出三藏記集 1.31

1.31

夫神理無聲。因言辭以寫意。言辭無跡。緣文字以圖音。故字為言蹄。言為理筌。音義合符不可偏失。是以文字應用彌綸宇宙。雖跡繫翰墨而理契乎神。昔造書之主凡有三人。長名曰梵。其書右行。次曰佉樓。其書左行。少者蒼頡。其書下行。梵及佉樓居于天竺。黃史蒼頡在於中夏。梵佉取法於淨天。蒼頡因華於鳥跡。文畫誠異。傳理則同矣。仰尋先覺所說。有六十四書。鹿輪轉眼筆制區分。龍鬼八部字體殊式。唯梵及佉樓為世勝文。故天竺諸國謂之天書。西方寫經雖同祖梵文。然三十六國往往有異。譬諸中土猶篆籀之變體乎。案蒼頡古文沿世代變。古移為籀。籀遷至篆。篆改成隷。其轉易多矣。至於傍生八體則有仙龍雲芝。二十四書則有揩奠鍼殳。名實雖繁為用蓋尠。然原本定義。則體備於六文。適時為敏。則莫要於隷法。東西之書源亦可得而略究也。至於胡音。為語單複無恒。或一字以攝眾理。或數言而成一義。尋大涅槃經列字五十。總釋眾義十有四音。名為字本。觀其發語裁音。宛轉相資。或舌根脣末。以長短為異。且胡字一音不得成語。必餘言足句。然後義成。譯人傳意豈不艱哉。又梵書製文有半字滿字。所以名半字者。義未具足。故字體半偏。猶漢文月字虧其傍也。所以名滿字者。理既究竟。故字體圓滿。猶漢文日字盈其形也。故半字惡義以譬煩惱。滿字善義以譬常住。又半字為體。如漢文言字。滿字為體。如漢文諸字。以者配言方成諸字。諸字兩合即滿之例也。言字單立即半之類也。半字雖單為字根本。緣有半字得成滿字。譬凡夫始於無明得成常住。故因字製義以譬涅槃。梵文義奧皆此類也。是以宣領梵文寄在明譯。譯者釋也。交釋兩國。言謬則理乖矣。自前漢之末。經法始通譯音胥訛。未能明練。故浮屠桑門言謬漢史。音字猶然。況於義乎。案中夏彛典誦詩執禮師資相授猶有訛亂。詩云。有菟斯首。斯當作鮮。齊語音訛遂變詩文。此桑門之例也。禮記云。孔子蚤作。蚤當作早。而字同蚤𧌡。此古字同文。即浮屠之例也。中國舊經而有斯蚤之異。華戎遠譯何怪於屠桑哉。若夫度字傳義則置言由筆。所以新舊眾經大同小異。天竺語稱維摩詰。舊譯解云無垢稱。關中譯云淨名。淨即無垢。名即是稱。此言殊而義均也。舊經稱眾祐。新經云世尊。此立義之異旨也。舊經云乾沓和。新經云乾闥婆。此國音之不同也。略舉三條。餘可類推矣。是以義之得失由乎譯人。辭之質文繫於執筆。或善胡義而不了漢旨。或明漢文而不曉胡意。雖有偏解終隔圓通。若胡漢兩明意義四暢。然後宣述經奧於是乎正。前古譯人莫能曲練。所以舊經文意致有阻礙。豈經礙哉。譯之失耳。昔安息世高。聰哲不群。所出眾經質文允正。安玄嚴調既亹亹以條理。支越竺蘭亦彬彬以雅暢。凡斯數賢並見美前代。及護公專精兼習華戎。譯文傳經不𮘧于舊。逮乎羅什法師俊神金照。秦僧融肇慧機水鏡。故能表發揮翰克明經奧。大乘微言於斯炳煥。至曇讖之傳涅槃。跋陀之出華嚴。辭理辯暢明踰日月。觀其為義繼軌什公矣。至於雜類細經多出四含。或以漢來。或自晉出。譯人無名莫能詳究。然文過則傷艶。質甚則患野。野艶為弊同失經體。故知明允之匠難可世遇矣。祐竊尋經言異論呪術言語文字皆是佛說。然則言本是一。而胡漢分音。義本不二。則質文殊體。雖傳譯得失運通隨緣。而尊經妙理湛然常照矣。既仰集始緣故次述末譯。始緣興於西方。末譯行於東國。故原始要終寓之記末云。

Now, the divine principle is soundless; it relies upon spoken words to convey meaning. Spoken words are traceless; they depend on written characters to depict sounds. Therefore, characters are the snare for words, and words are the trap for principle. When sound and meaning tally, neither may be neglected. Thus the practical application of writing encompasses the universe: though its traces are bound to brush and ink, its principle accords with the divine. In ancient times, there were three creators of scripts. The eldest was named Fan (Brahmā); his script ran to the right. The next was called Kharostha; his script ran to the left. The youngest was Cangjie; his script ran downward. Brahmā and Kharostha lived in India; the Yellow Emperor's scribe Cangjie was in the central lands of China. Brahmā and Kharostha took their model from the Pure Devas; Cangjie drew his patterns from bird tracks. The strokes of their writing indeed differed, but in conveying principle they are identical. Looking upward to inquire into what previous enlightened ones have expounded, there are sixty-four scripts: divided by the styles of writing like the deer-wheel and revolving eyes, while the dragon and ghost eightfold classes have distinct forms of characters. Only Fan and Kharostha are the supreme scripts of the world; therefore, the various countries of India call them divine scripts. Although the copying of sūtras in the western regions all traces its ancestry to the Fan script, the thirty-six kingdoms often show differences, comparable perhaps to the variant forms of seal and large-seal scripts in the Central Lands. When we examine the ancient script of Cangjie, it changed successively through dynasties: the ancient moved to large-seal, the large-seal shifted to small-seal, and the small-seal was altered into clerical script; its transformations were many. As for the eight derivative styles, there are the immortal, dragon, cloud, and lingzhi forms; among the twenty-four scripts, there are the stamped, set, needle, and spearhead types. Though the names and realities are numerous, their practical use is very limited. Yet when we trace the origin and establish the definition, their forms are fully provided in the six types of characters; and in being quick to adapt to the times, nothing is more essential than the clerical method. The sources of eastern and western writing can thus also be briefly investigated. As for the sounds of Hu speech, their single and compound expressions have no constancy: sometimes a single character encompasses many principles; sometimes several words form a single meaning. Examining the Mahāparinirvāṇa-sūtra, it lists fifty characters, fully explaining manifold meanings in fourteen vowels, calling them the roots of characters. Observing their emission of speech and articulation of sounds, they assist one another in modulating turns, differing by long and short, or by the root of the tongue and the tip of the lips. Moreover, a single sound in Hu characters cannot form an utterance: it must require additional words to complete the phrase, only after which the meaning is achieved. How could it not be difficult for translators to convey the intended meaning? Furthermore, Fan writing constructs texts with half-letters and full-letters. The reason they are called half-letters is that their meaning is not yet complete, so the body of the character is half-formed and partial, just as the Chinese character for 'moon' is deficient at its side. The reason they are called full-letters is that their principle is already ultimate, so the body of the character is complete and full, just as the Chinese character for 'sun' is filled in its shape. Therefore, half-letters have a negative sense to illustrate the afflictions (kleśa), while full-letters have a positive sense to illustrate the permanently abiding. Again, the body of a half-letter is like the Chinese character 'yan' (言); the body of a full-letter is like the Chinese character 'zhu' (諸). Only when 'zhe' (者) is paired with 'yan' (言) is the character 'zhu' formed; 'zhu' being a combination of the two is an example of the full. The single, independent standing of 'yan' is of the class of the half. Although the half-letter is singular, it is the fundamental root of characters: because of the existence of the half-letter, the full-letter can be formed. This is like how an ordinary person begins in ignorance and is able to attain the permanently abiding. Therefore, meanings are formulated from characters to illustrate nirvāṇa. The profound meanings of Fan literature are all of this kind. For this reason, the proclamation and comprehension of Fan texts depends upon clear translation. Translation (yi) means explanation (shi): mutually explaining between two countries; if the words are erroneous, the principle will be violated. From the end of the Former Han, when the scriptural Dharma first circulated, the translated sounds were all corrupt, not yet clearly refined. Thus 'futou' (Buddha) and 'sangmen' (śramaṇa) were mistakenly recorded in the Han histories; if sounds and characters were like this, how much more so the meanings! Examining the classical canons of the Central Lands, when chanting the Odes and upholding the Rites as transmitted from teacher to disciple, there were still errors and confusion. The Odes say: 'There is a hare, here is its head.' 'Si' (斯) should be 'xian' (鮮); because the sound in the Qi dialect was corrupt, it ended up altering the text of the Odes. This is an example of the 'sangmen' kind. The Book of Rites says: 'Confucius rose early (zao 蚤).' 'Zao' (蚤) should be 'early' (zao 早), yet the character is written identically to the flea (zao 蚤). This is an instance of an ancient character sharing the same form, which is an example of the 'futou' kind. If the ancient scriptures of the Middle Kingdom already had such variations as 'si' and 'zao', what wonder is there in 'tou' and 'sang' when translating from distant barbarian lands? As for conveying meanings by translating words, the choice of phrasing depends upon the pen. This is why the various new and old sūtras are broadly similar yet slightly different. In the language of India, one says Vimalakīrti: old translations explain it as 'Stainless Fame' (Wugoucheng); translations from the Guanzhong region say 'Pure Name' (Jingming). 'Pure' is simply 'stainless'; 'name' is simply 'fame'. Here the words differ while the meaning is equal. Old sūtras say 'Zhongyou' (All-Blessing); new sūtras say 'Shizun' (World-Honoured One). This is a difference in conceptual formulation. Old sūtras say 'Gantaphe' (Gandharva); new sūtras say 'Qiandapo'. This is a difference in regional phonetic transcription. Having briefly cited three items, the rest can be deduced by analogy. Therefore, the gain or loss of meaning depends upon the translator; whether the diction is plain or ornate rests with the scribe who holds the brush. Some are good at the Hu meaning but do not understand the Chinese intent; some are clear in Chinese text but do not comprehend the Hu sense. Though possessing partial understanding, they remain ultimately obstructed from full penetration. If both Hu and Chinese are clearly understood, and the meaning and principle flow in all four directions, only then can the profound mysteries of the sūtras be expounded and rectified. In ancient times, translators were unable to be thoroughly proficient, which is why the textual meaning of old sūtras came to have obstacles. How could it be that the sūtras themselves were obstructed? It was simply the failing of the translators. In the past, An Shigao of Parthia was exceptionally intelligent and outstanding; the various sūtras he produced were balanced and proper in plainness and ornamentation. An Xuan and Yan Fotiao were tireless and methodical; Zhi Qian and Kang Senghui were refined, elegant, and fluent. All these worthy men were admired in former generations. When it came to Master Dharmarakṣa, he was dedicated and meticulous, mastering both Chinese and barbarian traditions; his translation of texts and transmission of sūtras did not fall short of the old masters. Coming to Dharma Master Kumārajīva, his brilliant spirit shone like gold; the Qin monks Sengrong and Sengzhao had wisdom like a clear water mirror; therefore, they were able to reveal and write out [the texts], fully clarifying the sūtras' profound depths, so that the subtle words of the Great Vehicle shone brightly here. As for Dharmakṣema's transmission of the Nirvāṇa and Guṇabhadra's production of the Avataṃsaka, their phrasing and reasoning were fluent and perspicuous, outshining the sun and moon; observing their rendering of meanings, they followed in the tracks of Master Kumārajīva. As for the miscellaneous small sūtras, most were taken from the four Āgamas; some arrived under the Han, while some came out under the Jin; their translators are nameless, so they cannot be investigated in detail. However, if the text is overly ornate, it harms the elegance; if it is too plain, it suffers from coarseness. Both coarseness and over-ornateness are defects that equally lose the substance of the sūtra. Hence we know that masters who are discerning and faithful can rarely be met with in an age. I, Sengyou, note in examining the sūtras that heterodox treatises, mantras, speech, and scripts are all spoken by the Buddha. Thus, speech is fundamentally one, yet Hu and Chinese separate its sounds; principle is fundamentally non-dual, yet plainness and ornamentation differ in style. Although the success and failure of translation vary with conditions, the wondrous principle of the revered sūtras remains serenely, ever shining. Having respectfully compiled the initial circumstances, I next set forth the later translations. The initial circumstances arose in the Western Regions; the later translations circulate in the Eastern Land. Therefore, tracing the origin and summarizing the end, I lodge them at the end of this record.