Dharma · AI Companion

祇夜

geya

十二部经中重颂经义者AI-generated

Type GenreDifficulty IntermediateTruth-level span outside the four-fold two-truths frameworkInitial AI estimate · evolves with use

A Contemporary ReadingAI-generated

What This Concept Is About

To understand 祇夜, you have to first understand the larger cabinet it lives in. The twelve-fold classification of Buddhist scripture — 十二部經 (*dvādaśāṅga-dharma-pravacana*, the "twelve kinds of teaching") — is one of the oldest ways the tradition sorted its own literature. By the time the canon was being catalogued in India, the Buddha's words circulated in many registers: long discursive prose, terse memorable verse, parable tales, prophecies, inspired exclamations, dialogues, and so on. The commentators grouped these into twelve "kinds" so that readers would know what kind of speech they were reading. 祇夜 (*geya*) is one of those twelve, and its classical definition is narrow and precise.

The classical entry defines 祇夜 as: among the twelve divisions of scripture, the verse form (偈頌體) that reiterates the prior prose text (重頌前文). That is the whole definition in one line, and the depth is hidden in the brackets.

A few unpacking notes:

祇夜 is a phonetic transcription. The two characters 祇夜 are not chosen for meaning — they sound out the Sanskrit *geya* (sometimes spelled *geyya*), which means something like "to be recited together" or "to be rehearsed." Because Chinese translators often also gave semantic renderings, you will see 祇夜 referred to in commentaries as 重頌 ("repeating verse") or 應頌 ("responsive verse"). All three labels — 祇夜, 重頌, 應頌 — name the same form.

祇夜 is verse that comes after prose. This is the defining structural feature. A teaching is first delivered in prose (長行, *vaibhāṣika* / extended prose discourse, itself part of the 修多羅 category). Then a verse passage comes along and echoes the prose in compressed, metered form. The verse is not the source; it is the recapitulation. The prose is the full argument; the verse is its recapitulating echo.

祇夜 is one verse category, not the only one. This is where most modern readers stumble. There is a second verse category, 伽陀 (*gāthā*), and the two are not the same. 伽陀 is standalone verse — a teaching delivered directly in metered form, with no preceding prose section that it is recapitulating. Many *gāthā* in the early canon are independent utterances, sometimes attributed to kings, brahmins, monks, nuns, or even non-Buddhist wanderers. 祇夜, by contrast, depends on the prior prose for its meaning. Take away the prose, and the verse becomes an orphan — still poetic, but no longer a 祇夜.

祇夜 is a memory technology dressed as poetry. Why did the tradition produce this form at all? Because for centuries the canon was preserved not in books but in human throats. Prose is hard to carry verbatim; verse, with its meter, rhythm, rhyme, and musical phrasing, lodges in the body. By repeating the teaching in compressed verse, the reciting community guaranteed that the core message would survive the trip from one teacher to the next. The verse is the load-bearing pillar; the prose is the surrounding architecture.

祇夜 presupposes oral pedagogy. A 祇夜 is meant to be *chanted together with* its prose, not read silently on its own. The commentarial tradition sometimes describes monks reciting the prose, then the verse, then the prose again — a call-and-response of full and condensed form. The repetition is the point. The compressed verse is what gets carried; the prose is what gives the verse its meaning when it's unpacked again.

A small honest caveat: the twelve-fold classification is itself a commentarial overlay. When we look at compiled Chinese translations, we can usually identify which passages *function* as 祇夜 (verses that recap prior prose), but the historical certainty about whether any given verse was originally delivered this way is not always perfect. The categories were applied retrospectively to texts that had been redacted and combined. The *definition* is clean; the *attribution* is sometimes uncertain.

Life Walk-Through

祇夜 sounds like an obscure textual category. It is not. The form is so woven into how human beings transmit important things that you already use it dozens of times a day without the Sanskrit label. Let me walk through three contemporary scenes.

Scene 1: The meditation retreat.

You arrive at a ten-day retreat. On the evening of day three, the teacher gives a two-hour discourse on how attention moves through the body, how distraction arises, how to notice it without grasping, how to return. The teaching is long, careful, full of qualifiers — exactly the prose form 修多羅. At the end, the teacher says: "Now let me give you the verse." And they chant, in a clear voice:

*"This body itself is a city,* *its senses the gates,* *the mind the king who watches —* *what arrives, know it clearly."*

That short verse is the 祇夜. It echoes the prose teaching. It compresses two hours of nuance into a few lines that you can chant on the cushion. Over the next several days, when your mind wanders, what often returns is the verse — because the verse is what lodges in memory. The teacher knows this. They deliberately offered the verse at the end so that you would have a portable form of the teaching to take into the silent days.

Now, the mapping mistake to notice here. Many retreatants will, years later, remember only the verse and not the prose. They will live by the verse. They may even come to believe the verse is the *real* teaching and the prose was "just explanation." This inverts the canonical relationship. The verse is the recapitulating echo; the prose was the full teaching. The verse works *because* the prose preceded it. Take the verse alone, and you have lost the qualifications, the exceptions, the careful distinctions the teacher made. The verse carries the core; the prose teaches you what to do when the core is in tension with reality.

Scene 2: The grandmother's saying.

You grew up hearing your grandmother say, at moments of decision: "Measure twice, cut once." She said it in English, in Cantonese, in whatever language the moment arrived in. The saying is short, rhythmic, memorable. It is your family's 祇夜.

Behind that saying is a longer teaching your grandmother absorbed somewhere — perhaps from her own father, perhaps from a vocational school, perhaps from a hard lesson paid for in ruined fabric. The full teaching was something like: "When you are excited about a project, that excitement distorts your judgment. The right move is to slow down at the moment you most want to speed up. Review your work before committing resources. The cost of one careful pause is much smaller than the cost of a flawed decision. And this applies not only to cutting fabric but to almost every irreversible choice." That is the prose (修多羅). "Measure twice, cut once" is the verse that survives.

Notice what the verse preserves: a *commitment posture* (slow down at the moment of commitment), a *consequence picture* (ruined work), and a *transferability claim* (it generalizes). It does not preserve the reasons behind the commitment posture, the specific kinds of irreversible choices, the limits of when the rule applies. Those live in the prose, and most people never had access to the prose. The verse alone can mislead: it can become a compulsive checking ritual rather than a wise posture of review.

This is exactly the risk a 祇夜 runs when it gets separated from its prose.

Scene 3: The brand jingle.

A long, expensive advertising campaign runs for two years, telling the story of a brand's values, history, and promises. That is the prose. Then the agency produces a five-note sonic logo — a sound so short you can hum it after hearing it once. That is the 祇夜. It does not contain the brand's theology. It contains the brand's theology *echoed*. When you hear the sonic logo in a different country, in a different decade, with no contact with the original campaign, you still respond. You respond to the verse without knowing the prose.

Brands know that verses travel further than prose. They also know that verses can be hijacked — that a jingle separated from its meaning becomes a sound attached to whatever meaning the listener's culture pours into it. The same dynamic is true of a 祇夜 detached from its sutra. The verse is portable; the meaning is fragile.

The walk-through reveals three patterns worth naming:

- 祇夜 is the form your memory chooses to carry. It is almost never the form your full attention originally engaged. - 祇夜 gains power with distance but loses accuracy. The further a verse travels from its prose, the more people will receive it as if it were standalone truth. - 祇夜 requires its prose to function as Dharma. A 祇夜 heard without its prose is a proverb; heard with its prose, it is a teaching.

Why Contemporary People Need This

We are living through the inverse of the conditions that produced 祇夜. In the oral-recitation world, memory was scarce and the bottleneck was *transmission*. In our information environment, transmission is cheap and the bottleneck is *what survives*. We are drowning in prose — texts, articles, podcasts, threads, videos, captions — and starving for the compressed forms that could carry a teaching forward into life. Three practical uses of the 祇夜 principle for contemporary people:

1. As a discipline of distillation. The exercise "what is the verse for this teaching?" is itself a clarifying practice. When you try to put a long lecture, therapy session, book chapter, or hard conversation into a single sentence or verse, you discover which parts of it were decoration and which were load-bearing. Most people never do this; they carry the impression of a teaching without being able to say what it is. The practice of forcing a teaching into verse form reveals its skeleton. Many wisdom traditions have versions of this — the Buddhist 祇夜 form, the Zen *kōan*'s compressed line, the haiku of insight, the aphorism of the Greek philosophers. Each is a different packaging of the same technology.

2. As a check on slogan-thinking. We are surrounded by 祇夜-shaped objects — slogans, taglines, hooks, viral phrases — that circulate without their prose. The 祇夜 form, taken seriously, builds in a check: *this verse stands or falls with the prose that generated it.* A useful slogan points back to a real teaching; a dangerous slogan floats free. Contemporary people need the capacity to ask, of any short, powerful, repeatable phrase that lands: "What is the prose for this? What teaching generated this line? Is there a real teaching, or is this verse orphaned?" This is a version of the classical skill of knowing whether a verse is a 祇夜 (with prose) or a 伽陀 (standalone) — and what difference that makes.

3. As a recovery of the somatic dimension of learning. Reading is largely visual; recitation is bodily. When you chant a verse, you breathe it, your vocal cords shape it, your chest vibrates with it, your feet tap its rhythm. Contemporary neuroscience confirms what oral traditions always knew — embodied, rhythmic rehearsal produces deeper and longer-lasting encoding than silent reading. This is not a claim that neuroscience "proves" the Buddhist insight; it is a parallel observation. The brain does, in fact, encode rhythm and motor repetition differently than it encodes silent visual symbols. The 祇夜 form anticipated this. The verse is a body-technology for memory. In an age of disembodied scrolling, recovering some daily practice of chanting even a short verse out loud is a small but real reconnection to how human beings have always known things thoroughly — through the body, not just the page.

A deeper use, harder to name: 祇夜 is a model for the relationship between understanding and embodiment. A teaching delivered in prose is something you have comprehended; a teaching carried in verse is something you have *inhabited*. The verse has entered your breathing, your rhythm, your voice. There is a difference between knowing about impermanence and being a person whose body knows impermanence — and the 祇夜 is one of the technologies for getting to the second.

Common Misreadings and Clarifications

Misreading 1: "祇夜 just means any verse in a Buddhist text." This is the most common error, and it is a flattening. Not all verses in the canon are 祇夜. Some are 伽陀 — standalone utterances without preceding prose. Some are quotations embedded in narratives as dialogue. Some are refrains, summaries, or poetic flourishes within longer compositions. Only verses that specifically reiterate prior prose content qualify as 祇夜 in the strict classical sense. A 祇夜 is defined by its *relationship to a prior prose section*, not by its being in metered form.

Misreading 2: "祇夜 is the original teaching, and the prose is later commentary." The classical definition is the opposite. The prose is the full teaching; the verse is its recapitulation. This matters because the inverse mistake — taking the verse as the "real" sutra and the prose as explanation — leads people to skip the careful exposition and live by the catchy line. In some traditions the relative priority may be debated on historical grounds, but the *definition* of 祇夜 presupposes prose priority. A teaching delivered first in prose, then echoed in verse, treats the prose as the source.

Misreading 3: "祇夜 is just an ancient literary convention, irrelevant now." The opposite is closer to true. Every practice of compressing insight into memorable form — proverbs, hymns, mantras, slogans that actually work, even good haiku — is using the 祇夜 principle. The question is whether the verse is honestly anchored to a deeper teaching, or floating free. Recovering the 祇夜 pattern in your own learning — producing verses for the teachings you actually want to carry — is a practical contemporary skill. The classical form is alive as a pattern even if you never read a Buddhist sutra.

Misreading 4: "祇夜 proves the Buddhist teachings are 'just poetry.'" The reverse. The tradition's careful distinction between prose forms (修多羅) and the two different verse forms (祇夜, 伽陀) shows that the early Buddhist community was doing sophisticated textual categorization. They distinguished between verse that stands alone, verse that recapitulates prose, prose that expounds, prose that narrates, and so on. The fact that they named and classified 祇夜 specifically shows they understood the difference between a teaching and its echo. This is a sign of literary discipline, not literary naïveté.

Misreading 5: "If I just remember the verse, I have the teaching." You have the echo. Whether you have the teaching depends on whether you also carry the prose. The verse can be a doorway into the prose, or it can be a substitute for the prose — and only the first is doing the work the canonical form was designed to do.


A short note on what I have and have not claimed. The classical definition of 祇夜 as the verse that reiterates prior prose, set among the twelve-fold classification, is well-attested across the commentarial tradition. The distinction from 伽陀 is classical. The historical claim that specific passages in compiled Chinese translations were "originally" 祇夜 is sometimes less certain, since the categories were applied retrospectively. Where I have cited specific examples (a closing verse echoing prose teaching), I have framed them as *functioning as* 祇夜 rather than as definitely *labeled* 祇夜 by the commentarial tradition. The contemporary applications — verse as memory technology, slogan-thinking, somatic encoding — are interpretive extensions, not historical claims about the tradition's own self-understanding.

Canonical EntryAI-generated

十二部经中重颂前文的偈颂体。

By SutraAI-generated

大般涅槃經(南本)T0375 · 涅槃部

《大般涅槃經(南本)》將「祇夜」列為十二部經之一,作為菩薩摩訶薩「知法」的具體內容。其論述置於菩薩住於大乘大般涅槃、修七善法的脈絡中——七善法為知法、知義、知時、知足、自知、知眾、知尊卑,而「知法」之首義即知十二部經。

該經列舉十二部經依次為:修多羅、祇夜、授記、伽陀、優陀那、尼陀那、阿波陀那、伊帝目多伽、闍陀伽、毘佛略、阿浮陀達摩、優波提舍。對於「祇夜」,該經並未作抽象定義,而是採取「以事例為界說」的方式,呈現其典型樣貌。

該經所舉的祇夜事例為:佛告諸比丘——「昔我與汝愚無智慧,不能如實見四真諦,是故流轉,久處生死,沒大苦海。何等為四?苦、集、滅、道。」此段以偈頌形式表述四聖諦之要義,並未脫離先前所說之經義而獨立存在。

該經對「祇夜」的界定,核心在於其「重頌本經」之體性:佛先以契經(散文)方式為比丘說法,說法已竟,後有利根眾生為聽法而至佛所,向旁人詢問如來方才所說之事;佛時知已,即「因本經以偈頌」重述其義。可見該經強調祇夜乃由「散文契經」與「偈頌」前後對應而構成,先有長行、後以偈頌重頌之形式,是「祇夜」一詞「重頌經義」之名的具體落實。

值得留意者,該經對「祇夜」的界說與同列十二部經之「伽陀」形成明確區隔:「伽陀」經被界定為「除修多羅及諸戒律,其餘有說四句之偈」,乃獨立於修多羅與戒律之外的四句偈頌;而「祇夜」則必須緊隨本經之後,依本經而起,與「伽陀」之獨立成頌不同。該經此種區分,正顯示其對十二部經各各體性的細緻分辨。

最後,該經以「是名祇夜經」一句作結,隨即轉入對「授記經」的界說,全段採用逐項列舉、各各解說的方式陳述十二部經,並以「菩薩若能如是了知十二部經,名為知法」總結收攝。

大智度論T1509 · 釋經論部

《大智度論》於釋十二部經名義時,對「祇夜」之界定極為簡要而明確。論中先立「修多羅」為諸經中直說(長行散文)之總名,謂四阿含、諸摩訶衍經及二百五十戒經,乃至出三藏外諸經,皆名修多羅;緊接而言「諸經中偈,名祇夜」,即以祇夜為修多羅中所攝之偈頌部分,散文與韻文對舉而分屬二部。此處「祇夜」之定位,並非獨立於經外之體裁,而是就同一經中「直說」與「偈」兩種表達形式所作之區分:直說者歸修多羅,以偈出者歸祇夜。

論中復進一步說明偈之句數不定:「一切偈名祇夜,六句、三句、五句,句多少不定,亦名祇夜,亦名伽陀。」 此句透露兩層要義:其一,祇夜之形式不拘,六句、三句、五句皆可,無固定句數限制,凡偈頌體皆得稱祇夜;其二,論中明言祇夜與「伽陀」(gāthā)為同義互稱,二者在《大智度論》之用法中並無嚴格區隔,一切偈既可名祇夜,亦可名伽陀。此點與後世部分論師將祇夜專指「重頌」(以偈重述長行之義)、伽陀專指「孤起頌」(不依長行而獨立發起)之嚴格二分有所不同;《大智度論》此處係以寬義用之,凡經中偈頌通名祇夜,亦通名伽陀,未作重頌與孤起之別。

就術語形式而言,《大智度論》用「祇夜」之譯名,與「修多羅」「優陀那」等並列,皆取梵語音譯,不另立義譯(如「重頌」「應頌」等名),此為羅什譯本之一貫風格。論中又舉「優陀那」為例,謂「有法佛必應說而無有問者,佛略開問端」,以無問自說為優陀那之特徵,與祇夜之為經中偈頌形成對照:祇夜依附於經(修多羅)而存在,優陀那則因無問而自發,二者發生之因緣不同。

此外,該段原文於祇夜定義之後,大篇幅論及「授記」(記別)之相:佛欲與眾生受記先現微笑,光從四牙出,照六道而演無常、無我、涅槃之法,光還繞身七匝,隨所記之道從不同部位入(記地獄從足下入、記畜生從𨄔入、記人道從齊入、記天道從胸入、記聲聞從口入、記辟支佛從眉間相入、記得佛從頂入)。此段雖非直接論祇夜,但顯示《大智度論》釋十二部經時,各部門義往往相連而說,祇夜之偈頌定義置於修多羅與授記之間,作為十二分教名相辨析之一環。

總之,《大智度論》論祇夜,核心在「諸經中偈」四字,以偈頌為經文之韻文部分,句數不定,與伽陀通名互用,不作重頌、孤起之細分,術語取音譯「祇夜」,此即本論對該概念之獨特表述。

瑜伽師地論T1579 · 瑜伽部

《瑜伽師地論》以「應頌」為十二分教之第二,全篇不用音譯「祇夜」,亦不稱「重頌」,一律以義譯「應頌」立名。此為本論術語形式之要點:凡涉及十二分教之安立,皆書「應頌」,與「契經」「記別」「諷頌」等並列。

本論對應頌之定義見於兩處,表述互有側重。第一處云:「云何應頌?謂於中間、或於最後宣說伽他,或復宣說未了義經。是名應頌。」此將應頌析為二義:一、於長行(散文)之中間或最後所宣說之伽他(偈頌);二、宣說「未了義經」,即義理尚未究竟顯了之經。 第二處云:「應頌者,謂長行後宣說伽他,又略摽所說不了義經。」此處將位置限定為「長行後」,不復言「中間」;又將「未了義經」改書為「不了義經」,並加一「略摽」,謂應頌乃以偈頌形式略為標舉前文長行所說之不了義經。 兩處合觀,本論對應頌之理解兼攝「形式」(伽他偈頌之位置)與「功能」(標舉不了義經令得重述)兩面,而「略摽」一詞為第二處所獨有,強調偈頌對長行義理之提綱作用。

在術語層面,本論以「伽他」稱偈頌(梵 gāthā 之音譯),以「長行」稱散文體之直說,以「了義」「不了義」判經義之究竟與否。第一處用「未了義經」,第二處用「不了義經」,語序微異而義同,皆指尚未被記別(第三分教)廣為分別之經義。 此與本論記別之定義相呼應:「記別者,謂廣分別略所摽義,及記命過弟子生處。」 亦即應頌「略摽」於前,記別「廣分別」於後,二者構成一略一廣之對待關係。

本論復將應頌置於「正法」之總框架下:「所謂契經、應頌、記別,廣說如前十二分教。是名正法。」 應頌既為正法之組成,亦為說法者所當善巧運用之教體。論說「善順入言」時云:「唯善顯現解釋契經、應頌等法,終不引餘外道邪論。」 此謂善於說法者,於應頌等分教能善巧顯現解釋,而不雜引外道邪論,是應頌在論議實踐中之定位。

綜上,《瑜伽師地論》論應頌之要旨為:(一)術語上一律用義譯「應頌」,不用「祇夜」音譯;(二)定義上兼取「伽他偈頌」之形式義與「略摽不了義經」之功能義;(三)結構上與記別構成略廣對待;(四)實踐上為說法者善巧解釋之對象。

Related