, and upadhmānīya, in their catalogs of sounds; Kāt. 1.1.16–9, RP 1.22, TP 1.12, VP 8.24–7.

¹⁵ VP 1.63 has a similar rule but including pluta vowels: hrasvagrahaṇe dīrghaplutua pratīyāt ‘(When a) short vowel (is cited) one should understand (in its) stead (also) long and pluta vowels’. This resembles Pāṇini's procedure, but for morphophonemic purposes sim (see below) is defined to include only short and long vowels.

¹⁶ This rule and AP 3.42, TP 10.2 come under the regimen of a rule stating that a single sound takes the place of both sounds mentioned in the rules following (AP 3.41, TP 10.1, VP 4.49); cf. note 17.

¹⁷ This rule comes under the regimen of 6.1.84 ekaḥ pūrvaparayoḥ ‘One (member takes the place) of the preceding and following’. Cf. note 16.

¹⁸ See Thieme, ZDMG 107.665–6 (1957), who renders savarṇa ‘von gleicher Gattung’. Cf. D. S. Ruegg, JAOS 82.67 (1962).

¹⁹ The method adopted by Pāṇini is of course important also for the organization of sounds in the śivasūtras. Part of this question is discussed perceptively and lucidly by J. F. Staal, Lg. 38.1–10 (1962); but I think the materials allow a detailed, step-by-step reconstruction of Pāṇini's construction of the śivasūtras on the basis of an order of sounds as given in §2 (cf. Thieme, op.cit. 101–7), and that this cannot be done if the vowel and consonant sūtras are considered separately.