परिभाषाप्रकरणम्

Covers SK 34–46 (13 sūtras): all present.

Paribhāṣā-prakaraṇa — outline (p02, SK 34–46, all 13 sūtras)

These are the interpretive metarules — no sound is changed here; each fixes how a rule elsewhere is to be read: which element a named operation seizes, where an augment lands, how a sūtra's case-endings point, which substitute wins and how much it replaces, and how clashing rules are ranked. Grouped by that job, not strict SK order (reorderings noted): the three vibhakti-reading paribhāṣās (38, 40, 41) are pulled together; the four substitution-extent paribhāṣās (42, 44, 45, 43) are pulled together, with 43 placed beside the 45 it excepts.

A. Fixing which sound a named phonetic operation seizes — running example: mid + guṇa → med-, śrīpā → śrīpam

• 34 iko guṇavṛddhī 1.1.3 — wherever a rule prescribes an operation under the very words "guṇa"/"vṛddhi," the genitive ikaḥ presents itself: the guṇa/vṛddhi falls on an ik (i u ṛ ḷ). Worked in mider guṇaḥ (7.3.82): the guṇa lands on the i of mid. Does not present itself at a mere restatement like vṛddhir yasyācām (1.1.73), nor where the words guṇa/vṛddhi are absent (tyadādīnām aḥ 7.2.102).
• 35 acaś ca 1.2.28 — the twin rule for the other named family: an operation prescribed by the words "hrasva/dīrgha/pluta" seizes an ac (a vowel). Worked in hrasvo napuṃsake (1.2.47): the final ā of śrīpā (an ac) shortens → śrīpam; but not in supāt brāhmaṇakulam, since the stem supād does not end in an ac.

B. Fixing where an augment (āgama) lands — running example: bhav-i-tā, vār-ī-ṇi

• 36 ādyantau ṭakitau 1.1.46 — a ṭit augment is the initial part of the term it is prescribed for, a kit augment its final part (yathāsaṃkhya pairing). The ṭit augment iṬ of ārdhadhātukasyeḍ valādeḥ (7.2.35) attaches at the front → bhav·i·tā; the kit augments kuK/ṭuK of ṅṇoḥ kuk ṭuk śari (8.3.28) attach at the end → prāṅ·k·ṣaṣṭha. (Apavāda only of the immediately following ṣaṣṭhī sthāneyogā 1.1.49, itself blocked for taddhita/kṛt affixes by the later paraś ca 3.1.2.)
• 37 mid aco 'ntyāt paraḥ 1.1.47 — a mit augment is inserted after the last vowel of the whole aggregate it is prescribed for (hence it is that aggregate's final part, not a fresh initial). Reading acaḥ as a genitive-of-specification, nUM goes after the last ac of the stem so the aṅga still ends in n and sarvanāmasthāne ca (6.4.8) can lengthen it → vār·ī·ṇi; and mUM after the last ac keeps the pada m-final so mo 'nusvāraḥ (8.3.23) can act → vahaṃ·liha-. Bhāṣya: samudāyabhakto mit ("the mit belongs to the aggregate").

C. Reading a sūtra's case-endings as pointers — which neighbor a stated condition targets — running example: iko yaṇ aci → dadhy atra

• 38 ṣaṣṭhī sthāneyogā 1.1.49 — a genitive in a sūtra means "in place of." In iko yaṇ aci (6.1.77) the genitive ikaḥ marks the ik as the sthāna (= prasaṅga, the slot due to be displaced): the yaṇ replaces the ik rather than resting on it — so in dadhy atra the i is turned out for y. Restricted to genitives not otherwise determined: gohaḥ in ūdupadhāyā gohaḥ (6.4.89) is a partitive (fixed by co-occurring upadhā), so no sthāna-sense there.
• 40 tasminn iti nirdiṣṭe pūrvasya 1.1.66 — a locative condition (aci) targets the immediately preceding element. In dadhy atra the yaṇ falls on the i contiguous to the following a; it does not reach across the intervening dh in samidham, nor onto the following ik u in dadhi udakam (both would follow from a bare locative-absolute reading). Only varṇa-separation is barred, not time-separation (that comes from the saṃhitā-adhikāra).
• 41 tasmād ity uttarasya 1.1.67 — the mirror rule: an ablative condition targets the immediately following element. Under udaḥ sthāstambhoḥ pūrvasya (8.4.61) the pūrvasavarṇa is blocked in ut·pra·sthānam because pra intervenes; and the tiṅ ṅatiṅaḥ (8.1.28) nighāta hits the following verb — agnim ī́ḻe loses accent, but not in ī́ḻe agnim.

D. Choosing which candidate becomes the substitute — running example: vāg ghariḥ, cetā

• 39 sthāne 'ntaratamaḥ 1.1.50 — when several ādeśas are in prospect for one sthānin, the nearest (by sthāna > artha > guṇa > pramāṇa) is chosen. For the sthānin h in vāg ghariḥ, the savarṇa that comes in is gh — it alone shares all four efforts (ghoṣa, nāda, saṃvāra, mahāprāṇa) with h — not g/ṅ (lack mahāprāṇa) or kh (lacks the other three); the superlative -tama rules out the near-misses. And the appended paribhāṣā (sthāna beats the rest): in cetā/stotā the guṇa of i/u is e/o by place-nearness, not a, though a is nearest by measure (pramāṇa). (Placed after 38/40/41 as it presupposes the sthāna-sense; other worked cases in-unit: artha → kroṣṭuḥ, pramāṇa → adas short u / long ū.)

E. Fixing how much of the sthānin the substitute replaces — running example: tyadādīnām aḥ → yaḥ, saḥ; aster bhūḥ

• 42 alo 'ntyasya 1.1.52 — a substitute for a genitive-designated item replaces only its final al (sound). By tyadādīnām aḥ (7.2.102) the a replaces only the last sound of yad/tad → yaḥ, saḥ. Guards: "ādeśa" excludes the iṬ augment of 7.2.35 (an āgama is not an ādeśa); "of al" keeps vasusraṃsu…daḥ (8.2.72, under adhikāra padasya) from swallowing the whole final pada in paramānaḍudbhyām.
• 44 ādeḥ parasya 1.1.54 — an apavāda of 42: where the target is a following (para) element (fixed by 41), the substitute hits its initial al, not its final. Under dvyantarupasargebhyo 'pa īt (6.3.97) the ī replaces only the initial a of ap → dvīpam, not the whole ap.
• 45 anekālśit sarvasya 1.1.55 — an apavāda of 42 on the other side: a substitute that is anekāl (more than one al) or śit replaces the whole sthānin. bhū (anekāl) for as in aster bhūḥ (2.4.52) replaces the entire root. By paratva (vipratiṣedhe paraṃ kāryam 1.4.2, equal-strength → later wins) it overrides even ādeḥ parasya: at the aṣṭā as stage, auś (aṣṭābhya auś 7.1.21) replaces the whole ending, not just the initial a.
• 43 ṅic ca 1.1.53 — the scoped exception carving ṅit forms out of 45: a ṅit substitute, though itself anekāl, replaces only the final al. avaṄ, anaṄ and the like replace just the last sound. Explicitly an apavāda of anekālśit sarvasya, justified by yena nāprāpte and by "the śāstra of the particular is stronger" (a special rule for a subclass would otherwise be niravakāśa). (Placed beside 45, the rule it excepts, not in SK order.)

F. Marking a heading to recur, and ranking rules that clash — running example: tudati, dadhnā, ubhaye devamanuṣyāḥ

• 46 svaritenādhikāraḥ 1.3.11 — a word pronounced with the svarita accent is adhikṛta: it recurs (anuvṛtti) as a governing heading into the following sūtras; the svarita is known by the author's avowal (pratijñā), the anuvṛtti's far limit by received vyākhyāna. The unit then appends the rule-strength paribhāṣās (each later type stronger): para (from vipratiṣedhe paraṃ kāryam 1.4.2); nitya over para — in tudati the nitya affix śa (tudādibhyaḥ śaḥ 3.1.77) acts first and, being ṅit-treated, blocks the light-penult guṇa; antaraṅga over para — in ubhaye devamanuṣyāḥ the case-independent sarvanāma-saṃjñā (1.1.27) applies invariably, overriding the optional prathama-carama- (1.1.33), with seed asiddhaṃ bahiraṅgam antaraṅge; apavāda over the rest — in dadhnā the final-only replacement by ṅic ca (1.1.53) beats the whole-replacement anekālśit sarvasya (1.1.55). Closes with akṛtavyūhāḥ pāṇinīyāḥ (e.g. seduṣaḥ) — an operation is not run when its nimitta is about to perish. (Bālamanoramā colophon to the paribhāṣā-prakaraṇa follows.)