संज्ञाप्रकरणम्

Saṃjñā-prakaraṇa — outline (p01, SK 1–33, all 33 sūtras)

Covers SK 1–33 (33 sūtras): all present. No SK number unplaced. Two notes on order: (i) SK 2 (pratyāhāra formation) is presented after SK 1 and SK 3, the it-rules it consumes, though it stands textually between them; (ii) the orchestrator's suggested group "(G) pragṛhya prep" has no sūtra in p01 — pragṛhya is defined later (from A 1.1.11) — so no such group appears here.

A. Flagging a sound as it (the it-saṃjñā) — running example: the Śivasūtras a i u ṇ and l a ṇ, whose end-markers must be stamped it before any pratyāhāra can be read off them

• 1 halantyam 1.3.3 — a final hal (consonant) in upadeśa is it. In a i u ṇ the final ṇ (a hal) becomes it — a mere marker, not one of the denoted sounds; likewise the ṇ/k/ṅ/c… ending each of the fourteen sūtras. (BM construes it as the locative "the final in [the sūtra] hal" to avoid anyonyāśraya with the hal-saṃjñā.)
• 3 upadeśe 'j anunāsika it 1.3.2 — a nasalized (anunāsika) ac in upadeśa is it; which vowels are nasal rests on school-avowal (pratijñānunāsikya). The a of l a ṇ, pronounced nasal, becomes it, yielding the marker for the pratyāhāra ra (r + l); the rule also underwrites that it-sounds sitting inside a pratyāhāra are not counted in it (pratyāhāreṣv itāṃ na grahaṇam — else the k of ṛ ḷ k, taken as an ac, would trigger yaṇ and swallow the i of anunāsika).

B. Drawing sounds together into a pratyāhāra — running example: building aṇ (a i u) and ac (all nine vowels)

• 2 ādir antyena sahetā 1.1.71 — an initial sound pronounced with a final it is the saṃjñā of itself, the it, and every sound between. Initial a + final it ṇ → aṇ = a, i, u; initial a + the it c (end of ai au C) → ac = a i u ṛ ḷ e o ai au. The middle sounds are included because "initial" and "final" (per the Bhāṣya on 1.1.21) presuppose a middle; svam is drawn from 1.1.68 for "and of itself." Forty-four pratyāhāras are actually used in the grammar.

C. Length and pitch of the vowel — running example: the vowel a, sorted into its 3 durations × 3 pitches = 9-fold

• 4 ūkālo 'jjhrasvadīrghaplutaḥ 1.2.27 — an ac is hrasva / dīrgha / pluta as its duration equals u (1 mātrā) / ū (2) / ū3 (3). So 1-mātrā a = hrasva, 2-mātrā ā = dīrgha, 3-mātrā a3 = pluta.
• 5 uccairudāttaḥ 1.2.29 — an ac produced in the upper part of its place of articulation is udātta; e.g. the ā and e of ā ye (RV ā ye mitrāvaruṇā).
• 6 nīcairanudāttaḥ 1.2.30 — an ac produced low is anudātta; e.g. the initial a of arvāṅ.
• 7 samāhāraḥ svaritaḥ 1.2.31 — an ac in which udātta-ness and anudātta-ness are gathered together is svarita.
• 8 tasyādita udāttamardhahrasvam 1.2.32 — of a svarita the first half-mātrā is udātta, the remainder (by pariśeṣa) anudātta; e.g. the svarita ī of agnim īḷe (with udātta-śruti on its tail). With 3 lengths × 3 pitches, each single ac is thus 9-fold.

D. Homogeneity (savarṇa) and its exact scope — running example: the contrast daṇḍa + āḍhakam → daṇḍāḍhakam (savarṇa-dīrgha fires) vs dadhi + harati → dadhi harati (yaṇ blocked)

• 9 mukhanāsikāvacano 'nunāsikaḥ 1.1.8 — a sound uttered by mouth-with-nose is anunāsika. The nasal/non-nasal split doubles the 9-fold ac to 18-fold (so a, i, u, ṛ have 18 varieties each; ḷ 12, having no long; the ec 12, having no short).
• 10 tulyāsyaprayatnaṃ savarṇam 1.1.9 — sounds sharing sthāna (place) and ābhyantara-prayatna (internal effort) are mutually savarṇa; sāvarṇya turns on sameness of the whole set of oral places (the nose not counting). a and ā (both throat + vivṛta) are savarṇa, so in daṇḍa + āḍhakam savarṇa-dīrgha yields daṇḍāḍhakam.
• 11 a a 8.4.68 — the last sūtra of the grammar; it re-imposes the natural saṃvṛta (closed) a in finished forms, superseding the vivṛta a taught in the inventory. Because it is asiddha throughout (see SK 12), during derivation a stays vivṛta — hence still savarṇa with ā, keeping SK 10's savarṇa-dīrgha available.
• 12 pūrvatrāsiddham 8.2.1 — adhikāra: the Tripādī (8.2–8.4) is asiddha with respect to the earlier grammar, and even within it a later rule to an earlier. This is śāstra-asiddhatva, not kārya-asiddhatva; it is what makes a a (SK 11) inert in derivation, grounding that vivṛta a. (Its Kaumudī block also completes the sound-analysis with the 11-fold bāhya-prayatna and the ṛ/ḷ- and a/h-, i/ś-, ṛ/ṣ-, ḷ/s- sāvarṇya vārttikas.)
• 13 nājjhalau 1.1.10 — an ac and a hal are never savarṇa. In dadhi + harati, the i (ac) and h (hal) are non-savarṇa, so iko yaṇ aci cannot apply — the form stays dadhi harati (not *dadhy harati); likewise before ṣ, s, ś.
• 14 aṇudit savarṇasya cāpratyayaḥ 1.1.69 — an aṇ (or a u-marked sound), when not itself the item being prescribed, denotes its savarṇas as well as its own form. So the i of iko yaṇ aci stands for all 18 i-varieties — but, by SK 13, never for h. (This grahaṇaka rule's aṇ alone uses the later ṇ; the mention of aṇ rather than ac also gathers in nasal/non-nasal y, v, l.)
• 15 taparastatkālasya 1.1.70 — a tapara aṇ (one flanked by t) denotes savarṇas only of its own duration — a niyama restricting SK 14. Thus at, it, ut… denote just their 6 same-length varieties each (not the long), and ṛt denotes 12 (ṛ + ḷ classes). Completes the pratyāhāra + grahaṇaka apparatus.

E. The vowel-grade saṃjñās vṛddhi and guṇa — running example: the classes named off the pratyāhāras built above

• 16 vṛddhirādaic 1.1.1 — āt and aic (ā, ai, au) are termed vṛddhi. Though Pāṇini's opening sūtra, it is placed here because its tapara āt and its pratyāhāra aic presuppose all of A/B/D; naming vṛddhi is what later vṛddhi-operations invoke.
• 17 adeṅ guṇaḥ 1.1.2 — at and eṅ (a, e, o) are termed guṇa; the class summoned by every guṇa-operation.

F. Word-class saṃjñās: dhātu, nipāta, upasarga, gati — running example: prefix + root, e.g. pra + √nī → praṇayati; and the nis/nir split

• 18 bhūvādayo dhātavaḥ 1.3.1 — the items of the bhū-gaṇa, and those like vā (i.e. action-denoting), are dhātu. "Action-denoting" keeps the homonym particle vā (option) and hiruk (exclusion) out.
• 19 prāgrīśvarānnipātāḥ 1.4.56 — adhikāra: everything taught from here up to (but not including) adhirīśvare (1.4.97) is nipāta; nipāta runs by anuvṛtti through that stretch.
• 20 cādayo 'sattve 1.4.57 — the ca-gaṇa words, when their sense is not a substance (sattva/dravya), are nipātas; ca, eva… as particles, not when e.g. paśu names an animal.
• 21 prādayaḥ 1.4.58 — pra and the rest are (nipātas). Supplies the prādi list itself: pra, parā, apa, sam, anu, ava, nis, nir, dus, dur, vi, āṅ, ni, adhi, api, ati, su, ud, abhi, prati, pari, upa.
• 22 upasargāḥ kriyāyoge 1.4.59 — the prādis, when construed with a verbal action, are upasarga. pra + √nī, in construction, is an upasarga — licensing ṇatva: praṇayati / praṇeyam.
• 23 gatiś ca 1.4.60 — and, in construction with an action, they are also gati (both saṃjñās co-present, by the ca, overriding ā kaḍārād ekā saṃjñā). The nis/dus vs nir/dur split in the list explains nilayate / dulayate (l-substitution by upasargasyāyatau) against nirayate / durayate (blocked, the ru of sasajuṣo ruḥ being asiddha there).

G. Interpretive paribhāṣās governing how other rules are read — running example: correctly reading er ac and agner ḍhak

• 24 na veti vibhāṣā 1.1.44 — vibhāṣā names negation-and-option together, not mere option. This lets an ubhayatra-vibhāṣā (e.g. vibhāṣā śveḥ, saṃprasāraṇa of √śvi in liṭ) run uniformly — negation first (via na) levels the kit/apit forms, then a single affirmation-first option proceeds; vibhāṣā also stands as upalakṣaṇa for every option-word.
• 25 svaṃ rūpaṃ śabdasyāśabdasaṃjñā 1.1.68 — in a rule a word denotes its own form, except a grammatical saṃjñā (vṛddhi, guṇa…). In agner ḍhak the word agni means just the form agni, not synonyms for fire; but vṛddhi, guṇa etc. still reach their named sounds.
• 26 yena vidhis tadantasya 1.1.72 — a qualifier-word denotes what ends in it (and itself). In er ac, the affix ac attaches after a root ending in i and after a root that is i — cayaḥ, ayaḥ. Its two vārttikas (folded in: SK 26 k2/k4) bar this tadanta-vidhi in samāsa- and pratyaya-rules, except where an ugit or a single varṇa is mentioned — securing ṅīp in atipacantī and iñ in dākṣiḥ.

H. The junction and syllable-weight saṃjñās used across sandhi — running example: śikṣā (its cluster kṣ, and the weight of the preceding i); contrast sudṛṣatprāsāda

• 27 virāmo 'vasānam 1.4.110 — the cessation of utterance (absence of a following varṇa) is avasāna; the environment invoked e.g. in kharavasānayoḥ (visarga for repha).
• 28 paraḥ sannikarṣaḥ saṃhitā 1.4.109 — the utmost proximity of sounds (no gap beyond the unavoidable half-mātrā) is saṃhitā — the standing condition under which sandhi operates.
• 29 suptiṅantaṃ padam 1.4.14 — what ends in a sup or a tiṅ is a pada (sup being the pratyāhāra from su to the final p); e.g. rāmaḥ (sup-final), bhavati (tiṅ-final).
• 30 halo 'nantarāḥ saṃyogaḥ 1.1.7 — hals unseparated by an ac are a saṃyoga. In śikṣā the contiguous kṣ is a saṃyoga; the saṃjñā belongs to the aggregate/each pair, so in sudṛṣatprāsāda the t and p (an ac intervening) form no saṃyoga, and saṃyogāntalopa is rightly avoided.
• 31 hrasvaṃ laghu 1.4.10 — a short vowel is laghu; the i of śikṣ is intrinsically laghu.
• 32 saṃyoge guru 1.4.11 — a short vowel before a saṃyoga is guru (hrasvam carried from SK 31). The i of śikṣ, standing before the saṃyoga kṣ, is guru — which licenses the affix a by guroś ca halaḥ, giving śikṣā.
• 33 dīrghaṃ ca 1.4.12 — a long vowel too is guru, needing no following saṃyoga (saṃyoge not carried forward). This closes the saṃjñā-prakaraṇa.