中華佛學學報第07期 (p361-384)： (民國92年)，臺北：中華佛學研究所，http://www.chibs.edu.tw/
Chung-Hwa Buddhist Studies, No. 07, (2003)
Taipei: The Chung-Hwa Institute of Buddhist Studies
ISSN: 1026-969X

Issues in the Use of Electronic Markup for the

0. Āgama Literature

Of the two main collections of Āgamaliterature──in Pāli and Chinese──the Pāli Nikāyas have for variousreasons attracted a lot more scholarly attention during the past century.　Only the Pāli texts have beentranslated into English, and it is these that are generally held to be the “theearliest texts.” To this day, someare even convinced that the Chinese Ahan jing are mere translations of the Pālias it is found in the PTS edition.

However, the differences between thetwo corpora do not suggest that one is a translation of the other.　They are rather the result ofdifferent lines of transmission in the Buddhist order.　Both are the product of long processesof canonization and editing that came to an end (of some sort) only after the 5thcentury CE.　By that time thetranslation of most of the Chinese Ahan jing from the Prakrit Āgamas ofNorthern India had been completed and the Pāli Nikāyas had reached a new degreeof stability due to Buddhaghośa’s commentaries.

Almost hundred years ago, in order togive a western reader an idea of how large the differences between the twocollections are, Akanuma wrote: “if these deviations are not of the kind thatwe find in the four synoptic Gospels, or of such degree as those between theGospels and the Apocrypha, they are nevertheless more than the various readingsof Shakespeare in the Quarto and Folio present.”[1] Statements like this, and the assurancethat the doctrines are basically the same, may have lead to the impression thata comparative study would yield few relevant new insights.　 Perhaps this is true.　It all depends on what one believes tobe “relevant.”

Relevancy is also a central feature ofelectronic markup, the meta-data　

p. 364

1. Markup

Markup is the data that can, or rathermust, be added to an electronic text to ensure its usefulness in the digitalmedium.　 In order totransfer a text from paper, one needs to encode several features of thetext-on-paper which are usually taken for granted.　Merely to reproduce the words is notenough.　Paragraphs are notthere to make a page look nice, but are a conscious effort by an author oreditor to encode information regarding the structure of the text.　In a manuscript the hand of thescribe, the quality of paper, etc.　might be worthmentioning.　Therefore,standards have evolved that allow us to encode almost every kind of informationconcerning any level of a given text.　HTML is one ofthese standards, XML (eXtensible Markup Language) another.

For the examples given here we willuse the XML-conformant standard developed by the TEI (Text EncodingInitiative)-consortium (edition P4).

There are different standards, andeven within the same standard there is always more than one way to markup atext.　The designs used inthe present paper are meant as illustrations and do not suggest a final markupsolution for the problems in question.　They are ratherpreliminary suggestions toward a digital comparative edition of the ChineseAhan jing and the Pāli Nikāyas.

2. Why use Markup for Comparative Analysis?

We are rapidly moving into an erawhere the printed text will become a snapshot version of a constantly evolvingdigital text.　 Texts, especiallyscholarly texts, will not anymore be produced with the primary aim ofpublishing them in print.

p. 365

Markup is not only a sine qua nonin the process of digitization, but also a useful research tool for textualstudies.　 For comparativetextual studies of the kind the Āgama literature demands, where we have to dealwith large text corpora and a complex edition history, the use of markup forcomparative analysis offers several advantages.

2. It is flexible.　The researcher can encode comparativeinformation in one of the corpora (e.g. encoding Pāli equivalents in theChinese text), in both (by using various linking or alignment methods), or in athird place outside the texts (e.g. as in my digital comparative catalog ofPāli and Chinese sutras [ComCat]).

The researcher decides what and how toencode (in the framework of the standard) according to his/her aims.　The use of a standard makes it easier forresearchers to communicate with each other, work together, evaluate andstreamline their different approaches.

Errors caneasily be corrected, and different opinions and interpretations can be added to any one edition, all the whilekeeping track of all changes by various ways of version-control.

p. 366

3.What to encode?──Two examples

Here, we will, in two examples,consider the encoding of two different areas in comparative analysis:occurrence and content structure.　 Other importantfields, such as the markup of linguistic features, will have to wait untilanother occasion.

3.1. Occurence

Occurrence denotes the existence of acertain passage in a text.　In the followingexample, we will consider a gāthā found in the Za ahan jing (雜阿含經, T02n0099)in volume two, page 120b of the Taishō edition.　Thereis a closely corresponding gāthā in the Pāli canon in volume four, page 218 ofthe PTS (Pāli Text Society) edition.　The topic is ametaphor that likens the arising of sensations (vedanā) in the body to winds inthe sky.

We now want to encode informationabout which passages occur in both the Chinese and the Pāli version, and whichoccur only in one version.　For the sake ofsimplicity, we will ignore the problem of variant readings as they are given inthe apparatus of both editions and treat the passages as they appear in themain text.

Here is the Pāli text.　The underlined passages do not occurin the Chinese:

Yathāpi vātā ākāse// vāyanti vividhā puthū //

puratthimā pacchimācāpi // uttarā atha dakkhiṇā //

Sarajā arajā capi// sitā unhā ca ekadā //

adhimattāparittā ca// puthu vāyanti mālutā //

tathevimasmij kāyasmiṃ // samuppajjanti vedanā //

sukhadukkhasamuppatti// adukkhamasukhā ca yā //

yato ca bhikkhuātāpī // sampajaññaṃ nirūpadhi //

tato so vedanāsabbā // parijānāti paṇdito //

p. 367

So vedanā pariññāya// diṭṭhe dhamme anāsavo//

kāyassa bhedā dhammattho // sankhyaṃ nopeti vedagūti //

In the XML file ofthe CBETA (Chinese Buddhist Electronic Text Association) CD we find the Chinesetext together with some useful basic markup.　Underlined passages do not exist in the Pāli.

<lb n= “0120b26”/><lg><l>譬如虛空中</l><l>種種狂風起</l >

<lb n= “0120b27”/><l>東西南北風</l><milestone n=“0120b271b”/><l>四維亦如是</l>

<lb n= “0120b28”/><l>有塵及無塵</l><l>乃至風輪起</l>

<lb n= “0120b29”/><l>如是此身中</l><l>諸受起亦然</l>

<pb ed= “T” id= “T02.0099.0120c” n= “0120c”/>

<lb n= “0120c01”/><l>若樂若苦受</l><l>及不苦不樂</l>

<lb n= “0120c02”/><l>有食與無食</l><l>貪著不貪著</l>

<lb n= “0120c03”/><l>比丘勤方便</l><l>正智不傾動</l>

<lb n= “0120c04”/><l>於此一切受</l><l>黠慧能了知</l>

<lb n= “0120c05”/><l>了知諸受故</l><l>現法盡諸漏</l>

<lb n= “0120c06”/><l>身死不墮數</l><milestone n= “0120c06lb”/><l>永處般涅槃</l></lg>

What has beenencoded here are linebreaks <lb>s, one pagebreak <pb> (in fact acolumn break between 120b and 120c), lines <l> in the gāthā and the gāthāas a whole has been taken as a linegroup <lg>.　I have added two <milestone>elements, which we will need later.

Especially useful for us are theidentifying numbers <n> that enable us to adress each line.　It saves some work if there is alreadya basic markup structure with addressable identifiers that one can refer to.　If there is none, one can alwayscreate one’s own (TEI-conformant) structure. In English the text reads:

p. 368

p. 369

Now that the material is ready, let usconsider: what are the logical possibilities of “occurance” for the purpose ofcomparative analysis? Clearly, there are three.

Translatedinto TEI this set of possibilities could be expressed like this:

<interpgrp type=“occurrence”>

<interp id= “oc”resp= “MB” value= “Only in Chinese”>

<interp id= “op”resp= “MB” value= “Only in Pali”>

<interp id= “chpa”resp= “MB” value= “In Chinese and Pali”>

</interpgrp>

This is a group ofinterpretive tools <interpgrp>.　It assigns anidentifier <id> to each possibility.　 It also says thatsomeone with the initials of Mister Bean is responsible for the interpretation.

Now we can startwith the markup.　Here, we willinsert the markup in the Pāli text, but we could as well put it in the Chinesetext or in an altogether different location.

<seg ana=“chpa”> Yathāpi vātā ākāse, vāyanti vividhā puthū

puratthimā pacchimācāpi, uttarā atha dakkhinā.

<xptrana= “oc” doc= “T02n0099.xml” from= “0120b271b” to= “0120b28”/>

sarajā arajācapi,</seg><seg ana= “op”>sitā uṇhā ca ekadā

adhimattā parittāca,</seg><seg ana= “chpa”>puthū vāyanti māluta.

tathevimasmij kāyasmij, samuppajjanti vedanā

p. 370

sukhadukkhasamuppatti,adukkhamasukhā ca yā

<xptr ana= “oc”doc= “T02n0099.xml” from= “0120c02” to= “0120c03”/>

yato ca bhikkhuātāpī sampajaññaṃna riñcati.

tato so vedanāsabbā parijānāti paṇdito.

So vedanāpariññāya, diṭṭhedhamme anāsavo;

kāyassa bhedā</seg><seg ana= “op”>dhammaṭṭho</seg>,<seg ana= “chpa”>saṇkhyaṃ nopeti</seg><seg ana= “op”>vedagūti</seg>

<xptrana= “oc” doc= “T02n0099.xml” from= “0120c06lb” to= “0120c07”/>

Here we are sayingthat the segment <seg> from “Yathāpi” to “arajā capi” has the analyticalvalue (ana) “chpa”.　In our<interpgrp> above we have defined “chpa” as denoting a passage thatexists in both Chinese and Pāli.　The followingsegment sitā unhā caekadā adhimattā parittā ca appears only in the Pāli version, therefore it has the analyticalvalue “op.”

We indicate the existence of a passagethat appears only in Chinese by using an extended pointer <xptr> thatidentifies a passage in another document (doc) by using the (from) and (to)attributes.　 With the (ana)attribute we express that this passage exists only in Chinese “oc.”[6]

If we do not want to add theinformation into the texts themselves, we can use the <span> element toanalyse the two versions.　 This results in alist of <span> elements that can be attached to the text documents orstored in a different place.　The items on thelist would look like this:

<span value= “chpa” from= “IDREF for yathāpi” to=“IDREF for arajā capi”/>

<span value=“op” from= “IDREF for sitā” to= “IDREF for parittā ca”/>

<span value=“oc” from= “0120c02” to= “0120c03”/>

p. 371

The above examplewas to illustrate how the occurrence of shorter passages in both or eitherversion of our texts can be encoded.　While occurrencecan be encoded on the level of single words and passages it is also possible tostate occurrence for larger units.　To say that acertain sutra exists in both Pāli and Chinese can already be taken as aninstance of “chpa.” In many sutraslarge, distinct parts exist only in either Chinese or Pāli.　These could be marked with the samemechanism.　 But there isclearly more to do in a comparison of the two corpora.

3.2. ContentStructure

Similarity and difference in thecontent structure of related sutras is a relevant phenomenon for comparativeanalysis.　We need a mechanismto describe differences on this level, which is considerably more complex thanthat of occurence.　In order to accomplishthis, one has to define a set of structural components.　Again, what kind of structures areencoded depends on one’s interests and research aims.

In the following a tripartite contentstructure is assumed.　The beginning of asutra often outlines a “setting” that describes the where, who and why of thesutra (this could be further divided into “location”, “protagonists” and“speech instance” components).　Also, most Āgamasutras have a “topic,” which is usually, but not necessarily, a questionanswered by the Buddha.

Finally, let us call “argument” theargumentation as it is developed in its logical structure and the ideas andconcepts employed.

Comparing a Pāliand a Chinese sutra[7] one might want to say that the one orthe other of these structural parts are the same or different.　Expressed in TEI as <interpgrp>this becomes:

p. 372

<interpgrp type=“content similarity”>

<interp id= “sameset”resp= “MB” value= “same setting”/>

<interp id= “sametop”resp= “MB” value= “same topic”/>

<interp id= “samearg”resp= “MB” value= “same argument”/>

<interp id= “differentset”resp= “MB” value= “different setting”/>

<interp id= “differenttop”resp= “MB” value= “different topic”/>

<interp id= “differentarg”resp= “MB” value= “different argument”/>

</interpgrp>

With these tools,however, we can so far only produce statements of occurrence on a contentstructure level.　This does not takeus much beyond the previous example, where we could say about a <seg>that it existed in both versions or only in one.

If we want to gain a new take onstructural sameness and difference of content we will have to explore furtherwhat possible differences between sutras exist.

What then, canhappen, assuming there are setting, topic and argument? The logical possibilities are that two sutras have:

“same setting, sametopic, same argument”

“same setting, sametopic, different argument”

“same setting,different topic, same argument”

“samesetting, different topic, different argument” [not useful when one comparesPāli and Chinese][8]

p. 373

“different setting,same topic, same argument”

“different setting,same topic, different argument”

“different setting,different topic, same argument”

“differentsetting, different topic, different argument” [not useful, this is tantamountto saying the sutras are not related; comparison requires at least some form ofsimilarity]

If we include only the useful combinations into our <interpgrp> we get amore sophisticated scheme.　The <id>s canbe choosen freely by the encoder.

<interpgrp type=“content similarity”>

<interp id= “sameset”value= “same setting”/>

<interp id= “sametop”value= “same topic”/>

<interp id= “samearg”value= “same argument”/>

<interp id= “differentset”value= “different setting”/>

<interp id= “differenttop”value= “different topic”/>

<interp id= “differentarg”value= “different argument”/>

<interp id= “CS1”value= “same setting, same topic, sameargument”/>

<interp id= “CS2”value= “same setting, same topic, differentargument”/>

<interp id= “CS3”value= “same setting, different topic, sameargument”/>

<interp id= “CS4”value= “different setting, same topic, sameargument”/>

<interpid= “CS5” value= “differentsetting, same topic, different argument”/>

<interpid= “CS6” value= “differentsetting, different topic, same argument”/>

</interpgrp>

We can now not onlyencode if the topic of two (or more) sutras is the same or different.　We can also express which out of six types of content　

p. 374

Why would anyone want to know this?Simply, because this kind of structural comparison would lead to objective,quantitative data that could be used to gain new insights in the formation ofthe early Buddhist canon.　If it can be shown,for instance, that in a sizeable number of Pāli-Chinese sutra pairs that havethe same setting and topic there are differences in argumentation (“CS2”),this would constitute further evidence that during the centuries of oraltransmission the connection between topic and argument was especiallysusceptible to change.

To illustrate this the Arrow Sutta[9] will serve as example.　All my comments in [ ] could beexpressed with some form of markup.　 Merely for the sakeof simplicity I will keep the focus on content-structure.　 This does not require any markup inthe text (except <id>s).　I will show at theend how to encode our conclusion about the content structure in a simple way.

[1. Part: Setting and Topic]

Thus have I heard once the Buddha wasstaying at Rājagaha in the Karanda Bamboo Grove.

At that time he told a number ofbhikkhus: In an ignorant, untaught average person there arise pleasant,unpleasant and neutral vedanā [sensations].　Alsoin a well taught saintly follower, there arise pleasant, unpleasant and neutralvedanā.　What, bhikkhus, isthe difference between　

p. 375

The Bhikkhus said: The World-honoredOne is the root, the eye, the garment of the Dhamma, verily, if only theWorld-honored One were to elaborate on this, we would listen, remember andrevere [the teaching].

The Buddha told the bhikkhus: Whensensations arise in the body of the ignorant, untaught average person, whenthey are harrowed by pain and suffering, as if it were going to kill them.　They sob and cry sorrowfully, andloudly exclaim their anguish.

[1.3 might be a mistake in the text.　The passage is the same as the nextsection, but without the introduction (“Listencarefully……”).　 It looks like anunintentional reduplication of the following, perhaps a scribal error.　(I could encode this interpretation.　If later someone finds reason to thecontrary he / she could correct me, by making use of the same markupmechanism.)]

Assutavā,bhikkhave, puthujjano sukhampi vedanaṃ vediyati, dukkhampi vedanaṃ vediyati,adukkhamasukhampi vedanaṃ vediyati.　Sutavā, bhikkhave, ariyasāvakosukhampi vedanaṃ vediyati, dukkhampi vedanaṃ vediyati, adukkhamasukhampi vedanaṃ vediyati.　Tatra, bhikkhave, ko viseso ko adhippāyoso kiṃ nānākaraṇaṃ sutavatoariyasāvakassa assutavatā puthujjanenāti? Bhagavaṃmūlakā no, bhante, dhammā…… la ……

[As we see, 1.3 does not appear inPāli.　The restcorresponds exactly　

p. 376

[2. Part: Sensations in theuntrained person. Unique passages in bold font.]

The Buddha told the bhikkhus: Listencarefully and consider well.　I will explain itto you, oh bhikkhus.　When ignorant,untaught average persons come in touch with the arising of the varioussensations, they add to their pains and sufferings, as if they were going tokill them.　They sob and crysorrowfully, they loudly exclaim their anguish, and their minds get confused.　At these times their sensations becometwofold, physical and mental.

Assutavā,bhikkhave, puthujjano dukkhāya vedanāya phuṭṭho samāno socatikilamati paridevati urattāḷīkandati sammohaṃ āpajjati.　So dve vedanā vediyati ── kāyikañca, cetasikañca.

[For the level of this analysis, thisis the same.]

p. 377

This is like in a man whose body isstuck with two poisoned arrows in whom there arises intense pain.　An ignorant, untaught average person,having compounded his sensations to two layers, suffers intense pain on boththe level of physical and mental sensations.

Why is that? It is because theignorant, untaught average person does not understand that out of the fivesense-desires [五欲 pañcakāma?] he comes in touch with pleasant sensations.　By these pleasant sensations heexperiences the pleasures of the five senses [五欲樂 pañcakāmasukhā?].　 And because heexperiences these pleasures one becomes afflicted by the tendency to crave [為貪使所使].　When he experiences unpleasantsensations, then because of this there arises aversion.　With this aversion one becomesafflicted by the tendency to hate [為恚使所使].

He does not well understand theformation and the dissolution, the sweetness[10], the misery andthe abandonment of these two kinds of sensation.　Becausehe does not well understand this, when neutral sensations arise, one becomesafflicted by the tendency towards ignorance [為癡使所使avijjānusayo soanuseti].　Thus he is bound by pleasant sensations, not able to let go of them; bound by unpleasant sensations, not　

p. 378

[ThePāli version is somewhat different:]

Seyyathāpi,bhikkhave, purisaṃ sallena vijjheyyuṃ.　 Tam enaṃ dutiyena sallena anuvedhaṃ vijjheyyuṃ.　Evaṃ hi so, bhikkhave, puriso dve sallena vedanaṃ vediyati.　Evaṃ eva kho, bhikkhave, assutavā puthujjano dukkhāya vedanāya phuṭṭho samānosocati kilamati paridevati urattāḷiṃ kandati sammohaṃ āpajjati.　So dve vedanā vediyati ── kāyikañca,cetasikañca.　[This says the same as the Chinese.　 Not sothe following:]

Tassāyeva khopana dukkhāya vedanāya phuṭṭho samāno paṭighavā hoti.　Tam enaṃ dukkhāya vedanāya paṭighavantaṃ, yo dukkhāyavedanāya paṭighānusayo, so anuseti.　So dukkhāya vedanāya phuṭṭho samānokāmasukhaṃ abhinandati.　Taṃ kissa hetu? Nahi so, bhikkhave, pajānāti assutavā puthujjano aññatra kāmasukhā dukkhāyavedanāya nissaraṇaṃ, tassa kāmasukham abhinandato yosukhāya vedanāya rāgānusayo so anuseti.

[PTS -Translation:]

Touched by the painful feeling hefeels repugnance for it.　Feeling that repugnance for the painful feeling, the lurking tendency to repugnance fastens on him [patighānusayo so anuseti[11]].　Touched by the painful feeling, hedelights in pleasant feeling.　Why so? The untaught manyfolk,　

p. 379

So tāsaṃ vedanānaṃ samudayañcaatthaṇgamañca assādañca ādīnavañca nissaraṇañca yathābhūtaṃ nappajānāti.　Tassa tāsaṃ vedanānaṃ samudayañca atthaṇgamañca assādañca ādīnavañca nissaraṇañca yathābhūtaṃ appajānato,yo adukkhamasukhāya vedanāya avijjānusayo so anuseti.　So sukhaṃ ce vedanaṃ vediyati, saññutto naṃ vediyati.　Dukkhaṃ ce vedanaṃ vediyati,saññutto naṃ vediyati.　 Adukkhamasukhaṃ ce vedanaṃ vediyati,saññutto naṃ vediyati.　Ayaṃ vuccati, bhikkhave,assutavā puthujjano saññutto jātiyā jarāya maraṇena sokehi paridevehidukkhehi domanassehi upāyāsehi, saññutto dukkhasmāti vadāmi.

[The argumentationin this last part is again identical with the Chinese.]

[Here the Chineseand the Pāli both try to explain of how the experience of vedanā by an averageperson leads to bondage.　They differ,however, in their argumentation.　The way craving andaversion arise is explained differently. The Chinese states it relatively straightforward:

Sensual Desire -> pleasant sensations -> craving

Unpleasant sensations -> aversion

Neutral sensations -> ignorance

The Pāli makes an elegant little loop:

Unpleasant sensations -> aversion ＆ delight in sensual pleasure

Delightin sensual pleasure -> craving

Neutral sensations -> ignorance

Though both versions converge again in the conclusion, we take this to be an instance of “different argument”.]

p. 380

[3. Part: Sensations in the noble disciple]

Now if there arises pain in body ofthe learned, noble disciple, if he is harrowed by great pain, as if it weregoing to kill him, he does not give rise to exclaim his anguish full of griefand sorrow, does not sob and cry, and does not become mad and wild.　Because at this time there only arisesone [kind of] sensation, namely physical sensation and not mental sensation.

Sutavā ca kho,bhikkhave, ariyasāvako dukkhāya vedanāya phuṭṭho samāno na socati, nakilamati, na paridevati, na urattāḷiṃ kandati, na sammohaṃ āpajjati.　So ekaṃ vedanaṃ vediyati – kāyikaṃ, na cetasikaṃ.

[This is the same.]

Like a man struck by only one arrow,not by two.　 He at that time hasonly one [kind of] sensation.　What is called thephysical sensation does not give rise to mental sensations.

When he comes in touch with pleasant sensations, he is not defiled by the desire for pleasure.　Because he is not defiled by thedesire for pleasure one does not become afflicted by the tendency to crave for these pleasant　

p. 381

Seyyathāpi,bhikkhave, purisaṃ sallena vijjheyya.　 Tamenaṃ dutiyena sallenaanuvedhaṃ na vijjheyya.　Evañhi so, bhikkhave, purisoekasallena vedanaṃ vediyati.　Evameva kho, bhikkhave, sutavāariyasāvako dukkhāya vedanāya phuṭṭho samāno na socati, na kilamati, naparidevati, na urattāḷiṃ kandati, na sammohaṃ āpajjati.　So ekaṃ vedanaṃ vediyati –kāyikaṃ, na cetasikaṃ.

Tassāyeva khopana dukkhāya vedanāya phuṭṭho samāno paṭighavā na hoti.　Tamenaṃ dukkhāya vedanāya appaṭighavantaṃ, yo dukkhāy avedanāya paṭighānusayo, so nānuseti.　 So dukkhāya vedanāya phuṭṭho samāno kāmasukhaṃ nābhinandati.　 Taṃ kissa hetu? Pajānāti hi so, bhikkhave, sutavā ariyasāvako aññatra kāmasukhā dukkhāya vedanāya nissaraṇaṃ.　 Tassa kāmasukhaṃ nābhinandato yo sukhāya vedanāya rāgānusayo, so nānuseti.　So tāsaṃ vedanānaṃ samudayañca atthaṇgamañca assādañca ādīnavaṃ ca nissaraṇañca yathābhūtaṃ pajānāti.　Tassa tāsaṃ vedanānaṃ samudayañca atthaṇgamañca assādañca ādīnavañca nissaraṇañca yathābhūtaṃ pajānato, yoadukkhamasukhāya vedanāya avijjānusayo, so nānuseti.　So sukhañce vedanaṃ vediyati, visaññutto naṃ vediyati.　Dukkhañce vedanaṃ vediyati, visaññutto naṃ vediyati.　Adukkhamasukhañcevedanaṃ vediyati, visaññutto naṃ vediyati.　Ayaṃ vuccati,bhikkhave, sutavā ariyasāvako visaññutto jātiyā jarāya maraṇena　

p. 382

[Both the Pāli and the Chinese versionfollow their respective lines of argumention.]

[4. Part: Thefourth part in both sutras consists in a gāthā.　Somepassages are related, but the original of the Chinese must have differedconsiderably from the Pāli.　There are only twostanzas in Chinese versus three in Pāli.　Gāthās in theirsūtra contexts pose a number of special problems for comparative research,which I will skip here.]

The analysis of content structure, aswe have approached it above, concerns the text as a whole.　It is therefore not necessary to splitthe texts in segments <seg>.　What we can do hereis to simply link both texts with a <link> element and allow for ananalytical attribute like (ana) to carry our interpretation: “same setting,same topic, different argument” (“CS2” in our <interpgrp> above).

<link type=“contentstructure” targets= “ID chinese sutra _ID pali sutra” ana= “CS2”resp= “we”>

A number of these links would be aneffective way to refine the existing catalogues[13] that relate thePāli and the Chinese Tripitaka.　The use of markupallows us to bring structural analysis to new levels of precision andflexibility.　Especially whenlarge text corpora are concerned, the advantages of the digital medium aresubstantial.　 Where there is a　

p. 383

Using markup may not be everybody’scup of tea, but it has a number of solid advantages over the usual way toencode one’s research in the form of annotated essays or translations.　In an annotated, printed treatment ofa text the author’s strategy of what and why s/he annotates and if consistantlyso, is rarely transparent.　The formalism ofmarkup grammar will of course never replace the free flow of the essay, but itcan certainly provide us with new helpful tools for textual research,especially in the field of Buddhist Studies.

p. 384

漢巴經典之比較性分析──略說電子標記的問題

提要

本文是邁向創造一個先進的巴利經典與漢文阿含經電子版的第一步。以比較性目錄連接巴漢兩個文獻匯編之後，研究者可以TEI標記符號（markup）記錄其研究結果。透過型式的標記語言，電子版可以標準化且精確的方式告訴讀者某兩個經典的關係。由研究者自己決定記錄的內容、使用的標記語言及標記語言的表達方式。

本文1.,2.節敘述用標記符號的原因。接著以兩個例子在3.1及3.2說明比較性研究如何運用標記符號。

3.1說明如何使用標記讓所謂的「出現」（occurrence）概念形式化。「出現」代表某一個字、詞或句子是否僅在巴利文或僅在漢文或在兩種文獻皆出現。這種分析，舉例而言，可讓巴利文的讀者知道某一句巴利文在漢文有沒有平行的內容。說明此方法的例子源於《雜阿含經》（大2.99.120b）及相應部（PTS IV, 218）。

3.2節介紹「內容結構」（content structure）的概念。設計一個敘述內容結構差異的矩陣後，我們可以簡單的以一行標記符號來表現某兩個經典有何種差異。此研究方法可協助研究者從兩經結構性的差異獲得較具體的標誌，以探討巴漢文獻匯編的相同與相異之處。

關鍵詞：1.阿含經　2.巴利文阿含經　3.比較性分析　4.電子佛典　5.標記

[1] Anesaki, Masaharu: TheFour Buddhist Āgamas in Chinese ── A Concordance of their Parts and of the Corresponding Counterpartsin the Pāli Nikāyas.　 (1908), p. 2.

[2] I am grateful toDouglas Gildow for suggesting this translation.

[3] I take 有食與無食 to be sāmisā nirāmisā ca (lit.　withflesh and without flesh) in Pāli.　The term sāmisāwhen applied to vedanā denotes the vedanā that arise through contact with sensualobjects, thereby being defiled.　The sensations experienced in thejhānas, however, are said to be nirāmisā, undefiled.　See the paper by the Vipassana Research Institute: “Sāmisa andNirāmisa in Meditation” in The Importance of Vedanā and Sampajañña.　Nashik (Mahārashtra, India): VRI, 1990 [Reprint 2002], 53~55.

[4] Here the PTS main textPāli version does not match as well with the Chinese as the variant “sampajaññaṃ na riñcati”given in footnote 7.

[5] The last lines do notmatch well.　身死不墮數 translates kāyassa bhedā saṇkhyaṃ nopeti.　The Chinese omits dhammaṭṭho and vedagū.　永處般涅槃 “abiding forever in Nirvana”is not there in the Pāli, but does not look like a addition to fill the metereither.　The original must have been different.

[6] We have to tweak the DTDto allow the ana attribute into the <xptr>, but I think this is the bestsolution.　For this kind of comparative analysis it is desirable to have linksthat can associate interpretative values.

[7] It is well possible tocompare more than two sutras in this way.　For the sakeof simplicity here we will use only two.

[8] Comparing the Pāli withthe Chinese “same setting” does not mean much, because there is only a limitednumber of settings.　Thus, the constellation “samesetting, different topic and argument,” is true for too many sutra pairs, e.g.　all the sutras that Buddha told Ananda in Sāvatthī.　It can be a useful category, however, if one first searches forstructures inside one tradition and then compares the results.　E.g.　How many sutras were told by Sāriputta in the Pāli and the Chinesetradition respectively?

[9] Sallatena Sutta:PTS SN IV, 207; English IV, 139.　Chineseversion in Taishō vol. 2, p. 120a.

[10] 味 (“taste”) here probably renders assādaṃ.

[11] There are seven anusaya(inclinations, tendencies).　One can have anusaya towardskāma-rāgā (sensual passion), paṭigha (grudge), diṭṭhi (views), vicikiccā (doubt), māna (conceit), bhavarāga (cravingfor continued existance), avijjā (ignorance).

[12] 使 here probably a mistake for受.

[13] Anesaki (1908), Akanuma,Chizen赤沼智善: Kanpa shibu shiagon goshōroku 《漢巴四部四阿含互照錄》（The Comparative Catalogue of Chinese Āgamas and Pāli Nikāyas）.　Nagoya: Hajinkaku shobō破塵閣書房, (1929),and my digital “ComCat” (trial version 2002) available at the DBLM website.