[6/7] 南傳佛教研究中的“經典”與“理性”問題 

[Bilingual, Chinese & English] 

§6.

我的第二個說明試圖為已經描述過的那一困惑給予積極補救,但也將會說明一些來自三藏組織的進一步誤解。圖例後緊接有我建議的4個分類範疇,首先被列出,然後做出解釋(盡可能簡要)。在從左至右解讀圖例過程中,我們可以看到三個傳統“分籃”的分解,即三藏的組織構成(1.,《律藏》(Vinaya),2.《經藏》(Sutta),3.《論藏》(Abhidhamma),儘管以下的討論中會提到一些針對這一傳統的明顯例外情況);在圖例頂端我們可看到我所添加在經典中的三組分類範疇(自上而下)如下所示,第四個範疇未作闡明。

  1. “核心經典”(The Core Canon)在圖例中間的主圈上。
  2. 我所稱為邊緣經典的(Peri-Canonical)文本,要求以下解釋最多,緊位於核心的左右。
  3. 准經典文本(Quasi-Canonical),距離核心最遠(同樣,位於左右)。
  4. 非經典(Non-canonical)南傳佛教文學(例如,未聲明被包括在經典中的文本)。這些文本未有顯示在圖例中(因為我們的目的是說明經典中包含什麼,而非不包含什麼)。

我所指的“核心經典”僅僅包括前四部《尼柯耶經》(Nikāya)(參見圖例細節)。這即是我們現在稱為《經》(sutta)之一素材最古老的材料,而且(大致)類似於(一些)(未完好保存的)稱為《阿含》的漢文經典。把這四類(只有這四類)組織在一起並非新主張,但卻厘清了一個存在于巴利文研究中通常不甚明晰(且應用起來缺乏一貫性)的假定;如馬拉拉斯克拉(Malalasekera)所說的《尼柯耶經》,“前四部屬同類型,且文字同源。” [Malalasekera, 1937–8 (自此以後引用稱為 DPPN) ,*8 參見詞條《經藏》(Sutta Piṭaka)] 這是一種禮貌的方式來說第五部在某種意義上被排除在了和頭四部同樣具有的分類範疇之外(此區別以下將做討論)。

“巴利文經典”文本如上圖所示,都位於核心經典的左右。在左側,《具足戒》架起了核心經典與《律藏》(Vinaya)文學間的假設橋樑。這些規則被發現在《波羅提木叉》(Pātimokkha)和《經分別》(Suttavibhaga)中都有些微差別;規則為後來的《律藏》文本(顯示於圖例較左側的幾個圈中)完善提供了基礎。這些文本發展的基本順序和優先次序不存在爭議,因為可以通過文本內的證據進行闡明,例如,他們如何源於這裡或那裡(如圖例所示,《附篇》是最後添加進去的,故顯示於距離核心經典最遠處,且《犍度部》(Khandhaka)繼《經分別》之後,等等。)[參見 Hinüber, 2000, 第 20–22頁] 這一舊式歷史視角使《律藏》文學的完善進行了幾個世紀,同時規則為所有晚至西元前250年的佛教學派提供了一核心共識; [Nyanatusita, 2008, 第 24–25頁和注釋 11] 這一時間線上的細節與基本假設在如今的學術界中都存在爭論。

寺規在修造《律藏》文學以及將《經藏》文學重修為現存形式方面都有著特殊的優先考量;《經藏》本身的概念似乎已從這些戒律中轉移[Nyanatusita, 2008, 第 50–54頁] 且,我要補充的是,在一部分《律藏》文學和 (《長阿含經》(Dīgha Nikāya)的)Mahā-Parinibbāna-Sutta(相當於漢語的《大般涅盤經》之間存在著一種互反關係。[見 Obermiller, 1932] 這是南傳佛教學術界常現的情況:從起源處分離的常會在其修校時加以合併。

圖例左側顯示了一個有序的進程,而右側則相當混亂與費解。在核心經典右邊(位於圖例中)是第五部《尼柯耶經》,帶有一長串文本名單,這些名單被認為是當今斯里蘭卡正教中的南傳佛教經典的一部分。我特提斯里蘭卡是因為一些文本不在斯里蘭卡的經文中,卻被緬甸接受為是經典的;這些出現在圖例中作為“准經典的”,被組織在分開的一個圈裡(位於第五部《尼柯耶經》右側)第一標題為《彌蘭陀王問經》(Milindapañha)。

在斯里蘭卡正教典範中,第五部《尼柯耶經》處於邊緣地位,且在南傳佛教歷史中一直被認為其正統程度變化不定。巴利文評注(關於覺音,書於大約西元400-450年)記錄道,許多這些文本被(同時代的)斯里蘭卡某些僧侶教派所抵制;據報導,一派經院學者(《長阿含經》(Dīgha Nikāya)的持誦者)抵制這四部文本,認為是完全非經典的,並把剩餘的12部文本推給了《論藏》文學名目範疇。[Law, 1930, 第176和195頁; 參見 DPPN, 參見詞條《長部持誦者》(Dīghabhānakā), 參見詞條《所行經藏》(Cariyāpiṭaka)] 對古籍存在的這類爭論反映了佛教學者以其自我的標準,來不懈地爭取將他們自己的名目分類與巴利文語料庫中“准經典的”成分進行分割的現狀。

回顧當下關於圖例中每一部文本的文獻學爭論超出了本篇的範疇(儘管,我注意到,Hinüber的 [2000] 《手冊》試圖提供這樣一種調查);然而,我要強調的是,這些(我所稱作)“邊緣經典” 的相關年代測定多常依賴文本的內在證據,故而無可爭議。例如,在斯里蘭卡正教中,《本生經》(Jātaka)的詩歌被認為是經典的,但散文不是:文本中的詩歌與散文間存在著明顯的對比,(所以)無可爭議(且不會冒犯到任何宗教),即使認為散文後期增添的層次現在或存爭論。若我們接受這頭一個區別,我們就必須(因此)接受引用剛剛提到散文層的第三方文本其起源更晚(故而越來越“欠經典”)。這並非只是假設:在第五部《尼柯耶經》(Nikāya),如圖所示,被稱為《所行藏經》(Cariyāpiṭaka),的(例如)最後文本中已經得到了證明。[DPPN,

在參見詞條《所行藏經》] 顯然,如果《本生經》的散文是非經典的,那將在定義(部分從非經典的來源所派生的)更晚期文本為“經典的”問題上出現矛盾。這類矛盾情況存在於第五部《尼柯耶經》中,因此,如我所講,在沒有現代研究者介入的情況下,古代文物研究學者有理由認為這些文本欠完全經典。

再舉一個例子,我會注意到第五部《尼柯耶經》中另一,名為Apadāna(《譬喻經》)的文本,據稱引用了《論藏》(Abhidhamma)中一篇稱為《論事》(Kathāvatthu)的文本;[DPPN, 參見詞條Apadāna] 後一文本傳統上被認為屬於阿育王時代(即西元前250年左右)。*9 事實上這是一種“傳統上的”歸結,並不意味著實際如此,而是,意味著該文本的情況於現代化之前的歷史時期便已受到來自寺院學術方面的爭論,且其來源被認為很可能是佛陀圓寂的結果(在沒有來自現代學者的干涉之下,且因此現在提出同樣觀點不表明存在爭議。)在此情況下,一個文本派生出另一文本而後使雙方都變得含混不清,毫無疑問:如果《譬喻經》遲於《論事》,應在何種意義上稱其為“經典的”?儘管我提出這一“邊緣經典的”新分類範疇,我這裡的觀點是只有這個術語本身是新的:這些文本已經被本土學者作為了次級分類(在某種意義上),而沒有受到任何來自西方的影響。斷定一些(在經典裡的)文本較其他文本更具經典性既不新鮮,亦無冒犯。

總之,這些矛盾的情況不過是在一致地標明第五部《尼柯耶經》是作為經典外沿的繁雜分類(而前四部《尼柯耶經》的內容裡卻不存在這樣可能的變化。)

甚至更有教益的發現是,巴利文評述記錄著以擴展第五部《尼柯耶經》去涵蓋《律藏》文學與《論藏》的方式來組織整部經典的替代方法。[DPPN, 參見詞條 Khuddaka Nikāya (《小部經典》)] 若這一方法勝出,當(四部《尼柯耶經》中的)核心經典明確地分在一邊,而“其餘所有”放在第五部《尼柯耶經》的繁雜分類項中,則會令更為簡單的南傳佛教經典二元分類觀成為可能。我認為,這可能會不受歡迎,因為多少降低了寺廟規範的地位(即,如果我們只有兩個分類項,我們似乎會“降低”《律藏》規範為等同於和收集的詩歌與野史軼聞一個層次了)。*10 更為根本的是對這類經典組織的矛盾心態的存在表明了我們如今考慮的“經典結構”有著從(不斷變化著的)文化態度至文本的歷史衰落。這些態度不再能輕易改變,因為巴利語本身現已絕跡,且相當少的人熟諳其文學。

我將不再多佔用篇幅來描述准經典文本這第三分類範疇。 最難以討論的部分是《律藏》文學的結構(圖例左側),且之前對其(在勾勒寺院規範的作用,等等)已做充分解釋。說道《論藏》(Abhidhamma)文學(位於圖例右側),難度較少但爭議較多;這些文本的語料庫被所有專業學者明確承認為經典之中增加的最不經典成分,且據相關年表,是被最後添加進去的 [Law, 1930, 第 183頁; 參見 DPPN參見詞條《阿毘達摩論藏》(Abhidhamma Piṭaka)] Hinüber 評論說,“《論藏》文本的語言明顯與前兩部《藏》(Piṭakas)中發現的使用方法存在區別。” [Hinüber, 2000, 第 68頁] 這類證據很明顯,但僅僅是對非常少,在研究一手來源的人們而言。

將南傳佛教《論藏》歸納為准經典因兩個值得(簡要)提及的原因才體現出其“爭議性”,因為這兩個原因妨礙到現今在東南亞進行的實地考察。第一個是《論藏》文本在東南亞廣大的教派和文化中以其具有魔力而尊崇,且這類儀式中幾乎沒有涉及文本本身的解讀。*11 東南亞中極少有人認為《論藏》是個可以接觸到的文本語料庫;對大多數人,其詞彙體現的是捉摸不定的意象。第二個是,術語“論藏”具爭議就是因為詞語本身現在太經常地被誤用(和曲解),以至於人們常感困惑和冒犯行事,而無論我們如何去謹慎地談論有關這些話題。

正如我曾在圖例中注意到,我們今天對詞語“本生經”(Jātaka)有著同樣的問題,但當代佛教徒視之為一種流派風格(美學定義),他們有關《論藏》的思想反而是超自然且玄虛莫測的。某種意義上, 用同樣類似“本生經”風格的詞語去稱謂的情況並不令人感到驚奇,即,即便某個講故事的人知道故事最初寫于中世紀的泰國,並與古代印度無任何關聯的情況下。誠然,厘清一部文本為當地所著(有關較近世紀)的《本生經》並無冒犯之處。而另一部被認為起源是“經典的”,因為該術語現在被非正式地用於表現涵蓋野史軼聞和古代內容之共有風格。相比之下,當我向一位柬埔寨教授解釋,說他所認為的“論藏”事實上最初寫於二十世紀的緬甸,且從未存在有比現代緬甸語更古老的語言版本時而倍感惱火。他之所以感到詫異是因為他許多年前就已將該文本從英語譯成了柬埔寨語。他曾認為是在翻譯經典的《論藏》,但對該詞本意為何並無明確的概念。《本生經》與《論藏》這兩部文本,我要注明,是作為整個東南亞大陸喪葬儀式中的一個組成部分;這就是為何它們在這裡,而非其他地區,被嚴肅看待的簡單緣由(不考慮其具體內容)。

對比于我剛才述說的尊崇備至,許多西方學者將時間傾注在《論藏》文學上,留下了對文本所做的校訂批註。C.A.F. Rhys-Davids,(據她估計)這樣大約做了40年,把歐洲人努力研究《論藏》描述為僅僅捕獲了“一袋子老鼠”的狩獵探險;她反復地發問這整項事業是否是在“浪費時光”。[Rhys-Davids, 1923, 第 245–6頁] 她也的確將《論藏》的研究評論為(完全接在核心經典的“經部”之後那些文本的語料庫)證明了與早期經典相比較的一些有趣的智力進步;Rhys-Davids 稱這些後期的進步為一種新的“同思想和語言相一致的學科“,也是發生在“或許幾個世紀進程中”的“一種強烈而自省性成長” [同上, 第247頁] 然而,她對《論藏》文學的整體評價則是一座“滅絕文化的作品的紀念碑”。[同上. 第245頁] 她澄清自己的意圖是要吸引對“紀念碑”一詞在語源學方面的注意力:她並非只是斥其為墳墓,而是一座空墓。[同上]

在討論《論藏》時,我們在現代學術和(當下)佛教傳統兩方面態度間有著明確的比較。為了現代學術的目的,這些是公然的非經典文本,(多少)被事後添加到經典之中。而我們有著關於南傳佛教寺院傳統中這些來源之經典性持異議看法的評論記錄(類似於之前對第五部《尼柯耶經》的探討),超自然神話用來解釋經典中包含的這些文本,本身就是一個在更晚時代的捏造(即,神話本身最初出現在評論中,在經典裡並無根據)。[Hinüber, 2000, 第 66頁] 此處的問題不是神話和現實的對比,而是不同神話時代之間的對比。沒有人在真正閱讀這些文本時能夠接受它們是與其他經典同時起源的這個理論;理論是否超是超自然的幾乎不重要。

不再進一步偏離主題,我要注明,《論藏》在比較經典的每一版本和其他經典版本時,也是最富變化的:

“也許值得注意的是,這一梵文經典包含了七部《論藏》文本,正如同巴利文大藏經含有七部《論藏》文本。但這兩套文本似乎毫不相干。他們並非來自一個共同傳統的變體,而是獨立構成。” [Clark, 1930, 第137頁]

“而且, [漢語版本的經典] 包括了七部《論藏》文本,但這些看來與巴利文《論藏》文本截然不同……” [同上,第127頁]

我的觀點是,這種非對稱性因相對晚期學者的作用而變得更為重要,如玄奘,在中國其譯作(於西元七世紀)對(非南傳佛教)《論藏》文本起到了保護和促進。結果是,像《阿毘達摩俱舍論》(Abhidharmakośa) 這樣(無有與南傳佛教經典類同)的著作,在中國成為流行(且在那裡被認為是經典的重要組成部分)。剛剛提及的文本由天親(Vasubandhu)在西元四世紀所寫,與歷史上的佛陀沒有任何可能的聯繫是公開的事實;從南傳佛教觀來看,它是“偽《論藏》”著作;相反,鑒於所有我們回顧的所有上述現實,南傳佛教《論藏》整體可稱是“偽經典的”。提到的詞語“准經典”會涵蓋所有這些公開接在經典之後的文本,而且還被(不恰當地)包含在經典內;這讓我們避免了在一個又一個視角中做出哪些文本為“偽的”表述。

前面有關經典的“實踐指導”,其主要優勢實際上與所有我說過的相反:而我已經總結了在實地考察進行之前研究者們應該理解的歷史區別,這也讓他們能夠在面對印刷和談論兩種方式時,不去理會許多(謬誤的)歷史區別。

目前, 每一類南傳佛教學術研究對有關(我曾稱作的)核心經典中發現的文本所提出的偽歷史主張都極為寬容。關於“晚期”和“早期”特定文本(甚至特定觀念)的主張提出得散漫而不正式,經常缺少恰當的引用;學習何時忽視這些主張(以及何時去認真對待)本身就是一門科學。(來自一位知名學者的)一個重要例子,馬拉拉斯克拉(G.P. Malalasekera)駁斥了他自己較好的研究,提出閻王(King Yama)(也作閻羅)僅在“晚期文學”中對亡靈進行審判;[Malalasekera, 1967, 第85頁] 相反,同樣是他自己的一篇文章(幾十年前所作!)告訴我們閻王出現在南傳佛教大藏經中最古老的部分中。[DPPN, 參見詞條Yama] 無論如何,閻王作為亡靈的判官(以及地獄之王)出現於任何經典(即,我稱為的核心經典)裡的佛教文本中最古老的那部分,都無可爭議;*12 然而,在近幾代學者中,順便(沒有依據,或甚至引文地)提出這類偽歷史區別已成為了常規——即使他們自己更清楚這點。

當所有這些告誡都已(本著實用建議的精神)提出,我要重申,我這裡的資訊一直是肯定的:正如Dharani名稱的例子(上§4),經典材料裡有著無數對實地考察有用的應用,而且我們在實地考察中得到的發現也會激發出對經典新的探尋。遺憾的是,這並不時興。


§6. My second illustration tries to offer a positive remedy for the confusion I’ve already

described, however, it will also illustrate some further misconceptions arising from the organization of the Tipiṭaka (三藏). Immediately after the chart, the four categories I have proposed are first listed and then explained (as briefly as possible). In reading the chart from left to right we see a breakdown of the three traditional “baskets” that the Tipiṭaka is organized into (1., Vinaya, 2., Sutta, and 3. Abhidhamma, although, in the discussion below some significant exceptions to this tradition are mentioned); at the top of the chart we see the three categories that I’m imposing onto the canon (from the top down) as follows, with the fourth remaining unstated.

  1. The core canon, above the main circle in the middle of the chart.
  2. What I have called peri-canonical texts, requiring the most explanation below, to the immediate left and right of the core.
  3. Quasi-canonical texts, the furthest removed from the core (again, on both the left and the right).
  4. Non-canonical Theravāda literature (i.e., texts with no claim to being included in the canon). These texts are not shown on the chart (as our purpose is to show what is included within the canon, not what is excluded by it).

What I refer to as “the core canon” includes only the first four Nikāyas (refer to the chart for details). This is the most ancient of what we now call sutta (經) material, and is (roughly) analogous to (some of) what the Chinese canon has (imperfectly) preserved as the Āgamas (阿 含). Grouping these four together (and these four only) is not a novel claim, but simply clarifies an assumption that is often implicit (and applied inconsistently) in Pali studies; as Malalasekera says of the Nikāyas, “The first four are homogeneous and cognate in character.” [Malalasekera, 1937–8 (henceforth cited as DPPN),8 s.v. Sutta Piṭaka] This is a polite way of saying that the fifth one is, in some sense, excluded from being in the same category as the first four (a distinction discussed in the pages following).

The “peri-canonical” texts are shown on the chart both to the left and the right of the core canon. On the left-hand side, the monastic rules (具足戒) form the hypothetical bridge between the core canon and the Vinaya literature. These rules are found with minor differences in both the Pātimokkha and the Suttavibhaṅga; the rules provided the basis for the subsequent elaboration of the Vinaya texts (shown in circles further to the left in this chart). The basic sequence and priority of development of these texts is not controversial, because it is demonstrable from evidence internal to the texts, i.e., how they are derived from one another (as shown on the chart, the Parivāra is the last addition, and thus displayed as the furthest from the core canon, with the Khandhaka being subsequent to the Suttavibhaṅga, etc.). [cf. Hinüber, 2000, p. 20–22] The old-fashioned view of this history allows several centuries for the elaboration of the Vinaya literature, with the rules providing a nucleus common to all schools of Buddhism as late as 250 B.C.; [Nyanatusita, 2008, p. 24–25 & fn. 11] both the details and the fundamental assumptions of this timeline are disputed in current scholarship.

The monastic rules had a peculiar priority both in shaping the Vinaya literature and also in re- shaping the sutta (經) literature into the form that is now extant; the concept of the sutta itself seems to have devolved from these rules [Nyanatusita, 2008, p. 50–54] and, I would add, there is a reciprocal relationship between one portion of the Vinaya literature and the (Dīgha Nikāya’s) Mahā-Parinibbāna-Sutta (analogous to the Chinese Dà Bō Niè Pán Jīng, 大般涅盤經). [See Obermiller, 1932] This is a recurrent feature of Theravāda scholarship: things that are separated in their origins are often united in their revisions.

Whereas the left-hand side of the chart shows an orderly progression, the right-hand side is fairly messy and confusing. To the right of the core canon (on the chart) is the fifth Nikāya, with a long list of texts that are considered part of the Theravāda canon in Sri Lankan orthodoxy today. I specify Sri Lanka (斯裡蘭卡) because a few texts are excluded by the Sri Lankans but are accepted as canonical in Myanmar (緬甸); these appear on the chart as “quasi-canonical”, grouped in a separate circle (to the right of the fifth Nikāya) with the first title being the Milindapañha.

Within the paradigm of Sri Lankan orthodoxy, the fifth Nikāya has a peripheral place and has been regarded with varying degrees of legitimacy in the history of Theravāda Buddhism. The Pali commentaries (of Buddhaghosa, written ca. 400–450 A.D.) record that many of these texts were rejected by certain schools of monks in (contemporaneous) Sri Lanka; reportedly, one school of monastic scholars (the reciters of the Dīgha Nikāya) rejected four of these texts as entirely non-canonical and relegated 12 of the remaining texts to the category of Abhidhamma literature. [Law, 1930, p. 176 & 195; cf. DPPN, s.v. Dīghabhānakā, cf. s.v. Cariyāpiṭaka] The existence of debates of this kind in antiquity reflects the fact that Buddhist scholars were struggling with their own categories to separate out the “quasi-canonical” elements of the Pali corpus, in their own terms.

It is outside the scope of this essay to review current philological debates about each and every one of the texts on the chart (although, I note, Hinüber’s [2000] Handbook attempts to provide such a survey); however, I would emphasize that the relative dating of these “peri- canonical” texts (as I have called them) most often relies on evidence internal to the texts themselves, and is thus non-controversial. For example, in Sri Lankan orthodoxy, the poems of the Jātaka (本生) are considered canonical, but not the prose: there is a clear contrast between verses (詩) and prose (散文) within the text and (therefore) this is not controversial (and is not offensive to anyone’s religion), even if the assumption that the layers of prose were later additions may now be debatable. If we accept this first distinction, we must (therefore) accept that a third text quoting the prose layer aforementioned is even later in its origins (and is thus “less canonical” by degrees). This is not merely hypothetical: it has been demonstrated (e.g.) with the last text in the fifth Nikāya shown on the chart, called the Cariyāpiṭaka. [DPPN, s.v. Cariyāpiṭaka] Obviously, if the Jātaka’s prose is non-canonical, it would be contradictory to define an even later text (that is partly derivative of this non-canonical source) as “canonical”. Contradictions of this kind exist within the fifth Nikāya, and thus, as I say, there were reasons for scholars of antiquity to consider these texts as less-than-entirely- canonical, without any intervention from modern researchers.

As one further example, I would note that another text in the fifth Nikāya, titled the Apadāna, allegedly quotes an Abhidhamma text called the Kathāvatthu; [DPPN, s.v. Apadāna] the latter text is traditionally ascribed to the era of Emperor Aśoka (阿育王, i.e., around 250 B.C.).9 The fact that this is a “traditional” ascription does not necessarily mean that it is true, however, it does mean that the status of the text was already debated by monastic scholarship in pre- modern history, and its origins were decided to be very much subsequent to the death of the Buddha (without meddling from modern scholars, and, thus, we cannot be called controversial for pointing out the same thing now). In this case, the derivation of one text from another then marks both as dubious, without controversy: if the Apadāna post-dates the Kathāvatthu, in what sense can it be called “canonical”? Although I am proposing this new category of the “peri-canonical”, my point here is that only the term itself is new: these texts were already marked-out as a sub-category (in some sense) by indigenous scholars, without any Western influence in the matter. It is neither novel nor offensive to affirm that (within the canon) some texts are more canonical than others.

In sum, inconsistencies of this kind are nevertheless consistent in marking the fifth Nikāya as a miscellaneous category at the periphery of the canon (whereas there is no such variation possible in the contents of the first four Nikāyas).

It is even more instructive to note that the Pali commentaries record an alternate method of organizing the entire canon by extending the fifth Nikāya to include both the Vinaya Literature and the Abhidhamma. [DPPN, s.v. Khuddaka Nikāya] If this method had prevailed, it would allow a simpler two-category view of the Theravāda canon, with the core canon (of four Nikāyas) clearly set to one side, and “everything else” put into this miscellaneous category of the fifth Nikāya. I surmise that this could have been unpopular because it demotes the status of the monastic rules somewhat (i.e., if we have only two categories, we would seemingly “lower” the Vinaya rules to the same level as collections of poetry and apocryphal storytelling).10 What is more fundamental is that the mere existence of this type of ambivalence toward the canon’s organization demonstrates that what we now think of as “the structure of the canon” has historically devolved from (changing) cultural attitudes toward the texts. Those attitudes are no longer susceptible to change because the Pali language itself is now dead, and remarkably few people are conversant in the literature.

I will not devote very much space to describing the third category of the quasi-canonical texts (可疑經典). The most difficult aspect to discuss is the structure of the Vinaya literature (on the left of the chart), and enough of this has already been explained above (in delineating the role of the monastic rules, etc.). It is less difficult but more contentious to speak of the Abhidhamma literature (on the right-hand side of the chart); this corpus of texts is simply and bluntly acknowledged by all specialized scholars as the least canonical of additions to the canon and, in relative chronology, it was added last. [Law, 1930, p. 183; cf. DPPN s.v. Abhidhamma Piṭaka] Hinüber remarks that, “The language of the Abhidhamma texts is clearly distinct from the usage found in the first two Piṭakas.” [Hinüber, 2000, p. 68] Evidence of this kind is obvious, but only for the tiny minority of people who work with primary sources.

Categorizing the Theravāda Abhidhamma as quasi-canonical is only “contentious” for two reasons that are worth mentioning (briefly) because they directly interfere with fieldwork in Southeast Asia today. The first is that the Abhidhamma texts are revered as magical in a wide range of cults and cultures of Southeast Asia, and almost none of these rituals involve reading comprehension of the texts themselves.11 Very few people in Southeast Asia think of the Abhidhamma as a corpus of palpable texts; for most, the word suggests an impalpable ideal. Secondly, the term Abhidhamma is contentious simply because the word itself is now so commonly misused (and misunderstood) that people frequently get confused and take offense no matter how carefully we may speak about the subject.

As I’ve noted on the chart, we have a similar problem with the word Jātaka today, but whereas contemporary Buddhists treat the Jātaka as a genre (defined aesthetically) their idea of the Abhidhamma is instead supernatural and rather nebulous. In a sense, there is nothing surprising about a story that resembles the style of a Jātaka being called by this same term, i.e., not even if the person telling the story knows that it was first written in medieval Thailand, with no connection to ancient India whatsoever. Indeed, it is not offensive to clarify that one text is a locally-written Jātaka (of some relatively-recent century), whereas another is considered “canonical” in origin, because the term is now informally used to mean a genre that is inclusive of the apocryphal and the ancient alike. By contrast, one Cambodian professor was very much offended when I explained to him that a text he regarded as “Abhidhamma” was, in fact, first written in 20th century Myanmar and had never existed in any language more ancient than modern Burmese (緬甸語). He was surprised because he had already been translating this text from English into Cambodian for several years. He had assumed that he was translating the canonical Abhidhamma, but he had no clear notion of what the word is supposed to mean. Both Jātaka and Abhidhamma texts, I would note, are performed as part of funeral rituals throughout mainland Southeast Asia; this is one very simple reason why they are taken much more seriously here than elsewhere (i.e., regardless of their contents).

In contrast to the tremendous reverence I’ve just described, many of the western scholars who devoted their time to the Abhidhamma literature left castigating remarks against the texts. After some 40 years of work of this kind (by her own estimate) C.A.F. Rhys-Davids described the European efforts to research the Abhidhamma as a hunting expedition that only captured “a bag of mice”; she repeatedly asks whether or not this entire venture had been a “waste of time”. [Rhys-Davids, 1923, p. 245–6] She does also remark that the study of the Abhidhamma (as a corpus of texts entirely subsequent to the suttas of the core canon) demonstrated some interesting intellectual developments in contrast to the earlier canon; Rhys-Davids calls these later developments a new “discipline in consistency of thought and language”, and also “a kind of intensive and introspective growth” that took place “in the course of perhaps a few centuries”. [Ibid., p. 247] However, her evaluation of the Abhidhamma literature as a whole was of a “cenotaph of the works of a dead culture”. [Ibid. p. 245] She clarified her meaning by drawing attention to the etymology of this word “cenotaph”: she was not merely reviling it as a tomb, but as an empty tomb. [Id.]

In discussing the Abhidhamma we have a clear contrast between the attitudes of modern scholars and (ongoing) Buddhist tradition. For the purposes of modern scholarship, these are blatantly non-canonical texts that were (somehow) added to the canon after-the-fact. While we have commentarial records of dissent about the canonicity of these sources within Theravāda monastic tradition (similar to the foregoing discussion of the fifth Nikāya), the supernatural myth that is offered as an excuse for including these texts in the canon is itself an invention of an even later era (i.e., the myth itself first appears in the commentaries, and has no basis in the canon). [Hinüber, 2000, p. 66] The problem here is not the contrast between myth and reality, but the contrast between one era of mythology and another. Nobody who actually reads these texts can accept a theory that they are of simultaneous origin with the rest of the canon; it hardly matters if the theory is supernatural or not.

Without digressing further, I would note that the Abhidhamma is also the most variegated in comparing each version of the canon to the others:

“It may be noted that this Sanskrit canon contained seven Abhidharma texts, just as the Pali canon contains seven Abhidhamma texts. But these two sets of texts seem to be entirely unrelated. They are not variants from one common tradition but independent compositions.” [Clark, 1930, p. 137]
“Further, [the Chinese canon] contains seven Abhidharma texts, but these seem to differ in toto from the Pali Abhidhamma texts…” [Ibid., p. 127]

In my opinion, this asymmetry was made much more important by relatively late scholars such as Xuán Zàng (玄奘) whose translations both preserved and promoted (non-Theravāda) Abhidharma texts in China (in the seventh century A.D.). Resultantly, works like the Abhidharmakośa (that have no analogy in the Theravāda canon) became popular in China (and were perceived as important components of the canon there). It is an overt fact that the text just mentioned was written by Vasubandhu (世親, a.k.a. 天親) in the 4th century A.D., with no possible connection to the historical Buddha; from a Theravāda perspective, it is a “pseudo- Abhidhamma” treatise; conversely, in light of all the facts we have reviewed above, the whole of the Theravāda Abhidhamma could be described as “pseudo-canonical”. The proposed term quasi-canonical would encompass all of these texts that are overtly subsequent to the canon, and yet are (incongruously) included within it; this allows us to avoid statements about which texts are “pseudo-” from one perspective or another.

The main advantage of the foregoing “practical guide” to the canon is, in fact, inverse to all that I have said: while I have summarized historical distinctions that researchers should understand before fieldwork is undertaken, this also enables them to disregard many (spurious) historical distinctions that they will encounter both in print and in speech.

At present, Theravāda scholarship of every kind is extremely lenient about pseudo-historical claims that are offered about texts found within (what I have called) the core canon. Claims about specific texts (and even specific concepts) being “later” and “earlier” are made very lazily and informally, often without proper citation; learning when to disregard such claims (and when to take them seriously) is a science unto itself. As a remarkable example (from a remarkable scholar) G.P. Malalasekera contradicted his own better research in suggesting that King Yama (閻王, a.k.a. 閻羅) only performs judgement on the dead in “later literature”; [Malalasekera, 1967, p. 85] on the contrary, the same author’s own article (written decades earlier!) informs us that King Yama appears in the most ancient part of the Theravāda canon. [DPPN, s.v. Yama] There can be no debate whatsoever that King Yama appears as the judge of the dead (and the king of hell) in the most ancient stratum of Buddhists texts that exist in any canon (i.e., what I have called the core canon);12 nevertheless, in recent generations it has been routine for scholars to propose pseudo-historical distinctions of this kind in passing (without proof or even citations) —even if they know better themselves.

With all of these warnings having been offered (in the spirit of practical advice) I would reiterate that my message here remains positive: as with the example of the name Dharaṇi (§4, above) there are innumerable useful applications of canonical materials to fieldwork, and what we discover in fieldwork should inspire new inquiries into the canon as well. Unfortunately, this is not in vogue.


8. 在南傳研究中,DPPN是一個標準的縮寫 (即少數作者所採用的傳統); 是一本經常用到的參考書《巴利語專有名詞詞典》(Dictionary of Pali Proper Names)標題的首字母縮略詞。DPPN is a standard abbreviation in Theravāda studies (i.e., a convention used by a small number of authors); it is an acronym from the title of this often-used reference work, the Dictionary of Pali Proper Names.

9. 我假定這位帝王卒於西元前232年,於不到30年前開始統治,約在西元前268年。(任何關於這方面的爭論都不包括在此一文章範圍內) 。I am presuming the emperor’s death to be 232 B.C., and his reign to commence less than 30 years earlier, perhaps in 268 B.C. (any debate on this point is outside of the remit of the current article).

10. 再明確一些:如果評述確實允許認可第五部《尼柯耶經》(Nikāya)的兩種理解同樣有效(一種是“狹義”理解,另一種是對“尼柯耶經”一詞本身“更廣義”地使用)[根據“律法”,1930, 第184頁] 且如果在當時作者所處時代為一種被廣泛持有的觀點(約西元 400–450 年, 如上所述),那麼我們可以推測,其中一種理解因為當時的不流行而衰落(只剩下評述中的這一線索,如 Sumaṅgalavilāsinī 等等 )。To be more explicit: if the commentaries did allow both interpretations of the fifth Nikāya as equally valid (with one being a “narrow” interpretation and the other being a “wider” use of the term Nikāya itself) [as according to Law, 1930, p. 184] and if this had been a widely-held view during that era of authorship (ca. 400–450 A.D., as aforementioned) then we could surmise that one of the two interpretations has waned away due to its unpopularity since that time (leaving only this trace in commentaries, such as the Sumaṅgalavilāsinī, etc.).

11. 許多這些信仰與文本的任何內容均無關聯,並常象徵性地借用《經藏》,例如,通過對在任何語言中並無意義的魔法咒語的背誦。賈斯汀麥克丹尼爾(Justin McDaniel)發表了關於在老撾和泰北文化中的這個主題。[McDaniel, 2008, 尤其第8章]Many of these beliefs have no connection whatsoever to the content of the texts, and often invoke the Abhidhamma symbolically, e.g., through the recitation of magical formulas that are not meaningful in any language. Justin McDaniel has published on this subject in Lao and Northern Thai cultures. [McDaniel, 2008, esp. ch. 8]

12. 眾多例子之一,見 《天使經》(Devadūta sutta), 《中尼柯耶經》(Majjhima Nikāya)第130部經 (PTS MN, 卷3, 第178頁及以後) 。For one example among many, see the Devadūta sutta, the 130th sutta of the Majjhima Nikāya (PTS MN, vol. 3, p. 178 et seq.).