Xuan Zang’s Five Transliterations Revisited: A Corpus Linguistic Study of Vimalakīrti Nirdeśa

ABSTRACT


Introduction 1
The translation of Buddhist scriptures and sutras in China can be dated back to the introduction of Buddhism to China by the Indian monk Kasyapa Matanga in the first century CE.Kasyapa Matanga and Dharmaranta passed Indian Buddhist texts and images to China and first translated the Sutra of Forty-two Chapters into Chinese purportedly.Then An Shigao (c.148-180 CE), the "Parthian Marquess", translated many works including meditation, Abhidharma and basic Buddhist doctrines, more than a dozen of which are currently extant.Massive translations of Buddhist scriptures and sutras date back to the Eastern Han Dynasty (25 AD -220 AD) when Dao'an (312 AD -385 AD) spent the last years of his life supervising translation and interpretation of Buddhist scriptures and sutras.
The Buddhist translation activities with the participation of Chinese and overseas translators also inspired the development of translation theories.Even though Dao'an didn't engage in translation practice, he compiled the first Buddhist contents Zhong Jing Mu Lu (眾經目錄, Contents of All Sutras), which significantly contributed to the systematic translation of Buddhist scriptures and sutras later.He noticed the unfaithful translation in early versions of scriptures and concluded his thesis of Wu Shi Ben (五失本, the five losses from the original) and San Bu Yi (三不易, the three difficulties) in the Preface to the Prajnaparamita.He demanded translators to follow the format of the original text and keep the Chinese versions as closely as possible to the original.
The Buddhist translation peaked at the Tang Dynasty after Kumarajiva's (c.344-c.413) efforts in a massive translation of more than 300 volumes of sutras among others.Xuan Zang (c.602-c.664) was second to none on his contributions to the quality and quantity of translations of Indian Buddhist texts to Chinese in the rest of his life.His achievements in Buddhism were also known to the world for his translation methodologies.According to Dao Xuan's Xu Gao Seng Zhuan (續高僧傳, Supplemental Bibliography of Venerable Masters), the translation of sutras was traditionally done through word-by-word translation in the first step.Then the word order was adjusted by following Chinese grammar.Literati were invited to modify the discourses in the third step.Such translation procedure yielded problems of unfaithfulness or even misunderstanding of the doctrine.What Xuan Zang did in translation was that he interpreted the texts orally to accord with Chinese convention.Then his interpretation was recorded by other translators.Xuan Zang also had to collate the Sanskrit scripture with other versions for faithfulness.Xuan Zang's version was thus referred to as the "new versions" compared to previous versions or the "old versions" (Xie 2009, p.54).
It has been widely accepted that Xuan Zang's most significant contribution to the development of translation theory lies in the thesis of Wu Bu Fan (五不翻, the Five Transliterations), which refers to as "five categories of untranslated but transliterated Buddhist terms".The five categories include the esoteric terms, the polysemic terms, terms denoting things without equivalent in China, old established terms, and original terms which had Buddhist implications.However, the arguments with the Five Transliterations are found in none of Xuan Zang's works.The Five Transliterations was firstly introduced in the first volume of Fan Yi Ming Yi Ji (翻译名义集, Collection of Meanings and Terms in Translation) 2 authored by Fa Yun in the Southern Song Dynasty (1127-1279) as cited below 3 : Master Xuan Zang of the Tang Dynasty indicated five categories of untranslated terms.First, esoteric terms, such as dhāraṇī (mantras).Second, polysemic items, such as Bhagavan, enclosing six connotations (namely, powerful, burning, attractive, good reputation, auspicious and venerable).Third, terms denoting things without equivalent in China, such as jambū tree which is not found in China.Fourth, old established terms, such as anuttarā-samyak-saṃbodhi, which is translatable (as the unsurpassed correct enlightenment) but transliteration is widely used.Fifth, original terms which had Buddhist implications, such as prajñā."Prajñā" sounds respectful whereas "wisdom" informal, so it is better not to translate it.Delhey (2016, p. 53-54) indicates that none of the detailed arguments written by Xuan Zang on translation has been preserved, comparing to some statements from his collaborators and disciples.Murata (1975) doubts that the so-called Five Transliterations appears to be summarized through the Heart Sutra (Prajñāpāramitā hṛdaya).He concludes that Xuan Zang's translation tends to follow the old translations of Buddhist texts rather than empathizing with new methods of translation.In addition, the Five Transliteration is also applicable in the history of China's Buddhist translations.
Whether it was Xuan Zang who proposed the Five Transliterations remains debatable.Its origin has been widely discussed from perspectives of textual criticism, religious studies, as well as translation studies (Fang 2006; Li 2019; Wang 2017; Yang 2010; Yang 1986).However, Fa Yun's record of the Five Transliterations has a profound influence on the later practice of Buddhist translation in China.As Fang (2006) summarizes, the stances on the Five Transliterations in the history of China's Buddhist translation are threefold.First, the Five Transliterations is considered to be a technical convention instead of a translation theory (Liang 1988).Second, the Five Transliterations is considered to be the core of Xuan Zang's translation theories (Wang 1984).Third, the Five transliterations are deemed to be a sum-up of Xuan Zang's translation experiences, for transliteration is not the ultimate method to handle by a translator (Luo 1981).
In this paper, we attempt to revisit the Five Transliteration from the perspective of corpus linguistics for two purposes.On the one hand, we will present Xuan Zang's strategies of processing some Buddhist terms mentioned in Fa Yun's preface.On the other hand, we will try to better understand the tension between historically recorded translation theories (such as Fa Yun's assertion) and practice (such as Xuan Zang's operation) through a particular focus on the Sanskrit-Chinese Parallel Corpus of Vimalakīrti Nirdeśa Sūtra.This paper is organized into five sections.In the first section, we briefly introduce the background of the Five Transliterations in Buddhist translation.Then in the second section, we will briefly introduce the corpora and methodology.Based on the results shown in Section Three, we will discuss our findings so as to achieve the above-mentioned goals in Section Four.Section Five concludes this paper. 2It is commonly regarded that the Five Transliterations were found in the preface of Fan Yi Ming Yi Ji (翻譯名義, Collection of Meanings and Terms in Translation) authored by Zhou Dunyi.Cao (1979) indicates that this is not accurate, for Zhou simply paraphrased Fa Yun's assertion even with errors (see Fang 2006 for details). 3SAT Daizōkyō, Vol.54, p.1057.

Corpora
This paper is based mainly on two self-built corpora: the Corpus of Buddhist Translations by Xuan Zang (henceforth CBTX) and the Sanskrit-Chinese Parallel Corpus of Vimalakīrti Nirdeśa Sūtra (henceforth SCPC-VNS).
The CBTX is based on the SAT Daizōkyō Database, in which 1330 fascicles among 74 texts of Buddhist scriptures translated by Xuan Zang are included.The scriptures are separately saved into 74 TXT files, with 2,646,159 Chinese characters in total.We compare the translations of each term and report the frequencies in Section Three to discuss Xuan Zang's translation strategies in Section Four.In addition, we employ the SCPC-VNS to closely focus on how exactly Xuan Zang dealt with such Buddhist terms.The SCPC-VNS consists of the Sanskrit text, the three Chinese translations by Zhi Qian, Kumarajiva and Xuan Zang, and the English translation by Robert A. F. Thurmann (Thurmann 1976).We also refer to two online sources that associate with VNS: Bibliotheca Polyglotta by the University of Oslo and the Database of Chinese Buddhist translation and their Sanskrit parallels for the Buddhist Chinese Studies by the Education University of Hong Kong.
We select Vimalakīrti Nirdeśa Sūtra (VNS) as our target in our analysis for both religious and linguistic reasons.Religiously, VNS is a representative sutra among early Mahāyāna scriptures, containing Mahayana doctrines in its five facets (Kamata 1994).The insights of VNS had a significant impact on the development of Chinese Buddhism and the long-term cultivation of Buddhism in East and Southeast Asia (Cheng & Tse, 2014, p. 8).The three extant Chinese renditions by Zhi Qian, Kumarajiva and Xuan Zang were directly translated from the Sanskrit original (Cheng 2013; Lamotte & Boin 1976; Cheng & Tse, 2014), while the English renditions mainly from Chinese or Tibetan versions (Cheng 2013).Kumarajiva's version was the most widely circulated one in China before Xuan Zang's translation.However, it was replaced by Xuan Zang's translation in circulation afterwards.
Linguistically, as concluded in He (2007), Xuan Zang was unsatisfactory of neither Zhi Qian's nor Kumarajiva's translation of VNS.Xuan Zang argued that the previous translation of the title Vimalakīrti Nirdeśa (i.e., Wei Mo Jie 維摩詰) was not correct.Evidence comes from Xuan Zang's The Great Tang Records on the Western Regions (Da Tang Xi Yu Ji大唐西域記): To the northeast of the Sangharama 3 or 4 li is a stupa; this is the old site of the house of Vimalakirti (Pi-mo-lo-ki) 4 .
(Vimalakirti) is said to be Wu Gou Cheng (undefiled reputation) in Tang Dynasty.It's been called Jing Ming (pure prestige) before.Even though Jing (pure) means Wu Gou (undefiled), and Ming (prestige) means Cheng (reputation) semantically, there are nuances between them.The old translation was Wei Mo Jie, which was totally wrong and abridged.
In addition, Kui Ji (窥基), a prominent disciple of Xuan Zang, indicated in Shuo Wu Gou Cheng Jing Shu (说无垢称经疏 Explication of Vimalakirti Nirdesa Sutra) that many mistakes were made by Kumarajiva in his translation of VNS 5 .Kui Ji explained in detail how Kumarajiva mistranslated the title of VNS, as shown below: (Vimala) was transliterated as Wei Mo (维摩) (by Kumarajiva).But -ala was omitted to transliterate so denote Wu Gou (undefiled).Jie (诘, reputation) in Kumarajiva's translation was kīrti in Sanskrit.(Nirdesa) simply means Shuo (说, say).But somehow Kumarajiva translated it as Suo Shuo (所说 what's said).It is not necessary to do so.However, Kumarajiva didn't know Chinese nor Sanskrit well for he was a native of Kucha.His translation contained not only mistakes but also semantic misunderstandings.
Kui Ji also remarked that serious problems such as deliberate omission and semantic inaccuracy were widely found in the translations of many other Buddhist scriptures by Kumarajiva.One consequence of such linguistic problems in previous translations led to unfaithfulness, which was beyond Xuan Zang's baseline of Buddhist translation.

Methodology
In the previous part, we discussed the reasons we use VNS as the main target in our study.We refer to two online corpora that associate with VNS.The major corpus is Bibliotheca Polyglotta by the University of Oslo, which contains the Latin transliteration of Sanskrit text of VNS and the three Chinese translations by Zhi Qian, Kumarajiva and Xuan Zang.Besides, we also refer to the other parallel corpus "A Database of Chinese Buddhist translation and their Sanskrit parallels for the Buddhist Chinese Studies" by the Education University of Hong Kong, which contains the forms of Sanskrit without sandhi rules applied.
As what our research questions mentioned above, we take three typical examples in Fa Yun's statement as our major targets: Bhagavan for polysemic items, anuttarā-samyak-saṃbodhi for the old-established items, and prajñā for the original terms which had Buddhist implications.We exclude the examples for 'terms denoting things without equivalent in China' (无此故).The main reason is that transliteration per se is understood as a type of conversion of a term from the source language to the target language by swapping letters.The examples for 'esoteric terms' (秘密故) are also not included due to limited space in this paper.
In the following section, we will report the frequencies of the three typical terms in Xuan Zang's version by comparing them with Kumarajiva's version.

Anuttarā-samyak-saṃbodhi
The compound anuttarā-samyak-saṃbodhi consists of three words: anuttarā meaning 'unsurpassed'; samyak meaning 'perfect'; and sambodhi meaning 'complete enlightenment'.Thus, the compound literally means 'the unsurpassed, complete and perfect enlightenment', or 'the highest perfect awakening', to distinguish the enlightenment of a Buddha from that of an Arhat (Mäll 2003).As Fa Yun mentioned, the compound anuttarā-samyak-saṃbodhi was illustrated as an example of the 'old established terms' that Xuan Zang transliterated but not translated.
However, the results we found contradict Fa Yun's illustration.The appearance of anuttarā-samyak-saṃbodhi in VNS is 30, with its locative case form anuttarāyāṃsamyak-saṃbodhau generally.As the example shown in (3), the form samyaksaṃbodhaye is the dative form of samyaksaṃbodhi.Even though there's no prefix anuttarā (unsurpassed), the term was still translated as Wu shang zheng deng jue無上正等覺.However, the following locative form anuttarasyāṃ samyaksaṃbodhau was translated differently as Wu shang zheng deng pu ti 無上正等菩提.No direct evidence is found on the association between Xuan Zang's translation option and grammatical case.However, a distinct tendency is found by comparing the frequency of the term nuttarasyāṃ samyaksaṃbodhi in all the volumes of both translators that it was preferably transliterated by Kumarajiva but not Xuan Zang7 . ( Vkn Chi: Kumārajīva, T.475 541a5 -11 EN: At that moment, the Licchavi Vimalakīrti entered into such a concentration that those monks were caused to remember their various former existences in which they had produced the roots of virtue by serving five hundred Buddhas for the sake of perfect enlightenment.As soon as their own spirits of enlightenment had become clear to them they bowed at the feet of that good man and pressed their palms together in reverence.He taught them the Dharma, and they all attained the stage of irreversibility from the spirit of unexcelled, perfect enlightenment.It occurred to me then, the disciples, who do not know the thoughts or the inclinations of others, are not able to teach the Dharma to anyone.Why?These disciples are not experts in discerning the superiority and inferiority of the spiritual faculties of living beings, and they are not always in a state of concentration like the Tathāgata, the Saint, the perfectly accomplished Buddha.

Vkn Eng 29 3.3 Prajñā
The Sanskrit term prajñā is formed by a prefix pra-, meaning "supreme" or "premium" and the root jñā meaning "consciousness", "knowledge" or "understanding" (Keown 2003).It is defined as the intelligence that comprises all the higher faculties of cognition and intellect as conversant with general truths (David & Stede 1921).Prajñā was translated as Hui or Zhi Hui or transliterated as Bo Re波若, Bo Re鉢若, Pan Luo Re般罗若, Bo Luo Re波罗若, Bo Shen Rang鉢肾穰, or Bo Luo Zhi Rang鉢啰枳穰in early Buddhist manuscripts.In his translation of Abhidharma-kośa-bhāṣya, Xuan Zang commented on this principal term that a person owing prajñā can choose the right way based on his/her insights.Those who lack prajñā are of moha (ignorance), which is considered to be the root cause of all suffering.As an essential term in Buddhist sutras referring to the supreme intelligence, it is undoubtfully selected as a term that is better transliterated but not translated in Fa Yun's statements.
Comparatively, Xuan Zang chose to transliterate it as a whole into Bo Re Bo Luo Mi Duo 般若波羅蜜多 for its abundant Buddhist implications.
The term prajñā hits 53 times totally in VNS, as details shown in Table 3. Kumarajiva mainly employed Hui 慧 (17 times) or Zhi Hui智慧 (17 times) in his version.Comparatively, Xuan Zang flexibly processed his translation.Evidently, Xuan Zang didn't concentrate on transliterating the term as Bo Re 般若, but preferred a single Chinese word Hui 慧 or a phrase Miao Hui 妙慧 to translate it.

Discussion
We have examined three typical terms mentioned in Fa Yun's statement in the previous section.The findings show that the translation of these terms didn't strictly follow the so-called Five Transliterations.There seems to be a tendency that Xuan Zang would translate their internal meanings rather than transliterate them.In contrast, Kumarajiva's translation, to some extent, was more likely to follow the principle of Five Transliterations.The observation weakens the evidence that it was Xuan Zang who established the principle of the Five Transliterations, for its contradictory that one proposed translation strategies that s/he didn't follow at all.This is in line with Murata's (1975) suspicion based on his historical document review.The data in Section 3 reveals that Xuan Zang didn't follow what Fa Yun summarized as Buddhist translation conventions.
In order to have further insight on Xuan Zang's translation strategy on the terms illustrated above, we compare the translations/transliterations of the items of polysemy, of the old establishment, and of Buddhist implications with another three important sutras, namely Bhaiṣajya-guru-vaiḍūrya-prabhāsa-tathāgata-pūrva-praṇidhāna-viśeṣa-vistara-guṇa-sūtra The data in Table 4 reveal that Xuan Zang had the intention of interpreting the terms of adopting the phrases whose semantic meaning can be derived word by word, in comparison to using the transliterated ones.A typical example that violates the Five Transliteration is Bhagavān, which was quite frequently translated into Shi Zun世尊(literately meaning "the one who is respected by the world") or Fo Shi Zun佛世尊 ("Buddha who is respected by the world") instead of its transliteration Bo Jia Fan薄伽梵.
Meanwhile, the transliterations kept by Xuan Zang were limited in amount and distributed regularly in the texts.As described in Section 3.1, Xuan Zang kept the transliteration of the Sanskrit word Bhagavān at the beginning and the ending of the VNS for his respect for the Buddha.In the main texts of the sutras, bhagavān was systematically interpreted into a phrase more understandably with respect by the Chinese audience: Shi Zun世尊.The other two examples, i.e., anuttarā-samyak-saṃbodhi and Prajñā were handled similarly.We could speculate Xuan Zang's translation strategies on Buddhist terms through these examples.
On the one hand, Chinese scholars have focused on written characters that convey meanings, in contrast to Indian scholars whose focus was on sound 8 .Therefore, translators needed to find suitable combinations of Chinese characters to match both form and meaning.
The translation of Buddha's words from Sanskrit to Chinese was for people's better understanding of Buddhism rather than having Buddhist sutras isolated from its followers.His calling of Buddhist transmission resulted in his tendency of interpreting the items instead of transliterating them.Our speculation is also trackable from Yuan Hui's comments on Xuan Zang's translation of Abhidharma-kośa-bhāṣya, in which Yuan Hui mentioned that Xuan Zang's interpretations were no different from the original texts, and the discourses were of no errors.Those who were confused in reading previous translation versions could be clarified by reading Xuan Zang's translation 9 .Therefore, it seems that Xuan Zang had a purposeful attitude towards the translation or transliteration of given Buddhist terms.
Following our speculation, we regard the Five Transliterations as an inaccurate recapitulation of Buddhist translation strategies sorted by later generations.It contributed to the development of translation theories in China, even though it was not perfectly summarized.The transcriptions for the sounds of Indic words were not deliberately kept in Xuan Zang's versions.Besides, the Chinese characters recording Sanskrit sounds in Xuan Zang's versions were commented to be away from their predecessors' renderings, such as Kūkai's 空海 complaints that the mandarin pronunciations of mantras were far from the Sanskrit sounds it was meant to invoke.Lustaus (2019) argued that such comments are unaware of the historical phonetic change of Chinese characters over time and regions.
A coming question is how we could better understand the tension between historically recorded translation theories (such as Fa Yun's assertion) and practice (such as Xuan Zang's operation).Xuan Zang dedicated himself to the translation missions for the rest years after he returned from India.However, very few records are found about his experiences written by him on his initiative.Da Tang Xi Yu Ji 大唐西域記, the famous work about his traveling to India was under the requirements from the Emperor.Some of his words were recorded by his disciples in Da Tang Ci En Si Xuan Zang Fa Shi Zhuan 大唐慈恩寺玄奘法師傳.By the time he passed away, the translation work hadn't been done.Therefore, it is reasonable to deduce that Xuan Zang was not likely to have his own translation thoughts or perceptions documented in his lifetime, let alone translation theories.Even though there seemed to be no close associations between Xuan Zang's translation practice and translation theories under his name, we could conjecture that Xuan Zang's disciples or later followers might sort out translation strategies or policies by comparing Xuan Zang's versions and previous versions.Furthermore, the possibility should also not be excluded that later generations created their thoughts on Buddhist translation but misleadingly put them under Xuan Zang's name.As what Murata (1975) indicates, Xuan Zang had been regarded as "an existence in legend (伝説的存在)" from the Song Dynasty to the Ming Dynasty.Therefore, it was not unexpected that ideas such as the Five Transliterations were attributed to him in history.

Conclusion
So far, we have investigated how Xuan Zang processed Buddhist terms with a special focus on the Sanskrit-Chinese Parallel Corpus of Vimalakīrti Nirdeśa Sūtra.It is found that Xuan Zang seemed not to follow the Five Transliterations in his translation practice.We attempt to provide evidence that the Five Transliterations were not proposed by Xuan Zang, even though Fa Yun asserted so in the documents.Xuan Zang's translation strategies of processing Buddhist terms are nevertheless worthy of exploring.We speculate that Xuan Zang endeavored to balance the faithfulness and the readability in his translations, especially represented in the processing of Buddhist terms.He kept transliteration of terms in relatively fixed positions (generally at the beginning and the ending of a sutra) to show his respect and faith, meanwhile translated such terms in the texts to facilitate the followers' understanding.
Although the present study has yielded findings that have implications on people's understanding of Xuan Zang's insights of translation, its design is not without flaws.As mentioned at the beginning of this paper, this study is based mainly on two selfbuilt corpora: the Corpus of Buddhist Translations by Xuan Zang (CBTX) and the Sanskrit-Chinese Parallel Corpus of Vimalakīrti Nirdeśa Sūtra (SCPC-VNS).The result could be more convincing given all the Sanskrit texts and the parallel Chinese versions translated by Xuan Zang are included in a single large-scaled corpus.Even though most of the Sanskrit texts of Buddhist sutras are not found, the currently available resources are already very large.We are hopeful that future research based on corpora will provide more detailed results.Quantitative analysis of multiple translations is expected from both linguistic and rhetorical approaches.

Table 1 .
. The translation of Bhagavān in Xuan Zang's version and Kumarajiva's version is shown in

Table 1 .
The translation of bhagavānAs shown in Table1, Xuan Zang transliterated bhagavān into Bo Jia Fan (薄伽梵) only twice but translated it into Shi Zun(世尊, lit. the one who is respected by the world), Fo Shi Zun(佛世尊, lit.Buddha that is respected by the world), in most instances.Comparatively, there is no transliteration of bhagavān in Kumarajiva's version.

Table 4 .
'The Sutra on the Original Vows and Merits of the Medicine Master Lapis Lazuli Light Tathagata' (Yao Shi Liu Li Guang Ru Lai Ben Yuan Gong De Jing藥師琉璃光如來本願功德經), Yogācārabhūmi-śāstra 'On the Foundation for Yoga Practitioners' (Yu Qie Shi Di Lun 瑜伽师地论) and Abhidharma-kośa-bhāṣya 'Verses on the Treasury of Abhidharma' (A Pi Da Mo Ju She Lun阿毗达磨俱舍论), as shown in Table 4.The comparison of frequencies of items in various sutras.(* indicates the transliterated candidate that is proposed in Fa Yun's assertion of the Five Transliterations)