The Study of Sanskrit in Medieval East Asia: China and Japan*

34  Download (0)

Full text



The Study of Sanskrit in Medieval East Asia: China and Japan



University of British Columbia

Abstract: This paper explores the historical study of the Sanskrit language and its related systems of writing in ancient and medieval East Asia. It is argued that the varied availability of teachers and manuals in different time periods and environments led to uneven studies of Sanskrit in different generations. In some cases, we can point to significant understanding of Sanskrit in the writings of some monks. Although some monks had direct access to Indian teachers, the majority of students never had this opportunity, and instead relied on resources in Chinese, which primarily included word lists, rather than grammars. There is evidence for the systematic study of Sanskrit grammar, but this was apparently limited in time and faced a number of challenges. The script of Siddhaṃ became widely studied as a sacred system of writing, but I argue that this did not necessitate the learning of Sanskrit grammar.

Keywords: Sanskrit, China, Buddhism, Japan, Mikkyō, Jōnen, Siddhaṃ


Hualin International Journal of Buddhist Studies, 4.2 (2021): 240–273


* An earlier version of this paper was read on March 22, 2021 as ‘How did a Japanese Buddhist Monk Read Sanskrit?: Jōnen’s Understanding of Sanskrit and Siddhaṃ’, at the American Association for Asian Studies. I must thank Shuheng (Diana) Zhang for organizing the panel and inviting me to write this paper, and the following people for their valuable input and assistance: Nirajan Kafle, Peter Bisschop, Jayarava Attwood, Martin Gansten, Mimi Yiengpruksawan, Michel Gauvain, Jonathan Silk, and the anonymous peer reviewer.


his paper examines the study of Sanskrit in medieval China and Japan. Knowledge of Sanskrit and Indic languages was increas- ingly transmitted alongside Buddhism into China from the early centuries of the Common Era onward. The Japanese, who inherited Buddhism from the mainland, also acquired knowledge of Sanskrit to some extent, but never in a systematic fashion. We might wonder about the extent to which students of Sanskrit—both in China and Japan—comprehended Sanskrit texts, especially when native speakers of Indian languages were few in number, or otherwise simply un- available.

Another question to ask is what happened with Sanskrit studies over the centuries in East Asia. How did it evolve? How did China and Japan differ in this regard? The latter preserved until modern times extensive Chinese Buddhist literature alongside a tradition of utilizing an Indian writing system called Siddhaṃ in formal practices, particularly within Mikkyō traditions (i.e., Shingon and Taimitsu), but we might ask what that meant in terms of literacy and understanding of grammar and vocabulary. How much Sanskrit grammar, for example, did a monk in medieval Japan understand? To work toward an answer to this question, we can look at an analysis of a Sanskrit hymn by a monk from the twelfth century. This analysis combined with a broad look at the available manuals indicate that although Japanese monks studied Siddhaṃ and its pronunciation, there are only a few examples of Sanskrit grammar being studied.

There were, in contrast, more monks in China who studied Sanskrit grammar. This study points out, however, that although some Chi- nese monks in Tang China possessed advanced knowledge of Sanskrit grammar, such as those involved in translation projects, systematic


study of the Sanskrit language faced a number of obstacles and it was not maintained over time. Although Siddhaṃ took on an important role within hieratic contexts both in China and Japan, this did not mean many Buddhist monks in medieval East Asia necessarily read Sanskrit fluently, even though Siddhaṃ script was treated as a sacred writing system.

Historical Background: Indic Languages in China

Indic and other foreign scripts would have been seen in China as early as the beginning of the Common Era following the first trans- lations of Buddhist texts. These included the Kharoṣṭhī and Brāhmī scripts early on.1 One of earliest datable references to Kharoṣṭhī and Brāhmī is found in the Chinese translation of the *Vibhāṣā-śāstra (Piposha lun 鞞婆沙論), produced in 383 CE: ‘It is akin to quickly learning Kharoṣṭhī script when having already learnt Brāhmī script

如學梵書已速學佉樓書’.2 The translation of the Guoqu xianzai yinguo jing 過去現在因果經 [Sūtra on Past and Present Causes and Effects] by Guṇabhadra / Qiunabatuoluo 求那跋陀羅 (394–468) mentions that in Jambudvīpa there exist the Brāhmī and Kharoṣṭhī scripts, but there also exists a ‘Lotus Script’ (lianhua shu 蓮花書).3 M.

Nasim Khan has investigated an undeciphered script in Gandhāra, which he initially called Kohi. He points out that the Mahāvastu and the Lalitavistara refer to Brāhmī, Kharoṣṭhī, and Puṣkarasārī, the latter likely referring to this unique script of Gandhāra.4 Another Chinese translation, the Fo benxing ji jing 佛本行集經 [Sūtra of the Collection of the Past Activities of the Buddha] translated by

1 For a discussion of these scripts, see Falk, Schrift im alten Indian, 84–167.

2 T 1547, 28: 493b7–8. 佉樓 (Middle Chinese: kʰɨa ləw). Reconstructed read- ings of Middle Chinese (Pulleyblank) drawn from database on

3 T 189, 3: 628a15–16.

4 Khan, ‘Kohi or Puşkarasāri’, 7–8. I must thank Henry Albery (private com- munication, January 21, 2021) and Andrew Nguy (private communication, Feb- ruary 3, 2021) for pointing out the information related to Puṣkarasārī.


Jñānagupta/Shenajueduo 闍那崛多 (523–600/605?), lists sixty-four scripts, one of which is ‘Script Taught by the Sage *Puṣkara 富沙迦羅 仙人說書’ with a Chinese note translating this as ‘Lotus’ (lianhua ).5 The ‘Lotus Script’ in question, therefore, certainly refers to the Puṣkarasārī script. There is no mention in said list of Siddhaṃ (Chn.

Xitan; Jpn. Shittan 悉曇), which would later feature prominently in East Asia. This is in contrast to Chaudhuri who claims that ‘the Brāhmī scipt used for writing Sanskrit had regional variations, and the Chinese called the script form that was introduced to them as hsi-t’an 悉曇, a corruption of Siddham. They commonly used this word to mean the language also’.6 In reality, during the first five to six centuries of the Common Era, we see a Chinese awareness of Kharoṣṭhī and Brāhmī, but not Siddhaṃ. Siddhaṃ in these early cen- turies would have presumably just referred to the standard model of syllabic arrangement, rather than a specific system of writing, which came later.7

The Chinese would have been exposed to foreign languages and scripts during the early contacts with the ‘Western Regions’ (Xiyu 西 ). This is illustrated by the introduction of foreign loanwords and characters devised to phonetically represent them from the period of the Han dynasty onward, such as, for instance, tihu 醍醐. Pulley-

5 T 190, 3: 703c12. 富沙迦羅 (Middle Chinese: puwH ʂaɨH kɨa la). This same text gives what appears to be the earliest Chinese reference to the Greek lan- guage. 耶寐尼書 (Middle Chinese: jia miH ɳji) appears to be a transliteration of Yavanī (‘Ionian, Greek’), which presumably would refer to Bactria. The subse- quent note reads, ‘In Chinese called the script of Daqin 隋言大秦國書’. T 190, 3: 703c13. Daqin is a reference to the Levant and/or Byzantium, or in some cases to territories in Persia. For some recent points on Daqin, see Kotyk, ‘La nascita di Cristo’, 116–117.

6 Chaudhuri, Sanskrit in China and Japan, 9.

7 Salomon notes that ‘the terms siddham and Siddamātṛkā later came to be applied not only to the system of syllabic arrangement, but also to a particular local and highly influential script form which was current in northern India around the second half of the first millennium CE’. Salomon, Siddham Across Asia, 11.


blank connected this to Mongol čige(n) (kumiss) and concluded that

‘the assumption of a common borrowing from Hsiung-nu [Xiong- nu 匈奴] seems to be the best way to account for this’.8 Interest in foreign scripts, however, appears to have been largely limited to the Buddhist community. Some early translators of Indic texts, such as Faxian 法顯 (d. 418–423), who travelled to India and back between 399–414, became literate in Sanskrit and other Indic languages.

Faxian, for example, in Pāṭaliputra ‘studied Sanskrit texts and the Sanskrit language’ 學梵書梵語 for three years.9 The extent to which someone in China during Faxian’s time could have learnt Sanskrit is unclear. We might imagine that monks in China largely learnt Sanskrit and other Indic languages through direct instruction from foreign monks or even Brahmins resident in China, but only when this was possible.

We can point to the study of foreign languages in the capital from the early part of the Tang dynasty. According to an early ver- sion of the biography of Xuanzang 玄奘 (602–664), for example, at the age of twenty-nine ‘he stayed in the capital, widely familiarizing himself with foreign lands, and extensively studying scripts and languages’ 頓迹京輦廣就諸藩遍學書語. Information concerning under whom he studied is not given.10 There is no evidence that the Aṣṭādhyāyī by Pāṇini was ever translated into Chinese, but we can imagine that some Indian teachers in China would have possessed the background education to teach Sanskrit grammar. One of the five traditional sciences (Skt. pañca-vidyā; Ch. wu ming 五明) is the study of grammar and phonology (Skt. śabda-vidyā; Ch. sheng ming 聲明).

Xuanzang in his account of India mentions this as part of the general

8 Pulleyblank, ‘The Consonantal System of Old Chinese: Part II’, 255.

9 This is reported in his travelogue, the Gaoseng Faxian zhuan 高僧法顯傳 [Account of the Eminent Monk Faxian]. See T no. 2085, 51: 864b28–29.

10 See translation in Kotyk, ‘Chinese State and Buddhist Historical Sources on Xuanzang’, 529–530. This biography in question is that compiled by Daoxu- an 道宣 (596–667) sometime between 646–649. The early recensions of this text were preserved in Japan. See details in Ibid., 520–521.


curriculum of students there from the age of seven.11 We can indeed imagine a number of Indian monks during the Tang period offering guidance in Sanskrit studies in China. Xuanzang also relates that the Sanskrit language and script ‘were created by the god Brahma 梵天 所製’, and that the pronunciation of Middle India (in contrast to neighboring regions) was identical to that of the gods.12 The sanctity and divine power of the language, and the need for proper pronunci- ation, were instilled in the Chinese imagination through this concep- tualization of Sanskrit.

In a later generation, Yijing 義淨 (635–713), another monk who studied abroad in what are now geographically the nations of Indo- nesia and India, was confident that one could translate Sanskrit after studying the language for a few years. He explains as follows:

If you just learn this, you will understand all the rest of the language.

It isn’t the same as the older Thousand Word Prose. If you read San- skrit texts together with the Siddhaṃ manual(s), you will be able to translate in one or two years. 但學得此則餘語皆通, 不同舊千字文.

若兼悉曇章讀梵本, 一兩年間即堪翻譯矣. 13

Siddhaṃ script, which descended from Brāhmī script, became an important component within Buddhist Mantrayāna in China and later Japan. Mantrayāna became increasingly widespread in the years following Yijing’s death. We can observe that here specifically Yijing does not explicitly mention grammatical forms, declensions, conjuga- tions, etc., but Yijing does discuss Sanskrit grammar in an overview of the topic in chapter thirty-four of his account of foreign Buddhist realms from the year 691. 14

11 T no. 2087, 51: 876c17–18.

12 T no. 2087, 51: 876c9–14.

13 T no. 2133A, 54: 1190a20–21.

14 See translation in Li, Buddhist Monastic Traditions of Southern Asia by Śramaṇa Yijing, 145–155. See T 2125, 54: 228b1–229c27. The Chinese title is Nanhai jigui neifa chuan 南海寄歸內法傳 [A Record of Buddhist Practices Sent Home from the Southern Sea].


It is evident that Sanskrit grammar was also already known to contemporary Chinese monks to some extent. For instance, the noun cases in Sanskrit were known to Fazang 法藏 (643–712). These cases were literally called the ‘eight variable voices’ (ba zhuansheng 八轉聲) in Chinese (‘voice’ in this context does not involve verbs), although the original term in Chinese was zhuan , which means to chirp or sing. Fazang provides a relatively detailed overview of the noun cases in his commentary on the Avataṃsaka-sūtra (Huayan jing tanxuan ji 華嚴經探玄記), as follows:

The eight ‘voices’ are according to the [linguistic] rules of west- ern countries. If one wants to read Buddhist and non-Buddhist texts, one must understand the theory of voices [i.e., cases] and the rules regarding the eight variable voices. If not clearly under- stood, then one will be unable to know the distinctions in mean- ing. I. *puruṣa[ḥ]. This is the indicative voice [nominative case], as in ‘the man chops the tree’ indicates that man. II. *puruṣam.15 This is the voice [expressing] to what / whom the action is done [accusative case], as in ‘the tree to which the action of chopping is done’. III.

*puruṣeṇa. This is the voice expressing the instrument [by] which something is done [instrumental case], as in ‘to chop with a hatchet’.

IV. *puruṣāya. This is the voice conveying for what something is done, as in ‘to chop for the man’. V. *puruṣāt. This is the voice that conveys a cause [ablative case], as in ‘because the man builds a struc- ture, etc’. VI. *puruṣasya. This is the voice which conveys that which belongs [genitive case], as in ‘the servant belongs to the master’. VII.

*puruṣe. This is the voice that conveys that which is dependent [loca- tive case], as in ‘the guest is dependent upon the master’. The second [fascicle] of the Yogā[cārabhūmi] calls the above seven types as the

‘seven model phrases’, since with this one can understand the major models [of cases].16 The theory of voices is of eight variants. They

15 Read si as shan , as per note in Taishō.

16 This is quoting the second fascicle of the Chinese translation of the Yogācārabhūmi (Yuqie shidi lun 瑜伽師地論). See T no. 1579, 30: 289c1–3.


additionally include *[he] puruṣa.17 This is the vocative voice. Fur- thermore, these eight voices include three types: the masculine voice, feminine voice, and neuter voice. These above were explained with the masculine voice, since in Sanskrit a gentleman is called puruṣa.

Moreover, these eight further each include three: the singular voice, dual voice, and plural voice, which then comprise twenty-four voices. There are twenty-four when addressing a gentleman, and also twenty-four voices for the feminine and neuter [respectively].

There are altogether seventy-two voices. One can understand them accordingly with reference to the rules. However, here [in China] we mostly do not have this model. 聲者依西國法, 若欲尋讀內外典藉, 要解聲論八轉聲法. 若不明知必不能知文義分齊. 一補盧沙, 此是 直指陳聲, 如人斫樹, 指說其人. 二補盧私, 是所作業聲, 如所作斫 樹. 三補盧崽拏, 是能作具聲, 如由斧斫. 四補盧沙耶, 是所為聲, 如為人斫. 五補盧沙䫂, 是所因聲, 如因人造舍等. 六補盧殺娑, 是 所屬聲, 如奴屬主. 七補盧鎩, 是所依聲, 如客依主. 瑜伽第二名上 七種為七例句, 以是起解大例故. 聲論八轉, 更加 補盧沙, 是呼 召之聲. 然此八聲有其三種, 一男聲, 二女聲, 三非男非女聲. 此上 且約男聲說之, 以梵語名丈夫為補盧沙故. 又此八聲復各三, 謂一 聲, 二聲身, 三多聲身, 則為二十四聲. 如喚丈夫有二十四, 女及非 男女聲亦名有二十四, 總有七十二種聲. 以目諸法可以准知, 然此 方多無此例.18

Fazang gives the inflected forms of puruṣa (masculine, singular) transliterated into Chinese: bulusha 補盧沙 (*puruṣa[ḥ]), bulushan

補盧衫 (*puruṣam), buluzaina 補盧崽拏 (*puruṣeṇa), etc. Students of Sanskrit in China conceivably relied on this sort of system of phonetic representation, even when learning the noun cases. Later authors were also aware of case inflections. Huilin’s lexicon from 807, for example, explains that the different phonetic transcriptions

17 Read 醯補盧沙. Compare T no. 1831, 43: 614a2. See also T no. 2702, 84:


18 T no. 1733, 35: 149a28–b16. See alternative translation in Staal, A Reader on the Sanskrit Grammarians, 18–19. See also the earlier translation in van Gulik, Siddham, 19–20.


of ‘Magadha’ in Chinese stem from the varying inflections. 19

Staal notes that some of Fazang’s examples could possibly be traced back to the grammatical tradition of India. He notes, ‘The Kāśikā, for example, uses paraśunā cinatti “he cuts with an axe” to illustrate the instrumental (commenting on Pāṇini 1.4.42, 2.3.18)’.20 Fazang perhaps derived these similes from a Chinese commentary on the Indian treatise on logic, the Nyāyapraveśa (Yinming ruzhengli lun 因明入正理論; T 1630), in light of the Japanese monk Annen’s

安然 (b. 841) citation of a certain Qinggan yinming lun zhuchao 幹因明論註抄 [Notes on the Treatise on Logic by Qinggan], which provides a similar explanation:

The ‘eight variant voices’ [i.e., cases] are like when you chop a tree:

there is the tree [nominative], the tree to chop [accusative], the hatchet with which to chop this tree [instrumental], chopping it [the tree] to build a house [dative], chopping it because the king orders it [ablative], chopping it because one is serving an official [genitive], and chopping it on that land [locative]. This is called chopping the tree. 八轉聲者, 例如斫樹木時, 而言樹木, 而斫樹木, 是斫樹木之斧, 而爲造屋斫之, 而因王命斫之, 而屬官家斫之, 而依 其地斫之, 呼爲斫樹.21

19 T no. 2128, 54: 434b1–2. 摩竭提者, 或云摩伽陀, 或云摩揭陀, 或曰墨竭提, 此之多名由依八轉聲勢呼召致異, 然其意義大略不殊.

20 Staal, A Reader on the Sanskrit Grammarians, 18. See also earlier comments in van Gulik, Siddham, 19–20. Bronkhorst states that the Kāśikā ‘is the oldest surviving commentary on the whole of Pāṇini’s Aṣṭādhyāyī. It is our earliest tes- timony for all those sūtras of Pāṇini’s text that are not cited, used or referred to in Patañjali’s Mahābhāṣya. It is also the earliest text in the Pāṇinian tradition that contains a full Gaṇapāṭha, i.e., a complete collection of the lists (gaṇa) of words that accompany many sūtras. Being the earliest text of its kind that has sur- vived, the Kāśikā is an indispensable tool for all historical research into the early history of indigenous Sanskrit grammar, Pāṇinian and non-Pāṇinian.’ See Bronk- horst, ‘The Importance of the Kāśikā’, 129.

21 T no. 2702, 84: 385b18–21. Read er as suo throughout this line. The text cited appears to be the Yinming ruzhengli lun zhuchao 因明入正理論註鈔,


In light of this sort of interest in Sanskrit grammar, we might infer that systematic study of it was undertaken within a Chinese language medium, yet van Gulik was doubtful of this. He writes, ‘Chinese Buddhist monks could easily have collected the references to Sanskrit grammar scattered over the various translated sutras and supple- mented this information with what they could have learned in con- versation with foreign monks resident in China, so as to draw up an annotated version of the rules of the ancient Indian grammarians. As far as I know, however, this task was never undertaken in China’.22 This conclusion might have been premature. The lexicon for Sanskrit grammar was already well-developed in commentarial literature in Chinese on Yogācāra during and shortly after Xuanzang’s time, which was likely inspired by an interest in cultivating and emulating the śabda-vidyā in the Indian manner.23 Fazang appears to have read Kuiji’s 窺基 (632–682) commentary in particular, titled Cheng weishi lun zhangzhong shuyao 成唯識論掌中樞要 (Essentials of the Treatise on the Theory of Consciousness-Only in the Palm of the Hand).

Therein we see an outline of the ‘theory of voices, subanta 蘇漫多聲 ’, which refers to the eight cases.24 The following table reproduces Kuiji’s presentation of eight declensions. Note that he only provides Chinese characters and I have added the proposed translated words in Latin script directly beneath the Chinese characters. Kuiji appears to decline bhavan (‘being’), present participle bhavat, although in an irregular manner. Annen in his Shittan zō 悉曇藏 (Siddhaṃ Repository) reproduces these lines from Kuiji’s work along with the Siddhaṃ letters, which might have been part of the original text, but the letters in Annen’s work appear corrupted (although, again, this

which is listed in the Tendaishū shōsho 天台宗章疏 [Account of Tendai Texts] by Gennichi 玄日 (846–922). See T no. 2178, 55: 1137a16.

22 van Gulik, Siddham, 21.

23 See, for example, Yugaron gi 瑜伽論記 [Commentaries on the Yogācārabhūmi]

by Dullyun 遁倫: T no. 1828, 42: 414a9–22.

24 T no. 1831, 43: 613c3. Compare Fazang’s remarks with T no. 1831, 43:



might have been part of Kuiji’s original work and then recopied into latter manuscripts). I have included Annen’s Siddhaṃ below each declension for reference.25

The Theory of Voices, Subanta 蘇漫多聲說26 Masculine Voice

男聲 Feminine Voice

女聲 Neuter Voice

非男非女聲 一儞利提勢

1. nirdeśa 一婆婆那1. bhavan bhabhana

一婆婆那帝 1. bhavantī bhabhanati

一婆婆多1. bhavat bhabhata


2. upadeśana 二婆婆那擔2. bhavantam bhabhanataṃ

二婆婆那底摩 2. bhavantīm


二婆婆䫂2. bhavat bhabhatta


3. kartṛkaraṇa 三婆婆多3. bhavatā bhabhanatā

三婆婆那底夜 3. bhavantyā bhabhanatya

第三囀下稍近男 聲From third case onward, it is quite close to the masculine voice.


4. sāmpradānika 四婆婆羝4. bhavate bhabhanate

四婆婆那帶 4. bhavantyai bhabhanate


5. apādāna 五婆婆多褒5. bhavantaḥ bhabhanataḥ-ā

五婆婆那底夜褒 5. bhavantyāḥ

bhabhanatya- ā

25 For Annen’s text, see T no. 2702, 84: 385a3–15. The Siddhaṃ letters and Chinese text here are extracted from The SAT Daizōkyō Text Database (

26 The following website was helpful in interpreting the declensions in Chinese:




6. svāmivacana 六婆婆那多阿6. bhavantaḥ bhabhanataḥ-a

六婆婆那底夜阿 6. bhavantyāḥ

bhabhanatya-aḥ 七珊儞陀那囉梯

7. saṃnidhānārtha 七婆婆底7. bhavati bhabhani

七婆婆那底夜摩 7. bhavantyām

bhabhanatyama 八阿曼怛羅泥

8. āmantraṇa 八於初囀上加醯字8. Add he ( ) to first case.

八於初囀上加醯字 8. Add he ( ) to first case.

This sort of presentation of Sanskrit grammar in Chinese suggests to me that all declensions and conjugations were most likely avail- able as part of handbooks, even if these were not widely circulated, although the garbled quality of the Siddhaṃ reproduced by Annen could indicate that precise and accurate handling of the script were lacking in the original Chinese materials. This sort of approach to learning Sanskrit—with transliteration into Chinese and unreliable Siddhaṃ spellings—appears to have been what Chinese and Japanese monks would have had available to them. In the table above, it is possible that the Siddhaṃ letters were added only after the Chinese characters were used to record the declensions. In other words, the Chinese transliteration of the Sanskrit came first (perhaps recorded from oral recitation) and then the Siddhaṃ letters were added after- ward. We might imagine a Chinese student learning declensions and conjugations through an oral medium and then writing down what they heard in Chinese characters. Attempting to read a Sanskrit text with this sort of system would have been conceivably quite difficult, but in the majority of cases, translation from Sanskrit into Chinese was generally undertaken with the assistance of foreign scholars, although this was not always so. 27

Another point requiring consideration is that the Chinese under-

27 For a study of how Sanskrit Buddhist texts were translated into Chinese, see Funayama, Butten ha dō kanyaku sareta no ka.


standing of Sanskrit underwent further development with particular interest in Siddhaṃ as a sacred system of writing, which was further used in visualizations.28 This was in large part due to the interest in Mantrayāna, which emphasized the orthodox pronunciation of mantras while also greatly utilizing the Siddhaṃ script in various contexts. This interest is represented by the Xitanzi ji 悉曇字記 [Ac- count of Siddhaṃ Letters] by Zhiguang 智廣 (d.u.), which dates to sometime before 806 when Kūkai 空海 (774–835) returned with it to Japan. Zhiguang states that he wrote his work after he recited dhāraṇīs but discovered many errors when attempting to reproduce the proper pronunciation. He met the monk Prajñābodhi (Boreputi

般若菩提) from Southern India, who had brought with him dhāraṇī texts to Mount Wutai. Zhiguang’s work deals primarily with the phonetics of Siddhaṃ based on guidance from Prajñābodhi, but only in one brief fascicle.29 This text appears in Kūkai’s catalog of items brought back from China, although it does not appear to be men- tioned in Chinese sources. Manuals on Sanskrit grammar, assuming they existed, might have also similarly remained unrecorded in the extant literature of Chinese Buddhism. Kūkai also records a text titled Xitan shi 悉曇釋 [Explanation of Siddhaṃ].30 Annen in 885 cites this work in two instances, but only in one of these does the citation mention Siddhaṃ, and this is just Sanskrit letters with kanji (Chinese characters) used for phonetic transliteration. We cannot determine whether this was a guide to grammar.31

28 The deities depicted in maṇḍalas are generally each assigned a seed syllable (Skt. bīja). These were preserved in Japan. For an encyclopedic overview of the two primary maṇḍalas of East Asia with their various deities, seed syllables, and other features, see Somekawa, Mandara zuten.

29 T no. 2132, 54: 1186a10–13. A text by Prajñābodhi (Nan Tianzhu Boreputi xitan yishiba zhang 南天竺般若菩提悉曇一十八章; Eighteen Chapters on Sid- dhaṃ by South Indian Prajñābodhi) is recorded in Annen’s catalog: T no. 2176, 55: 1130c19–20.

30 T no. 2161, 55: 1064a27–28.

31 See T no. 2702, 84: 407c8, T no. 2397, 75: 541b21. Annen also lists the Xitan shi in his bibliography: T no. 2176, 55: 1131a5.



Looking at Japan, the first probable transmission of Sanskrit studies into Japan based on the extant record dates to 736 during the Nara Period, the year when Bodhisena (Bodaisenna 菩提僊那; 704–

760) of India and Buttetsu 佛哲 (d.u.) of Linyi 林邑 arrived. Japan was increasingly exposed to Sanskrit in varying degrees in subsequent generations via Buddhist texts and monks returning from abroad.

Hatsuzaki also points out that the study of Siddhaṃ in Japan was historically limited due to the nature of the language differing from Japanese (this was also the reality with Chinese and Sanskrit) and the relevant literature remaining largely unavailable in Japan, even though monks of Taimitsu and Shingon both studied the doctrinal and symbolic significances of Siddhaṃ letters and phrases. Monks in the early Heian period who travelled to China had advantages over their successors, since Indian teachers were available in China.

There are recorded instances of Japanese monks in China who had the opportunity in China to learn Sanskrit directly from Indians and also Chinese specialists. These monks included Kūkai and others (see below).32

Kūkai’s proficiency with Siddhaṃ and the associated lore is demonstrated in his Bonji Shittan jimo narabini shakugi 梵字悉曇 字母并釋義 [Letters of Sanskrit and Siddhaṃ, and Their Exegesis].

Some myths surrounding Kūkai, which are often held to be true even by modern scholars, suggest that he capably understood both Chi- nese and Sanskrit, but Kobayashi in 2009 called into question wheth- er Kūkai really possessed a solid grasp on Sanskrit itself. Kobayashi further challenges modern scholarship that uncritically accepts the traditional account which explains that Kūkai learnt Sanskrit under Huiguo 惠果 (746–805), and assumes Huiguo, and by extension Kūkai, must have capably understood Sanskrit, otherwise they could not have transmitted the esoteric teachings. Kobayashi also points out that Kūkai really did not have so much time in China—which could be counted in months—to study Sanskrit and adequately

32 Hatsuzaki, ‘Kōbō Daishi no shittangaku’, 154.


master the noun declensions and verb conjugations, and furthermore what he would have read was dhāraṇīs, which are merely incanta- tions, rather than literature, that only require basic knowledge of the Siddhaṃ script.

Another figure of note was the Tendai monk Ennin 圓仁 (794–

864). His travelogue written in China, the Nittō guhō junrei kōki 唐求法巡禮行記 [The Record of a Pilgrimage to China in Search of the Dharma], gives the follow account on 28th of June, 842:

I studied Siddhaṃ again and orally received the proper pronuncia- tion from Tripiṭaka Master *Ratnacandra of India at Qinglongsi [in the capital, Chang’an]. 於青龍寺天竺三藏寶月所, 重學悉曇, 親口 受正音.33

Ennin mentions this tutorial, but this does not indicate he immersed himself in the study of Siddhaṃ for more than a day. In this case, he simply reviewed the pronunciation of letters with an Indian teacher, rather than having studied Sanskrit grammar. Ennin’s junior col- league, Enchin 圓珍 (814–891), in autumn of 853 studied Sanskrit and acquired related texts from *Prajñātara (Boredaluo 般若怛羅).34 Again, the extent to which he studied Sanskrit is unclear, since Enchin only relates that he ‘studied the Siddhaṃ manual of Brahma, and then received Sanskrit texts 學梵天悉曇章竝授梵夾經等’.35

Although Ennin, Enchin and others had opportunities to learn directly under Indian teachers in China, later Japanese monks had no such access. As we will see below, later generations of monks in Japan understood the pronunciation of Siddhaṃ letters primarily through kanji. The Japanese could also indicate the pronunciation of kanji with phonetic kana, but the limitations of this script would have prevented the preservation of the original pronunciation of Sanskrit.

Apart from the few who studied in China, Japanese monks would have never heard or ever been able to study the ‘true pronunciation’

33 CBETA B18, no.95: 93b16–17.

34 T no. 2172, 55: 1101c6–13.

35 T no. 2172, 55: 1101c12.


of Siddhaṃ as Ennin and Enchin had experienced.

Moving to a later century, we will focus on Jōnen 靜然 (d.u.). His Gyōrin shō 行林抄 [T 2409; Summary of the Forest of Practices], compiled in 1154, offers a detailed analysis of a Sanskrit stanza in an attempt to decipher the meaning of the individual words.36 The stanza in question also appears in some ritual manuals in the Taishō canon, in Siddhaṃ and/or transliteration into kanji.37 One of these is the Beidou qixing humo fa 北斗七星護摩法 [Homa Ritual for the Seven Stars of the Dipper of the North], which is nominally attribut- ed to Yixing 一行 (673–727), but this is spurious because this sort of practice postdates 727.38 This stanza is labelled zhutian zan 諸天 [‘Hymnal Praise for the Gods’] and was, it seems, used to evoke worldly deities for their blessings toward the end of a ritual.39 It seems this stanza was treated in East Asia as a dhāraṇī to be recited, although it might not have originally been regarded as a dhāraṇī, i.e., a sacred incantation like a mantra.

Jōnen’s Analysis in Gyōrin shō of the ‘Hymnal Praise for the Gods’

The following is a translation and analysis of Jōnen’s commentary on the ‘Hymnal Praise’ that includes his citation of Sanskrit in both Siddhaṃ and kanji.40 The point of this exercise is to show how Jōnen read and deciphered the lines of Sanskrit. One of the main points to which we should pay attention is the absence of reference to gram- mar altogether in Jōnen’s analysis.

36 Jōnen was affiliated with the temple Mudōji 無道寺. He was a disciple of the Sōshitsu 相實, the progenitor of the Hōman-ryū 法曼流. See Dolce,

‘Taimitsu’, 763.

37 T no. 924C, 19: 32c18–22. T no. 1287, 21: 357b20–c4. T no. 1290, 21:


38 T no. 1310, 21: 458b3–8. Kotyk, ‘Yixing and Pseudo-Yixing’, 27–30.

39 See example of this: T no. 1287, 21: 357b20.

40 T no. 2409, 76: 409c13–410a33. See also the work on this hymn by Kiyota,

‘Shaka-zan (ōshin-zan) to shoten bongo zan’, 24–28.


Below I include the Siddhaṃ and kanji provided by Jōnen for each word or phrase alongside his notes, which I have translated. The individual vocabulary cited by Jōnen does not always match up with the initial full stanza provided at the beginning (presented immedi- ately below), which seems to reflect the fact that he was compiling his material from multiple manuscripts.

Siddhaṃ 41 1.

2. 3.

4. 5.


Latin Text 1. ayaṃtudevacagasura 2. kindaradarakṣakranaya 3. prapradharmagritadhikra 4. vidharmacapraśamaśaikhya 5. nemetabhūtametaprakaśaya 6. tanehaśramaṇayadhahaṃ

Kanji 1. 阿都泥嚩左誐素羅

2. 緊那羅那囉鑠迦囉二合那野 3. 鉢囉嚩羅達磨蘖哩二合多地迦囉 4. 尾達磨左鉢囉二合捨磨操企也二合四 5. 儞銘多部多銘多鉢羅二合迦捨夜 6. 怛儞賀室囉二合麼拏也駄𤚥引 六 Middle Chinese

(Pulleyblank) 1. ʔa jianX tɔ nɛj bwaH t͡saX ŋa sɔH la 2. kinX naX la naX la ɕɨak̚ kaɨ la naX jiaX 3. pwat̚ la bwaH la dat̚ mwa ŋat̚ lɨX ta diH kaɨ la 4. mujX dat̚ mwa t͡saX pwat̚ la ɕiaX mwa t͡sʰaw kʰjiə̆ X jiaX 5. ȵiə̆ X mɛjŋ ta bɔX ta mɛjŋ ta pwat̚ la kaɨ ɕiaX jiaH 6. tat̚ ȵiə̆ X ɦaH ɕit̚ la mwaX ɳaɨ jiaX daH mamX

Jōnen breaks down the hymn into individual components with reference to both the Siddhaṃ and kanji available to him based on a few different editions. Jōnen carried out a careful examination of the materials at hand and, as a result, was able to generally decipher

41 The Siddhaṃ letters and Chinese text here are extracted from The SAT Daizōkyō Text Database (


the meaning of the original Sanskrit, albeit with some misunder- standings. The tone of Jōnen’s writing, however, shows that he was uncertain about certain elements.

ayāntu 句義未尋 . 諸請召呪有此句 . 大底請赴句歟 .

I have not investigated the meaning of the phrase. Evocatory incantations [of sentient beings] have this phrase. It is perhaps generally a phrase for summoning.

Jōnen infers the meaning of the phrase in question by referring to other dhāraṇīs, although he does not state which ones. 42

deva 天也 .

泥嚩 Gods.

In some instances, the meanings of individual words are apparent to Jōnen without reference to other works. In other cases, as we will see below, Jōnen guesses at the meaning.

bhūjagā 龍也 . 義釋云部惹誐 . 唐院讃一本云冐左迦 . 一本云胞若虎 .


左誐 Dragons. The Exegesis [of the Vairocanābhisaṃbodhi] gives 部 惹誐. One edition of the Praise from Tō-in gives 冐左迦, while one gives 胞若虎. Here it is perhaps an error where it gives 左誐.

The Siddhaṃ word here is clearly referring to bhujaga or bhujaṃga (snake, serpent), although the Siddhaṃ here differs from the line given at the beginning in original stanza. The Chinese transcription (t͡saX ŋa左誐) is missing a character to phonetically represent bhu-.

Jōnen critically referred to other editions, such as those from the

42 There are clear examples of other dhāraṇīs in Chinese transliteration that commence with kanji phonetically representing ayāntu. See, for example, T no. 873, 18: 304a1 & 874, 18: 315c30.


Tō-in 唐院, in which kanji representing bhu- are given (mawH and paɨw ). Jōnen then notes the error in the original transcription.

The diversity of transcriptions of Sanskrit is informative with respect to the scribal practices of copyists. The Siddhaṃ and kanji could both be reproduced in different forms, a point which likely reflects the fact that copyists (i.e., an amanuensis) were often writ- ing what they heard dictated. The variation in Siddhaṃ spellings is further explained by the fact that Japanese scribes did not use the original Indic pronunciation, but instead they used phonetic transcriptions based on kanji and kana (e.g., hūṃ, written in kanji as , is pronounced un in Japan). Detailed works on Sanskrit grammar and phonology, such as Pāṇini or others for example, were evidently not available in medieval East Asia, although as mentioned earlier, there likely existed handbooks on Sanskrit grammar written in Chinese. In Japan, Siddhaṃ and Indic vocabulary were basically studied through a Sino-Japanese medium. There consequently exist- ed considerable variations in spellings of Siddhaṃ in some instances, even for well-known mantras and dhāraṇīs, such as that of the Heart Sūtra. Dreitlein notes that ‘the Siddham in Kūkai’s text reads

*pragate (where the standard text has pāragate) and *prasugate (in- stead of pārasaṃgate). This may be a mistake on Kūkai’s part, a copy- ist’s error, or Kūkai may be using a different text from the standard one known today. Note that, however, the oldest extant manuscript of the Heart Sūtra in Siddham, the Hōryū-ji manuscript, gives the standard form’.43 Variations in Siddhaṃ spellings clearly existed from early on in Japan.

Jōnen’s citation of the exegesis of the Vairocanābhisaṃbodhi is important to note.44 This work was mined for authoritative

43 Dreitlein, ‘An annotated Translation of Kūkai’s Secret Key to the Heart Sūtra’, 36, fn. 127.

44 The Dari jing yishi 大日經義釋 [Exegesis of the Vairocanābhisaṃbodhi] is a revised version of the commentary compiled by Yixing 一行 (673–727) on the basis of an oral testimony by Śubhakarasiṃha 善無畏 (637–735). Kano, ‘Vairo- canābhisaṃbodhi’, 383. For further discussion regarding the complex history of the commentaries, see Mano, ‘Kan’yaku Dainichikyō no chūshakusho’, 218–223.


definitions of Indic vocabulary.45 Other lexicons of Chinese-Indic vocabulary were available in Japan, such as the Fanyu zaming 語雜名 (Miscellaneous Sanskrit Words), which was compiled by a monk from Kucha named Liyan 禮言, and later brought to Japan by Ennin.46 This text is a long list of Indic words in Siddhaṃ and Chi- nese characters, together with each word’s meaning in Chinese. This type of document would have been consulted by Japanese monks who studied the vocabulary of dhāraṇīs and verses in Siddhaṃ and even those transliterated into kanji.

sura 上引 字即 也 . 云非天 .

素囉一 The above elongated gā letter is the a. An asura is a non-god [i.e., the Asuras who battle the Devas].

Jōnen here shows an awareness of word boundaries, specifically long vowels, which can be a feature of sandhi, although the concept of sandhi itself does not appear to have been studied or known.

kintarendra 疑神也 . 點即 也 . 王也 . 字衆本皆爾 . 唐院本或

云緊駄 . 或本云緊曩哩曩捺囉 . 今直云那羅者謬歟 . 緊那羅那囉 Kintara, I suspect, is a spirit. The mark [in the manuscript

viewed by, myself, Jōnen] is i. Indra is the king. The letter re is like this in all editions. Some of the Tō-in editions give 緊駄. Some editions give 緊曩哩曩捺囉. Now here it is perhaps an error where it gives 那羅. 47

Jōnen here is grappling with multiple manuscripts. He could not, it seems, confidently identify the first word here, but we can infer that it is kiṃnara. Monier-Williams defines this as ‘a mythical being with a human figure and the head of a horse (or with a horse’s

45 Jōnen appears to be citing X 438, 23: 365c18 (部若伽龍也). See parallel line at T 1796, 39: 667b25.

46 See Ennin’s catalog of items brought back from China: T 2165, 55: 1075b18.

47 Read as .


body and the head of a man … celebrated as musicians’.48 Normally, the word kiṃnara would have been transliterated into Chinese as 那羅, and this would have been immediately recognized, but Jōnen was perhaps confused by the following term, which through sandhi had modified the immediately preceding vowel (kiṃnara + indra = kiṃnarendra), ‘Lord of the Kiṃnaras.’ In the manuscript available to him, there was the letter i ( ), which perhaps was a notation to indicate that endra was to be read as indra without the sandhi.

śakradaya 帝釋歟 . 字或本 . 唐院本或云舍羯羅那野 . 梵字即今本也 ,

或本云鑠掲羅跢夜叉 .

鑠迦羅二合那野 It is perhaps Śakra [Indra]. The letter kra is krā in some editions. The Tō-in editions give 舍羯羅那野. The Sanskrit letters are as in the present edition. Some editions give鑠掲羅 跢夜叉.

The original Sanskrit here seems to have read as śakra (Śakra the god) + ādayaḥ (‘others’).

pravaradharma 上法也 . 嚩羅云二合謬也 .

鉢羅二合嚩羅二合達麼 The supreme Dharma. The merging of 嚩羅 is an error.

Jōnen displays an awareness of errors in the transliteration of Sanskrit words into kanji. These annotations are typically written in superscript, such as 二合 which show that the pronunciation of the preceding two kanji are merged. This practice was carried over from China. This would have resulted in consonant clusters that do not normally exist in the Japanese language (or Chinese for that matter).

For example, the kan’on 漢音 reading (the borrowed pronunciations from Sui-Tang China) of 鉢羅 is hatsu ra (h was pronounced as p in Old Japanese). In this case, the consonant ending is dropped and the pronunciation would have approximated p[a]ra. There appears to have been an awareness that the vowel following the first consonant is

48 Monier-Williams, A Sanskrit-English Dictionary, 283.


dropped, a point that is reflected in the Siddhaṃ. The kana syllabary, which was designed for the Japanese language, does not allow for a consonant cluster such as pr.

kṛtādhikārā 此句衆本梵字皆同 . 唐院本或云紇栗多地迦囉 , 或云吃哩二合

駄地迦跢 . 今漢字謬歟 .

蘖哩二合多地伽囉 This phrase in the editions [at hand] all have the same Sanskrit letters. Some of the Tō-in editions give 紇栗多地迦 囉. Some give 吃哩二合駄地迦跢. Here perhaps the kanji are erroneous.

Jōnen appears to have not understood this part of the dhāraṇī and how it relates to pravaradharma. In this case, it would refer to the aforementioned beings, who are established (kṛtādhikārāḥ) in the supreme Dharma.

voddhaṃvacaḥ 佛語歟 . 字或作 . 唐院本云冐淡嚩左 , 或云謨朕麼惹 . 尾達麼左 This perhaps means speech of the Buddha. The letter

ddhaṃ is sometimes written as dhvaṃ. The Tō-in edition gives 冐淡嚩左. Some give 謨朕麼惹.

The letter vo is an error for the graphically similar bo , al- though here we might normally expect bu , as in buddha.

praśama 寂也 , 能除也 .

鉢羅二合捨麼 Calm. To absolve.

saukhya 安樂也 . 唐院本或云素契也 .

梵字同今 . 或云鉢羅嚩囉素迦 . 或作 , 今作 本 , 私法本同 之 .

操企也二合四 Peace. Some of the Tō-in editions gives 素契. The Sanskrit letters are identical to the present version. Some give 鉢羅嚩囉 素迦 [pravarasukha?]. Some give so. The present edition has sau. The edition of Kōbō[daishi Kūkai] is identical to this.

Here shi hō 私法 ought to be read as kō bō 弘法, based on the appearance of the latter below. This refers to the edition, or a copy


thereof brought to Japan by Kūkai, i.e., Kōbōdaishi 弘法大師. This would presumably refer to the Bonji tenryū hachibu zan 梵字天龍 八部讚 [Hymnal Praises of the Eight Divisions of Nāgas and Devas in Sanskrit], which is recorded in Kūkai’s list of items brought back from China in 806. Jōnen, citing the bibliography of esoteric works compiled by Annen in 885–902, also mentions this text alongside three others under the same heading of ‘Hymnal Praises to Worldly Deities’ (sho seten zan 諸世天讃). These three texts were carried to Japan by Ennin and Eun 惠運 (798–869).49 These all deal with the Eight Divisions of Nāgas and Devas (tenryū hachi bu 天龍八部).

These may have included different versions of the Sanskrit hymn that Jōnen investigated. 50

nimita 相也 . 唐院本云儞弭多 . 或云儞弭駄 . 今二合謬歟 .

儞銘二合Mark. The Tō-in edition gives 儞弭多. Some give 儞弭駄. The present merging of the two characters is perhaps an error.

Here nimita would normally be nimitta in standard Sanskrit.

Although interpreting this word as ‘mark’ (sō ) would not be totally incorrect, in this context it has the sense of cause, ground, or reason. 51

bhūta 實也 . 唐院本云部駄儞銘二合多 .

部多 Reality. The Tō-in edition gives 部駄儞銘二合.

Again, Jōnen is not entirely incorrect to translate bhūta as re-

49 For Kūkai, see T no. 2161, 55: 1063c18. For Annen, see T no. 2176, 55:

1130b19–22. For Jōnen, see T no. 2409, 76: 409c11–14.

50 This sort of hymnal work was apparently used in the liturgy at Qinglongsi in Chang’an, based on its appearance in the liturgical prescriptions of Faquan (fl. 838–847), titled Gongyang hushi batian fa 供養護世八天法 (Method for Offering to the Eight Guardian Deities). T no. 1295, 21: 382c17.

51 Monier-Williams, A Sanskrit-English Dictionary, 551.


ality. Here, however, nimitta-bhūta would have originally meant

‘being a cause or reason or means’, specifically with regard to praśa- ma-saukhya: thus, ‘the speech of the Buddha is the cause for calm and peace’. 52

metaprakāśya 歟 . 唐院本或云弭 , 以迷達音 , 云如是歟 . 次句

直 鉢囉羯捨也 , 此云開示歟 .

銘多鉢羅二合迦捨夜 This is perhaps metata. Some of the Tō-in editions give metaḥ. Perhaps the meaning is ‘thus’ with the pronunciation 迷達. The following phrase prakaśya is 鉢囉羯捨. This per- haps means ‘to reveal’.

The manuscript appears to have been corrupted. Here, meta, which Jōnen understands as ‘thus’ was conceivably itthaṃ originally.

The alternative kanji provided by Jōnen would have been read as miə̆ X danH ( ) in Middle Chinese (Jpn. mi dan). It is possible that nimittabhūtam-ittha[ṃ] was erroneously copied as mettha and thereafter meta. T no. 1287 gives nimeta bhuta meta prakaṣaya.53

tadiha入聲音如也 . 云如此歟 . 唐院或本云 . 多印賀.

弘法本云 .

怛儞賀 The pronunciation is like (entering tone). I ha perhaps means ‘thus’. Some of the Tō-in editions give ta i ha 多印賀. The Kōbō edition gives i hā śra.

Using the classical system of Chinese tones to indicate the pronunciation of foreign words or mantra elements is a feature of Buddhist lexicography in East Asia.54 The resulting system of pro- nouncing Sanskrit might be regarded as a type of ‘Sinicized Sanskrit’

and this was subsequently imported to Japan. Japanese monks capably

52 Ibid., 551.

53 T no. 1287, 21: 357b27–c1.

54 See, for example, the Yiqiejing yinyi 一切經音義 [Sounds and Meanings of the Scriptures]: T no. 2128, 54: 369a23.


read Chinese, but their pronunciation would have been generally based on phonetic Japanese readings of Chinese characters and also kana. The Japanese preservation of Chinese pronunciations of kanji was therefore only approximate, so their pronunciation of Sanskrit vocabulary, based on borrowed Chinese conventions for representing the sounds of Sanskrit, was similarly approximate. The word svāhā within the mantra of the Heart Sūtra (Jpn. Hannyashin-gyō 般若心 ), for example, is read sowaka 薩婆訶 in Japanese (in Middle Chi- nese sat̚ bwa ha).

Moving on, Jōnen’s cited variations of tadiha in the manuscripts available to him, together with the following lexical item, again point to scribal errors and confusion.

vaṇaya 肉也 . 唐院本 .

Meat. The Tō-in edition [gives] śravaṇi.

Jōnen’s interpretation here is clearly based on guesswork. More- over, it is unclear how he derived ‘meat’ from vaṇaya or śravaṇi (assuming that niku is not an error for another kanji, which is certainly possible; bun / mon , ‘to hear’ potentially could have been the original kanji). Judging from the Siddhaṃ, we might speculate that the original word was śravaṇāya (‘for hearing’), but Jōnen does not actually suggest this anywhere.

dharma 法歟 . 弘法本 . 今云駄𤚥謬歟 . 唐院一本云 愬怛囕 , 此

云經 . 唐院一本十四字爲一句 , 今依之 . 此讃集唐院梵本二本 , 漢字本二本 , 弘法梵本校定了 .

𤚥引六 Dharma? The Kōbō[daishi] edition [gives] dharma. Here it gives 駄𤚥, which is perhaps an error. One of the Tō-in editions gives sutraṃ 愬怛囕. This means scripture. One of the Tō-in editions has fourteen letters as one line. Here I have relied on this. This hymnal praise brings together two Sanskrit editions and two kanji editions from Tō-in, which were corrected based on the Sanskrit edition of Kōbō[daishi].

Based on the material provided by Jōnen and his running com- mentary, we can attempt to tentatively reconstruct the original


Sanskrit as follows. Having shown this reconstruction to a few San- skritists, I received varying opinions and critical pointers, so I con- cede that this attempt is flawed and problematic from the beginning, but it is still a useful exercise because we can, I argue, get an idea of what the original hymn might have been.

āyāntu deva-bhujagāsura-kiṃnarendrāḥ śakrādayaḥ pravara-dharma-kṛtādhikārāḥ |

buddhaṃ vacaḥ praśama-saukhya-nimitta-bhūtam itthaṃ prakāśya tad iha śravaṇāya dharmam ||

Let come the Kings of the Gods, Snakes, Asuras, and Kiṃnaras, and Śakra and others who have been admitted to the best Dharma. The word of the Buddha, the cause of calm and happiness, having thus shone forth, here the Dharma is to be heard. 55

This stanza is in vasantatilakā meter. With regard to the second last line, Gansten writes to me, ‘If you want to make this mean “the word/speech of the Buddha” (which seems reasonable) without violating the metre, you would need to emend buddhaṃ to baud- dhaṃ, making it an adjective (which would be perfectly idiomatic).

As it stands, it can only mean “awakened speech”. Dharmam at the end would have to be taken as a neuter noun, which surprised me (dharma is normally treated as masculine), but according to Monier-Williams it is rare but not unknown, so let it stand. Another option would have been to make that, too, into an adjective—

dharmyam, qualifying “speech”’.56 One concern with this process of reconstruction is the assumption that the original stanza was, in fact, written in entirely orthodox Sanskrit, but it is possible that this was not the case and it could have been composed in a hybrid form. Ide-

55 I must thank Nirajan Kafle, Peter Bisschop, Jayarava Attwood, and Martin Gansten for their assistance in reconstructing these lines of Sanskrit. This reconstruction is a revision of what I presented in Kotyk, ‘Yixing and Pseu- do-Yixing’, 29. Any fault in this reconstruction is my own. See also Kiyota,

‘Shaka-zan (ōshin-zan) to shoten bongo zan’, 24–28.

56 Private communication (February 10, 2021).


ally in the future a better version of the hymn will become available and we can investigate this matter further in order to demonstrate how the original lines changed over time as they were transmitted from India to China to Japan.


Jōnen did not approach the Sanskrit stanza with any systematic grammar, at least judging from his presentation, but instead he large- ly relied on definitions of words derived from an array of sources.

In some instances, he was relying on guesswork, but nevertheless he still critically approached the Sanskrit at hand. Detailed knowledge of Sanskrit grammar, however, was not unknown in East Asia, as we explored earlier, but it is unclear whether Jōnen had access to the relevant training and materials.

One tentative conclusion to take away from Jōnen’s work is that, if he was in fact representative of Mikkyō scholars of his time, then perhaps study of Sanskrit had declined in Japan since the ninth century when figures such as Annen in particular carried out com- prehensive studies of the Sanskrit-related materials available to them.

Monks in Tendai and Shingon certainly continued to study Siddhaṃ as a sacred system of writing, but perhaps the expertise in the subject of Sanskrit had faded over time, particularly after Annen. This situa- tion would be comparable to early Song China, where although in- terest in dhāraṇīs persisted and translation activities occurred under state supervision, local interest in Sanskrit and the opportunity to study it declined. Even when new texts were translated from Sanskrit, they were not so influential or widely read. On this point, we should note that Sen argues that ‘the shifting doctrinal interest among the members of the Chinese Buddhist community towards indigenous schools and practices rendered most of the new translations and their contents obsolete in China’. 57

We ought to recall Kobayashi’s remarks concerning Kūkai, that he

57 Sen, ‘The Revival and Failure of Buddhist Translations’, 31.


did not understand Sanskrit. In light of that, we might also wonder about the other monks who understood Siddhaṃ. What level of knowledge did they possess when it came to analysis of grammar?

If there was no substantial tradition of Sanskrit grammar in Japan from the ninth century, then it is perhaps unsurprising that Jōnen only pieced together the meaning of the hymn in question through reference to individual terms.

When we compare the study of Sanskrit in Japan to China, it is evident that the latter, particularly during the Tang period, had a clearly existent tradition which studied the grammar of Sanskrit, albeit with a number of limitations. My present sense is that this tradition was initially strong amongst students of Yogācāra, which no doubt followed Xuanzang’s legacy, yet the relevant literature was primarily read through Chinese translations. The Chinese lexicon for grammatical terms from Sanskrit was established, which was nec- essary to translate the relevant terminology as it appeared in Sanskrit works of Yogācāra. Although Yijing in a later generation encouraged the study of Sanskrit, it does not seem that such studies were widely taken up. We know that there was knowledge of declensions, but the extant table we presented above indicates that the spelling was cor- rupted and likely influenced by the recording of Sanskrit sounds with Chinese characters. This would have been an obstacle to accurate reading of Sanskrit texts. Nevertheless, the Chinese monks during the period in question often had access to Indian teachers who were resident in China, so their guidance was likely indispensable. The Japanese, however, did not have this opportunity apart from rare instances, such as when Bodhisena stayed in Japan during the eighth century, or when a Japanese monk stayed in China.




Related subjects :