宋高僧傳 3.36

3.36

二、雪山之北是胡，山之南名婆羅門國，與胡絕，書語不同。從羯霜那國字源本二十餘言，轉而相生，其流漫廣，其書竪讀同震旦歟！至吐貨羅，言音漸異，字本二十五言，其書橫讀。度葱嶺南，迦畢試國言、字同吐貨羅，已上雜類為胡也。若印度言、字，梵天所製，本四十七言，演而遂廣，號《青藏》焉。有十二章，教授童蒙，大成「五明論」，大抵與胡不同。五印度境彌亘既遙，安無少異乎？又以此方始從東漢傳譯，至于隋朝，皆指西天，以為胡國，且失梵天之苗裔，遂言胡地之經書。彥琮法師獨明斯致，唯徵造錄，痛責彌天。符《佛地》而合《阿含》，得之在我，用胡名而迷梵種，失則誅誰？唐有宣公，亦同鼓唱。自此若聞彈舌，或覩黑容，印定呼為梵僧，雷同認為梵語。琮師可謂忙於執斧捕前白露之蟬，瞢在迴光照後黃衣之雀。既云西土有梵、有胡，何不南、北區分？是非料簡，致有三失：一改胡為梵，不析胡開，胡還成梵，失也；二不善胡、梵二音，致令胡得為梵，失也；三不知有重譯，失也。當初盡呼為胡，亦猶隋朝已來總呼為梵，所謂過猶不及也。如據宗本而談，以梵為主；若從枝末而說，稱胡可存。何耶？自五天至嶺北，累累而譯也，乃疑琮公留此以待今日，亦不敢讓焉。

Second, north of the Snow Mountains is Hu; south of the mountains is called the land of Brahmans, which is completely separate from Hu, their scripts and spoken languages being different. From the country of Kesh, the script originally had twenty-odd letters, which shifted and generated [others], their variants spreading widely; its script is read vertically, just like that of Cīnasthāna! Reaching Tokhara, the spoken sounds gradually differ, its script having originally twenty-five letters, read horizontally. Crossing south of the Congling Mountains, the language and script of the country of Kapiśa are identical to Tokhara; the miscellaneous kinds above are Hu. As for the language and script of India, they were created by Brahmā, originally having forty-seven letters, which were then expanded and widely propagated, called the Śabdavidyā. It has twelve chapters to teach beginners and culminated in the five treatises of the five sciences (pañcavidyā), being in general different from Hu. Given the vast extent of the territory of the five Indias, how could there not be slight variations? Furthermore, from the time when translation first began in this land during the Later Han down to the Sui dynasty, people all pointed to the Western Heavens and regarded them as Hu countries, thereby losing the lineage of Brahmā and speaking of them as scriptures of the Hu lands. Dharma Master Yancong alone clarified this principle: he merely examined the catalog and severely reproached Mitian (Daoan). In according with the Buddhabhūmi and matching the Āgamas, the gain was ours; but in using Hu names and confusing the Brahman stock, upon whom should the fault be blamed? In the Tang there was Master Daoxuan, who likewise joined in this chorus. From this point on, if people heard the trilling of a tongue or saw a dark visage, they would firmly designate the person a Sanskrit monk, and uniformly identify [the tongue] as the Sanskrit language. Master Yancong may be described as being so busy wielding the axe to catch the cicada before the white dew that he was blind to the light glancing upon the yellow-coated sparrow behind him. Since it is said that the Western lands have both Sanskrit and Hu, why did he not distinguish between north and south? Through his analysis of right and wrong, three errors arose: first, changing Hu into Sanskrit without analyzing and unpacking Hu, so that Hu reverted to Sanskrit—this is an error; second, not being skilled in the two sounds of Hu and Sanskrit, thereby causing Hu to be taken for Sanskrit—this is an error; third, not knowing that re-translations existed—this is an error. Originally calling everything Hu was just like calling everything Sanskrit since the Sui dynasty: this is what is called going beyond being just as bad as falling short. If speaking on the basis of the fundamental source, one takes Sanskrit as primary; if speaking from the branches, the designation Hu may be preserved. Why is this? From the five Indias to the north of the mountains, texts were translated through multiple stages; thus I suspect Master Yancong left this to await today, and I do not dare decline [to clarify it].