ancestor, male warrior, and maleness have been derived” (all italics in the quotations here are added by me).39 Such intrinsic maleness of the deer image on the Steppe resonates naturally with the Mulan-Xiong name-style correspondence cited earlier.
Chinese Transcription Notes
Despite the strong evidence for the existence of a “Barbarian” unicorn name in the Tuoba realm, a phonetic difficulty for linking Pulan/Mulan to this name is the Middle Chinese ending -k in the first character. It is therefore legitimate to question whether the Chinese forms could have been a transcription of bulān or its variants, which does not seem to show any trace of a velar.
The two Chinese characters of concern, namely pu and mu, belonged to the old rusheng, “entering tone,” category, characterized by a final stop (-t, -k, -p), which still exists today in several southern dialects, especially Cantonese. Nearly fifty years ago Edwin Pulleyblank published a study40 demonstrating the wide use of “entering-tone” characters to represent a single consonant in a foreign consonant cluster or a single final consonant. An additional conclusion of his is that the Chinese name Tujue may have simply represented Türk, not the commonly assumed Mongolic plural form *Türküt ever since the time of Joseph Markwart and Paul Pelliot, prompted by the final -t in the Chinese transcription.
This last conclusion faces a technical difficulty that, in all Pulleyblank's well-substantiated examples of foreign consonant clusters, the rusheng character always transcribed the first consonant rather than the final as the case of Türk would require. This point notwithstanding, his general contention that such a character so often transcribed a single consonant is certainly valid. What I propose is to advance his observation one step further. If a rusheng character could represent only its initial consonant in transcribing a foreign word, it would seem no less natural that it could also represent its initial consonant plus the vowel, that is, the full syllable minus the final stop.
A careful study of the vast Chinese transcription data confirms that such uses of rusheng characters, which, by the way, cannot be easily explained by assimilation or other types of absorption into the next syllable, indeed abound. Space considerations prevent me from listing many more than a few illustrative cases most pertinent to my study. Unpronounced final stops are highlighted in boldface. To avoid extensive notes, except as otherwise indicated, the Sanskrit transcription data are based on the several Sino-Sanskrit glossaries in the Taishō Tripitaka and P. C. Bagchi's two-volume study of Sanskrit-Chinese transcriptions.
• The ancient Central Asian city transcribed as Mulu (muk-luk) (Hou Han shu 88.2918 and Xin Tang shu 221b.6245) has been identified by almost all scholars (F. Hirth, E. Chavannes, P. Pelliot, etc.) as the modern city of Merv, known as Mûlu in ancient times.41
• The Sanskrit name Bhuţa for Tibet (Bod) was transcribed as Puzha (b'uk-ta)42
• Here is a sample of cases from Taishō Tripitaka (Chinese transcription < Sanskrit word):
◊ boqifu (puât-kiət-b'iwak) pakva, “cooked, ripe.”
◊ yutaimo (jiu-t'âi-muât) < utma, “high.”
◊ shejiedi (siät-kiät-tiei) < çakti, “spear.”
◊ salishabo (sät-lji-şat-puâ) < sarşapa, “mustard.”
◊ nalameluo (nâp-lât-muâ-lâ) < duravala/durbala, “weak.”43
◊ naqu (nâp-kiwo) < dukha, “pain, hardship.”
• Even the famous Buddhist pilgrim-scholar Xuanzang (600–664), well known for having mastered the Sanskrit language and for his rigor in translating Buddhist sutras (as well as for likely being responsible for translating a sutra from Chinese back to Sanskrit)44 left many such cases. Two examples are juduo(k'iuk-tâ) for gupta in many personal names (the Pali form gutta cannot reconcile with the final -k of the first character either), and biboluo (piët-puět-lâ) for pippala, “pepper.”45
• The “Nine (Uighur) Tribes” as recorded in the two official Tang histories46 provide us with a telling pair: Yaoluoge (iak-lâ-kât) and Yaowuge (iak-miuat-kât). The first was attested in the Saka portion of the famous Staël Holstein scroll as yah:idakari and in Uighur Runic inscriptions as yaγlaqar. The second, however, was rendered as yabūttikari in the Holstein scroll. W. B. Henning, for one, already wondered how the same Chinese character yao (iak) could be used in rendering both yaγ and ya.47
• Pelliot48 has identified the Pa-lan-ba Sum-pa people (-ba is a suffix) mentioned in the Tibetan version of the Inquiry of Vimalaprabha with the Bailan (b 'vk -lân) tribe in Chinese records.
It should be noted that the first two examples were about the same initial characters as in the name Pulan/Mulan, whereas the last example is a striking parallel to our contention that Chinese b'uk-lân ~ Altaic bulān.
What may be more revealing is a group of “Barbarian” tribe names with known variants:49
1. Chili (tsiet-lji) ~ Chilie (tsiet-liät).
2. Qifu (k'iat-b'iu) ~ Qibu (k'iat-b'uo) ~ Qifo (k'iat-b'iuat) ~ Qifu (k'iat-b'iuk).
3. Pugu (b 'uk-kuo') ~ Pugu (b 'uk-kuat).50
4. Dabu (dât-b'uo) ~ Dabo (dât- b'uat).
5. Hesui (γâ-zwi) ~ Heshu (γâ-dz'iuět).
6. Mozhe (mâk-tsia) ~ Mozhe (mâk-tśiât).51
At first glance, the forms with a final -t would seem to strengthen the old Markwart-Pelliot theory that the Chinese name Tujue for Türk represented some (proto-)Mongol plural form *Türküt. But after closer examination, in addition to the equivalent forms with an open final syllable, a (proto-)Mongol intermediary cannot be substantiated in most cases, much less accommodate the difficulty with the -k endings and the Qiang/Tibetan names (6). On the contrary, these variants lend strong support to the contention that a rusheng's final stop corresponded to an open syllable in many a Chinese rendition of foreign words. The frequent -t ending can easily be explained by the simple fact that, among the rusheng characters, this type is more numerous, at least in transcriptions. One may note, for instance, that one “Barbarian” clan name was recorded by two different dynastic histories as Fulugu (b'iuk-luk-ku) and Buliugu (b'uo-luk-ku) respectively,52 a cognate of Bulgar, the name of the people who founded the medieval Bulgarian kingdom, as we shall see. The avoidance of initial consonant clusters by the Altaic languages makes it clear that the first character in these two Chinese renditions must be transcribing more than just a consonant b-. Incidentally, these cases and the Saka form bākū for the tribe name (3) Puku (b 'uk-kuat) strongly support Clauson's seemingly idiosyncratic reading Türkü as the Old Turkic form of the name Türk, a notion heavily criticized by, among others, Pulleyblank.53
The clan name transcription Fulugu, by the way, reminds me of the use of the first character fu (b'iuk) as a frequent alternative, especially in Buddhist theophoric names, to the character fo (b'iuat), the standard transcription of “Buddha.” The final stop -k in the former, therefore, had to be silent in such transcriptions.
My thesis that, in transcribing a foreign word, the medieval Chinese “enteringtone” characters often have a silent final stop is in fact not new. After examining similar data, J. Harmatta also reached the conclusion that the Chinese rusheng syllables “must have had two phonetic variants” (CVC and CV), so that “[these] two phonetic realizations of the [rusheng] syllables permitted their alternative use for representing both foreign syllables with final stop and those without it.”54
In addition, more than sixty years ago, while studying the Uighur