Definition
The provisional / definitive distinction (neyārtha / nītārtha; literally “meaning to be drawn out” / “meaning already drawn out”) is the Mahāyāna hermeneutical device for classifying teachings of the Buddha by their interpretive status. Neyārtha teachings are pedagogical, expedient, and require further interpretation to disclose what the Buddha intended; nītārtha teachings already articulate the final meaning and require no further drawing-out. The Pāli antecedent, the Aṅguttara passage usually cited as AN 2.25 (Karunadasa cites it as A. II, 60), uses the same pair: a statement whose meaning is already drawn out (nītattha) should not be understood as one whose meaning has to be drawn out (neyyattha), and vice versa; one who disregards the distinction misrepresents the Tathāgata (karunadasa-theravada-abhidhamma-2010, Introduction p. 10).
The distinction is hermeneutical, not evaluative: a teaching being neyārtha does not mean it is false but that it was given to a specific audience for a specific purpose and is to be read accordingly. Selflessness preached to one who clings to the self is not the same level of statement as the analysis of dharmas in their universal lack of intrinsic nature. The device is the Buddhist analogue of the Brahmanical pūrvapakṣa / siddhānta layering but operates internally to the Buddha’s own teaching rather than between schools.
In the Madhyamaka tradition the distinction does load-bearing interpretive work, and two ways of applying it need to be kept apart. One sorts teachings by content. For Candrakīrti, sūtras explaining something other than suchness teach expedient truth and are to be interpreted, while those explaining emptiness teach definitive truth (MA 6.97; dzongsar-khyentse-madhyamakavatara-2003, p. 207); Gorampa heads the passage “These procedures of interpretability and definitiveness are applicable to all sacred writings” (gorampa-removal-wrong-views, p. 279). Mabja likewise divides sūtras by content into those of expedient meaning, “which explain the vast relative truth”, and those of definitive meaning, “which teach the ultimate truth of profound emptiness” (mabja-ornament-of-reason, Preliminary Discussion). The other way sorts by turning: the Saṃdhinirmocana itself calls the first two turnings “surpassable and adapted to the circumstances” and “of provisional meaning”, and the third “of definitive meaning” (Toh 106, 7.30). The Mādhyamikas do not take over that second sorting. Candrakīrti counts the Saṃdhinirmocana’s own three-natures teaching among the sūtras of interpretable meaning (MA 6.95, as Gorampa explains it, p. 280), and by the content criterion the first-turning teachings on the existence of dharmas and the third-turning mind-only and tathāgatagarbha teachings come out neyārtha, the Prajñāpāramitā teaching of universal niḥsvabhāva nītārtha. The Jonang school keeps the sūtra’s ordering of the turnings, reading the third as nītārtha and the second as neyārtha. Where the cut is made is therefore one of the principal axes of Tibetan doctrinal disagreement — yet the device itself is presupposed by every voice in the dispute.
Comparison matrix
| Thinker | What is nītārtha | What is neyārtha | Source | Wiki link |
|---|---|---|---|---|
| Nāgārjuna | Does not name the distinction in the MMK, VV, VP or ŚS. On Candrakīrti’s reading of MMK 18:6, “neither self nor non-self” is taught to the advanced | On the same reading: a self taught “for teaching purposes” to materialists who deny that actions bear fruit, no-self to those bound by attachment to self. ŚS v. 44 says only that the Buddha speaks through intention | Mūlamadhyamakakārikā 18:6–8 with the Prasannapadā (Sprung 1979, pp. 174–182); Śūnyatāsaptati v. 44 | |
| Candrakīrti | Madhyamaka teachings on emptiness | Cittamātra / Vijñānavāda and tathāgatagarbha as expedient | Madhyamakāvatāra 6.94–97 | Candrakīrti |
| Tsongkhapa | Second-turning emptiness, read through Prāsaṅgika Candrakīrti | First turning, third turning incl. tathāgatagarbha and cittamātra (per Laṅkāvatāra’s self-gloss); buddha-nature literature neyārtha | Drang nges legs bshad snying po (the canonical Geluk hermeneutical reduction); tsongkhapa-illuminating-intent-1418 on MA 6.94 | Tsongkhapa |
| Mabja | By content, sūtras “which teach the ultimate truth of profound emptiness”; by time of teaching, the third wheel is “the final wheel of Dharma on the definitive meaning” | By content, sūtras of expedient meaning, “which explain the vast relative truth”. He does not class the third turning as provisional; he rejects the Yogācāra reading of its intent and reconciles the first two wheels by the two truths. Applied to the MMK itself: chapter 1 “proves that teachings on the absence of arising are of definitive meaning”, while the teachings on arising “are of expedient meaning and concerned with what is false and relative” | mabja-ornament-of-reason Preliminary → “Classifications of the Words of the Buddha”; on chapter 1 | Mabja |
| Gorampa | Second turning | Tathāgatagarbha literature interpretable in the freedom-from-extremes register at MA 11.34, but zhentong ontologisation is rejected | gorampa-distinguishing-views-1469; gorampa-removal-wrong-views | Gorampa |
| Ninth Karmapa | Second turning Madhyamaka | Buddha-nature and zhentong explicitly classified as provisional meaning — directly opposes Dolpopa / Tāranātha | karmapa-feast-fortunate-1578 | Ninth Karmapa |
| Mipham | Second turning, with third-turning convergent under Yogācāra-Madhyamaka two-step | First turning neyārtha; Cittamātra neyārtha (using the Laṅkāvatāra’s self-gloss) | mipham-introduction-middle-way-2002 (Cittamātra refutation MA 6.45–97) | Mipham |
| Dolpopa | Third turning (tathāgatagarbha / zhentong) | First and second turnings neyārtha; the Prajñāpāramitā universal-emptiness teachings as preparatory — self-emptiness of the ultimate read as a technique for non-conceptual meditation (Hopkins 2006, p. 205) | dolpopa-mountain-doctrine-1333 (Hopkins 2006, pp. 199, 202, 206, 394); synthesis at taranatha-essence-other-emptiness-2007 | Dolpopa |
| Shakya Chokden | Third dharmacakra essential for identifying the meditative object; both second and third dharmacakras admissible as nītārtha under different aspects | Neither turning is exclusively neyārtha; the Niḥsvabhāvavāda and Alīkākāravāda both lead to the same direct experience | komarovski-visions-unity-2011 | Shakya Chokden |
| Siderits (analytic) | The Madhyamaka prajñā-perfection register (BCA 9; MMK 18:6) | The Reductionist register (BCA 8; non-self preached to those who cling to self) — a “graded” or progressive teaching in which chapter 8’s arguments are “provisional”, offered “as no more than a suggestion”; he does not use the Sanskrit terms | siderits-reality-altruism-2000 pp. 421–422; siderits-buddhism-philosophy-2007 p. 206 | Siderits |
| Karunadasa (Theravāda) | Nītattha — statements “whose meaning is already drawn out,” to be taken as they stand (explicit/definitive) | Neyyattha — statements “whose meaning has to be drawn out” (to be interpreted) — but in Pāli neither is ranked above the other (Edgerton); the device sorts direct vs indirect meaning, not higher vs lower truth; it is the canonical antecedent of the Theravāda two truths | karunadasa-theravada-abhidhamma-2010 Ch 3 (AN; Aṅguttara commentary) | Karunadasa |
Textual loci
- Mūlamadhyamakakārikā 18:6–8 — the locus classicus in the Madhyamaka corpus, though Nāgārjuna does not name the distinction. MMK 18:6: the Buddhas taught a self, taught no-self, and taught neither self nor non-self. MMK 18:8: “Everything in this world can be taken as real or not real; or both real and not real; or neither real nor not real. This is the Buddha’s teaching” (sprung-lucid-exposition-1979, p. 181). Candrakīrti reads 18:6 under “the universal rule that a distinction between truth for the initiated (nītārtha) and what is merely truth for beginners (neyārtha) exists in the teaching” of the Buddhas: the self is taught “for teaching purposes” to materialists who deny that actions bear fruit, no-self to those tied by attachment to the self, and neither to those near nirvāṇa (Sprung pp. 175–177). He introduces 18:8 as “this graduated (anupūrvī) teaching” and assigns its four alternatives to hearers at successive stages (pp. 181–182). As a modern parallel, Garfield reads 18:6 by audience and 18:8 by levels of truth (garfield-fundamental-wisdom-1995); Siderits cites 18:8 for “graded” teaching (siderits-reality-altruism-2000, p. 422).
- Śūnyatāsaptati v. 44 — in the root, the Buddha speaks of “existence”, “non-existence” and both through his intention (དགོངས་པ་), and this is hard to understand. The translation’s further words, that this intention is the two truths and that the verse “must be interpreted in this light”, are the translators’ interpolations from the commentaries (komito-seventy-stanzas-1987, pp. 156, 182). The verse is Nāgārjuna’s own statement that the Buddha’s words carry an intended sense; it does not name the two truths or the neyārtha / nītārtha pair.
- The MMK’s purpose, as the commentators state it. Candrakīrti says Nāgārjuna composed the treatise because “he wanted to demonstrate the proper distinction” between sūtras of definitive and of provisional meaning, and quotes the Akṣayamatinirdeśa’s criterion: those spoken “for the sake of entering the path” are provisional, those spoken “for the sake of attaining the final goal” definitive (sprung-lucid-exposition-1979, pp. 44–45). Sprung’s note calls the distinction “indispensable to Madhyamika’s interpretation of the mass of Buddha statements”, while objecting that the renderings “‘provisional’ (neya) and ‘final’ (nīta) truth cannot hold for Madhyamika” (p. 44 n. 1). Tsongkhapa repeats the point: the treatise is written “distinguishing interpretable from definitive meanings” (tsongkhapa-ocean-of-reasoning-1408, Sarnath ed. p. 24).
- Gorampa’s apparatus for interpretable teachings (on MA 6.44). Teachings of interpretable sense require a threefold explanation, “of their intention ground, their purpose and their obstruction of explicitness” (gorampa-removal-wrong-views, p. 227) — in the Tibetan, the intended purport (དགོངས་གཞི་), the purpose (དགོས་པ་) and the objection to the literal meaning (དངོས་ལ་གནོད་བྱེད་); the Prāsaṅgika is “this supreme system, which by way of the two truths differentiates between the interpretable and the definitive” (p. 228). Here the two truths and the provisional / definitive pair are joined explicitly, in the commentator’s voice.
- Bodhicaryāvatāra 8.101–103, 9.1 and 9.7 — Śāntideva’s argument from impartial benevolence (Ch 8, Perfection of Meditation) is read as provisional Reductionism; Ch 9 (Perfection of Understanding) supplies the nītārtha corrective. BCA 9.1 is the text’s own internal warrant: “all this was taught by the Buddha for the sake of gaining prajñā.” See siderits-reality-altruism-2000 pp. 421–422. The Nyingma side gives the same reading (see kunzang-pelden-nectar-manjushri-2007, ch. 9 following Mipham’s Norbu Ketaka): the commentary on BCA 9.1 reads the five prior perfections as “branches… for the sake of wisdom,” and at BCA 9.7 gives an explicit neyārtha reading inside chapter 9. The Buddha’s teaching that the aggregates exist and are momentary “was speaking on the level of expedient meaning; his real intention was only implied… thinking only of the mode of appearance — his purpose being to lead the worldly (as yet unable to understand emptiness) gradually onto the path of the authentic Middle Way.” Here the graded-teaching device operates verse-internally, in the tradition’s own voice.
- Madhyamakāvatāra 6.94–97 — Candrakīrti’s classification of the Laṅkāvatāra Sūtra’s mind-only teaching as expedient. The Laṅkāvatāra self-glosses cittamātra as neyārtha, which grounds the Madhyamaka hermeneutical reading in the sūtra’s own words. Cited by both Mipham (in his Cittamātra refutation) and Thuken.
- Ākāśagarbha Sūtra 1.15 — the graded ordering stated by the Buddha in a Kangyur sūtra, with its purpose named. Beginner bodhisattvas are to be taught the six perfections “with a reference point, with the notion that suchness is expressible,” and the great elements as arising and perishing; only then familiarised with all phenomena as inexpressible, non-arising, non-ceasing and “not in the slightest way existing.” Taught in that order, they “definitively abandon eternalism (རྟག་པའི་ལྟ་བ་) and nihilism (ཆད་པའི་ལྟ་བ་), and they will not be at all frightened or terrified.” Two features distinguish this from the general commendations of graded teaching. What is graded is the hermeneutical posture accorded to the words — suchness taken as expressible at the first stage, that posture withdrawn at the second — rather than the difficulty of the material, which puts the passage closer to neyārtha / nītārtha proper than to a bare pedagogical recommendation. And the terminus is specified as neither extreme, tying the device to the anti-nihilist result instead of leaving the two doctrines separate. The reason for the staging is given at 1.34: a direct display of unborn ultimate reality would leave the world’s inhabitants “confused and defiled,” a limit that holds for bodhisattvas up to the eighth ground. See akasagarbha-sutra-84000-2014.
- Ākāśagarbha Sūtra 1.60–1.61 — the same ordering with a sanction. Teaching the profound-emptiness sūtras to the unprepared, who are then frightened and turn back from complete enlightenment, is the first of the eight root downfalls of a beginner bodhisattva; and bodhisattvas must first know others’ inclinations and “teach the Dharma step by step.” The text specifies that the offending transmission is faithful “in word and meaning,” so what is faulted is the ordering, not the doctrine. The list this heads is the canonical source of the fourteen root downfalls used across the Tibetan schools (Bodhicaryāvatāra 5.104; Sakya Paṇḍita’s Thub pa dgongs gsal). This is the strongest available evidence that the graded device is an operative commitment rather than an exegetical convenience — see mis-sequenced-emptiness-teaching-is-a-downfall.
- Aṣṭasāhasrikā Prajñāpāramitā — not a locus for the neyārtha / nītārtha pair, which is not named in this text, but the clearest statement in the wiki of the reading practice the pair formalises. The 84000 translators record that this literature is studied in the Indian and Tibetan scholastic traditions for “the framework it sets out for the ‘path,’ the implicit meaning behind the texts’ explicit statements” — an explicit / implicit division of a sūtra’s content that is the operative hermeneutic of the Abhisamayālaṅkāra commentarial line and is applied to the Prajñāpāramitā by Indian commentators. Two related passages show the practice at work. At 1.6 the criterion of a successful hearing is affective and practical rather than propositional, and “precisely this state is the instruction.” At 6.15 the sūtra states outright that a certain point is “not to be taught to beginning bodhisattvas because it might undermine their faith, appreciation, trust, or devotion” — the graded device operating within the text, in its own voice, without the vocabulary. See astasahasrika-84000-2026.
- Aṅguttara Nikāya (Pāli; usually AN 2.25, cited by Karunadasa as A. II, 60) — the antecedent: two people misrepresent the Tathāgata, the one who presents a discourse whose meaning is to be drawn out (neyyattha) as one whose meaning has been drawn out (nītattha), and the one who does the reverse (karunadasa-theravada-abhidhamma-2010, Introduction p. 10). The device is canonical, not a Mahāyāna innovation.
- Akṣayamatinirdeśa Sūtra — standard Mahāyāna source for the rule that nītārtha sūtras teach emptiness while neyārtha sūtras teach selves, persons, sentient beings, etc. Candrakīrti quotes it in stating the MMK’s purpose (see above; Sprung 1979, p. 45), and it is cited downstream by the Geluk hermeneutical tradition.
Interpretations
The graded-teaching reading (Nāgārjuna, Candrakīrti, Tsongkhapa, the broad Madhyamaka mainstream): neyārtha and nītārtha describe a teaching strategy in which the Buddha meets students where they are, then progressively withdraws the conceptual scaffolding he provided. To one clinging to the self, no-self is taught; to one who has reified no-self into a metaphysical position, the four extremes are negated; at the end-point even the nītārtha teaching is to be let go of (the raft simile). The device is therefore not a list of which sūtras are which but a function on the relation between teaching and student — the same verse can be neyārtha for one audience and the boundary of the nītārtha for another.
The Jonang inversion (Dolpopa, Tāranātha; see dolpopa-mountain-doctrine-1333): here the second-turning Prajñāpāramitā teachings on universal emptiness are neyārtha, because they describe only rangtong (self-emptiness) and so do not address what truly exists — the zhentong-empty tathāgatagarbha. Dolpopa concedes that the middle wheel teaches that even the ultimate does not ultimately exist (Hopkins 2006, p. 199), but reads this as a meditative technique (Hopkins 2006, p. 205) and faults the middle wheel for “overstating” self-emptiness by extending it to the ultimate (Hopkins 2006, p. 206; introduction, p. 24). The third turning “teaches directly,” the other two “obliquely by way of intentional speech” (Hopkins 2006, p. 394; introduction, p. 24). In Dolpopa’s own words, “all statements—in these and those texts of the middle wheel of doctrine—of the non-self-empty as self-empty are just of interpretable meaning with a thought behind them” (Hopkins 2006, p. 364). This is the principal axis of Jonang departure from rangtong Madhyamaka, and the target of Tsongkhapa’s Drang nges legs bshad snying po.
An objection from inside (Gendün Chöpel). In the Adornment (see lopez-madmans-middle-way-2006) Gendün Chöpel turns the device on itself: correct reasoning is found in the definitive scriptures, yet the provisional and definitive are told apart by reasoning — “If one understands with reasoning, why search for the definitive meaning? If one does not understand with reasoning, how does one find the definitive meaning?” (). He does not abandon the device: the distinction between assertions for oneself and assertions for others, he says, “is not different in the other tenet systems that assert a presentation of the provisional and definitive and the two truths” (). The circle is a challenge the necessity thesis has to meet, posed by a framework-internal author.
The Kagyü middle position (Ninth Karmapa): tathāgatagarbha literature is explicitly classified as provisional meaning, directly opposing Dolpopa. The Kagyü position therefore aligns with Geluk on the neyārtha / nītārtha sorting of the third turning while differing from Geluk on other axes (qualifier, four koṭis, pramāṇa). This is one of several places where the standard “rangtong vs zhentong” framing fails to capture the actual Tibetan picture: the Karmapa retains zhentong-style buddha-nature material as expedient rather than rejecting it.
The Sakya / Nyingma freedom-from-extremes reading (Gorampa, Mipham): tathāgatagarbha is not collapsed to deflationary med dgag (Tsongkhapa) and not reified as zhentong truly-existent (Dolpopa) but read as mtha’ bral — beyond all four extremes. Whether this counts as a third reading of nītārtha or as a refusal of the binary sorting is itself contested. See tathagatagarbha-and-four-extremes and tathagatagarbha.
Śāntarakṣita / Mipham — the device structuring a synthesis, not just a sorting: the neyārtha/nītārtha device usually ranks the turnings (which is nītārtha, which is neyārtha). The Madhyamakālaṅkāra’s two-step method shows it can instead structure a synthesis of them. Cittamātra supplies the conventional account and Madhyamaka performs the ultimate analysis on it: “On the basis of the Mind Alone… outer things do not exist… mind is utterly devoid of self” (MA v. 92). The graded structure is built into the architecture of the path (v. 93, “those who ride the chariot of the two approaches”). The approximate ultimate functions as the neyārtha-side reasoned negation, the actual ultimate as the nītārtha terminus beyond all assertion. The Padmakara translators qualify the ranking reading. The Cittamātra accepted conventionally is “scriptural” Cittamātra (བཀའི་སེམས་ཙམ་), not the tenet system, and it “need not be considered inferior to” Madhyamaka; the Nyingma approach “resists the tendency to distinguish hierarchically the scriptures of the second and third turnings”; and the association of Cittamātra with the relative and Madhyamaka with the ultimate is not “meant to suggest that the latter is more important” (shantarakshita-madhyamakalankara, Translators’ Introduction). The two steps are therefore an order of teaching, not a ranking of the turnings. It is the clearest case of neyārtha/nītārtha as framework-as-pedagogy doing constructive rather than merely sorting work, and the device that makes Madhyamaka the definitive final move on a received (here, Cittamātra) provisional conventional truth (see madhyamaka-is-method-not-standalone-system and shantarakshita-madhyamakalankara).
Siderits’s analytic re-invocation: in siderits-reality-altruism-2000 (pp. 421–422), Siderits defends Śāntideva against Williams by reaching, tentatively, for the graded-teaching device — BCA 8 read as provisional Reductionism, BCA 9 as the Madhyamaka nītārtha corrective, with cross-reference to MMK 18:8. Siderits does not name neyārtha / nītārtha and offers the reading “as no more than a suggestion” (p. 422), but the shape is the device’s. This is corroboration from the analytic side that the device is doing load-bearing interpretive work even for readers without antecedent commitment to it. He invokes it independently again at siderits-buddhism-philosophy-2007 p. 206 on MMK 18:6.
Karunadasa’s Theravāda-internal reading — the device is continuous, and it is the elder; the ranking is not: In karunadasa-theravada-abhidhamma-2010 (Ch 3; Introduction p. 10), Karunadasa treats the Aṅguttara nītattha / neyyattha distinction as the canonical antecedent of the Theravāda two truths, and reports that the Aṅguttara commentary itself “seeks to establish a correspondence between the original sutta passage and the Theravāda version of the two truths.” Two statements in that chapter matter to the argument here and should be quoted rather than paraphrased.
“The distinction is so important that to overlook it is to misrepresent the teachings of the Buddha.”
“It seems very likely that this distinction between nītattha and neyyattha has provided a base for the emergence of the subsequent doctrine of double truth, not only in Theravāda but also in other Buddhist schools.”
The second is a genetic claim, and it inverts the dependency the Mahāyāna material usually suggests. On Karunadasa’s account, which he hedges as “very likely”, the sorting device is not derived from the two truths; the two truths are in part derived from it. So the device does not presuppose a two-truths doctrine — it presupposes only that a teaching may have an intended sense distinct from its surface sense. The reverse direction holds too, and separately: in the Madhyamaka corpus the commentators make the two truths the content the sorting sorts by — Gorampa’s system “by way of the two truths differentiates between the interpretable and the definitive” (p. 228), and Mabja reconciles the first two turnings by the relative and ultimate perspectives. Nāgārjuna’s own Śūnyatāsaptati v. 44 says only that the Buddha speaks through intention. The pair supplies what the sorting sorts by; the sorting is what makes the pair usable on a text.
How far the two usages coincide, and the one place they part. Since a great deal now rests on this device, the extent of the overlap needs stating rather than assuming. They coincide on what is sorted — statements whose meaning is already drawn out (nīta-attha), to be taken as they stand, against statements whose meaning is still to be drawn out (neyya-attha) — and on the penalty, which the Aṅguttara puts as making “a false statement with regard to the Blessed One.” Karunadasa glosses the pair “in a broad way to mean the difference between the direct and the indirect meaning,” which is exactly the operation the Indian and Tibetan commentaries perform on the Mahāyāna sūtras.
They part on ranking. Karunadasa is explicit that the sutta passage “makes no preferential value-judgement in respect of the two statements. One statement is not singled out as higher or lower than the other,” and, citing F. Edgerton, that in Buddhist Hybrid Sanskrit literature “a nītārtha text… is recommended as a guide in preference to one that is neyārtha,” whereas “in Pāli neither is ipso facto preferred to the other; one errs only in interpreting one as if it were the other.” So the Pāli device sorts modes of statement; the Sanskrit device additionally grades them. The grading is what allows teachings to be arranged in an order along which a practitioner is led, and it is therefore the part of the device the framework-as-pedagogy reading actually needs. The honest statement of the continuity is: continuity of the sorting instrument, which is canonical and not a Mahāyāna innovation, with the evaluative ranking as a Sanskrit-Buddhist development. An argument that leans on the Pāli warrant to carry the grading is over-claiming.