e Syllable yi in Old Japanese
Tom Dougherty Department of Linguistics University of Hawai‘i at Mānoa December ,
Introduction
It is widely held that there was no syllable yi or wu in Old Japanese (cf. Vovin , p. ). However, Vovin () discusses the possibility of both yi and wu, providing sev- eral examples in whi we can infer yi and wu (Vovin , p. ). Since this paper is looking at yi and not wu, we shall set discussion relating to the possible syllable wu aside. ere are two possible solutions via triangulation whi we could use to solve the question of whether Old Japanese had the syllables yi and wu. e first would be a Middle Chinese transcription of Old Japanese whi unambiguously uses yi as opposed to i and wu as opposed to u. e second would be a loan from Old Japanese into Ainu whi is wrien with i in Old Japanese but ye in Ainu (note that Ainu lowers /i/ to [e] before /j/ and /w/), or with u in Old Japanese but wu in Ainu.¹ As far as I am aware, neither a Middle Chinese nor an Ainu triangulation is to be found. Further, there may be serious accuracy issues with any contemporary Chinese transcription of Old Japanese (Miyake , pp. -). Since these “magic bullet” solutions are not to be found, we must rely soley on in- ternal evidence. e main pieces of internal evidence are first i wrien in violation of the phonotactic rules of Old Japanese (i.e. [C]VCVCV…), and second, a seemingly free variation between i and yu in various words (Vovin , pp. -):
¹While the segment ye—from both *ye and *yi diaronically—is relatively common in Ainu, the segment wu is extremely rare.
() 和我勢故波 / 多麻尓母我毛奈 / 手尓麻伎鵜 / 見都追由可牟乎 / 於 吉鵜伊加婆 乎思 wa -Nka se -kô pa tama n- -i möNkamô na TE -ni I - beloved - jewel - - arm - mak- -î -te MÎ -tutu yuk- -am- -u -wo ok- wrap- - - look()- - go- -- - - leave- -î -te ik- -aNpa wosi - - go- - be.regreable “[I] want my beloved to be a jewel! Although [I] would go wrapping [him] around my arm and looking at [him], [it] is regreable if [I] go leaving [him].” ( : ) (Vovin , p. )
In this paper, I will further explore whether or not we have reliable evidence for the existence of the syllable yi in Old Japanese. First, I will examine the man’yōgana used for both i and yu, looking at both their frequency across the corpus as well as their reconstructed forms in Early Middle Chinese. en I will examine the verb yuk- /ik- (“to go”), again looking at the raw frequencies of yuk- versus ik-, as well as some distributional peculiarities. Finally, I will look at the various words whi vary between yu and i in the Nihonshoki Kayō, because there is reason to believe that it would provide the most persuasive Old Japanese-internal evidence.
Man’yōgana for i and yu
I consulted both digitized as well as critical versions of the following texts: the Kojiki Kayō (), the Nihonshoki Kayō () , the Fudoki Kayō () , the Man’yōshū (), and the Bussoku Seki Ka (). I ose to exclude other Old Japanese texts, notably the Norito and the Senmyō, as these texts are extremely long, but have very few man’yōgana. I also omied man’yōgana used to write proper nouns whi continued to use the same Chinese aracters even aer the use of man’yōgana ceased. I feel that these fall into a special class of not-quite ongana, not-quite logograms. Further, only one place name that I know of is of any interest to the present study: Yukî Island, mentioned several times in the Man’yōshū and once in the Nihonshoki Kayō, is called Iki Island in modern Japanese. I consulted with Takagi, Gomi, and Ōno (; ; ; ) for the Man’yōshū, and Tsuihashi and Konishi () for all other texts. ere are, of course, limitations to this; some texts are interpreted differently in different critical editions, and critical editions oen la script variants from different manuscript traditions. For this present brief study, however, I do not feel this should not be too mu of an issue. My oices in terms whi man’yōgana forms to include follow directly from Vovin (, pp. -). Many forms given in Vovin () are unaested in the digitized cor- pus I use. is is almost certainly because not every manuscript tradition is represented in my corpus of Old Japanese. If I had consulted manuscript sources, these numbers would certainly be different, but I am fairly confident that the overall paern would still hold.
Man’yōgana total 伊 *ʔjij , 以 *yiX 異 *yiH 移 *ye 因 *ʔjin 印 *ʔjinH 夷 *yij 怡 *yi 壱 *ʔjit
Table : A list of ongana man’yōgana for i by frequency.
Man’yōgana total 由 *yuw 喩 *yuH 遊 *yuw 油 *yuw 愈 *yuX 瑜 *yu 臾 *yu 庾 *yuX 踰 -
Table : A list of ongana man’yōgana for yu by frequency.
e Early Middle Chinese () values presented follow Baxter (). ese forms are especially interesting for examining the Nihonshoki Kayō, as the writing system had not gone through a Middle Korean intermediary like in all of the other texts (Vovin , pp. -). e Early Middle Chinese forms fall into two categories: those that have a palatal approximant as an initial and those that have a gloal stop as an initial (Baxter , pp. -). As one might expect, the man’yōgana for yu are all some variation of *yu in Baxter’s reconstruction of Early Middle Chinese (Baxter , pp. -). Curiously, of the man’yōgana for i aested in my corpus, only 伊 begins with what Baxter re- constructs as a gloal stop (Baxter , p. ). e rest agree in their initial, *y, but differ in their rhyme: 以 is reconstructed as *yiX, 異 is reconstructed as *yiH, and 移 is reconstructed as *ye (Baxter , pp. -). Note that Baxter () does not give a reconstruction for the aracter 踰 , but it is likely something along the lines of *yu, agreeing with the other aracters with similar phonetic radicals. My conclusions here are summarized in Table and Table .
i-yuk- , , , : ; : , ; : , ; : uti-yuk- : ; : ; : taNpî yuk- : ; : , , , ; : , ; : ; : , ,
Table : A listing of various co-occurrences with the verb yuk-. Multiple instances of the same poem indicate multiple co-occurrences in the poem.
Distributional Peculiarities of yuk-/ik-
It goes without saying that the absence of evidence is not the evidence of absence. However, there are some distributional peculiarities whi, when coupled with the orig- inal Middle Chinese reading of the man’yōgana involved, could lend weak support for a syllable yi for the verb yuk-/ik- in Old Japanese. In this section I deal with this verb alone because it is by far the most frequent of all of these yu/i variable words, and thus provides mu more information with whi we can make inferences. Note that this listing is not intended to be exhaustive; there are likely other distributional oddities.
. e Prefix i- When the verb yik- is prefixed with the directive-locative focus marker i-, in all examples that I have found, it is spelled as yuk- rather than ik-. All phonographic co- occurrences can be found in Table . For example:
() 於伎都之麻 / 伊由伎和多里弖 / 可豆久知布 / 安波妣多麻母我 / 都々美弖夜良牟 okî -tu sima i- yuk- -î -watar- -i -te kaNtuk- open_sea -. island - go- - -cross- - - dive- -u t- ip- -u apaNpî tama möNka tutum- -î -te yar- - say- - abalone pearl hide- - - send- -am- -u -- - “I will hide it and send it, the abalone pearl [you] want that I said I would go towards the island in the open sea and dive [for].” ( : )
All examples in the Kojiki and the Man’yōshū use 由 for yu, whi Baxter recon- structs as *yuw (Baxter , p. ). All examples in the Nihonshoki Kayō use 喩 for yu.
Baxter reconstructs this as *yuH (Baxter , p. ). However, except in the Nihonshoki Kayō, this distinction may not be important; the man’yōgana used in documents other than the Nihonshoki, while originally from Late Middle Chinese, went through a Mid- dle Korean intermediary whi did not have a distinction between /i/ and /ji/ (Miyake , pp. -; Vovin , p. ). is construction is not aested in the Fudoki Kayō or the Bussoku Seki Ka. I have also omied the numerous logographic examples from the Man’yōshū, whi tell us nothing about how this co-occurrence might have been pronounced.
. e Prefix uti- Similar to when yik- co-occurs with the prefix i-, when yik- co-occurs with the pre- fix uti-, it shows the same spelling bias towards yuk- rather than ik-. e only examples are from the Man’yōshū, and are found in Table . One— : —is wrien logo- graphically. e others are all spelled with 由 for yu.
() 波萬へ余里 / 和我宇知由可波 / 宇美邊欲理 / 牟可倍母許奴可 / 安 麻能都里夫祢 pama PÊ -yöri wa -Nka uti- yuk- -aNpa umî PÊ -yôri mukapë bea side - I - - go- - sea side - meeting mö kö- -n- -u ka ama -nö tur- -i -N- come- -- - fisherperson - angle- - -- pune boat “If I went from the edge of the bea, would I also meet the fisherman’s boat whi came from the seaside?” ( : )
. e Bigram taNpî yuk- e phrase taNpî yuk-, meaning “to go on a journey”, also displays the same tendency towards yuk- spellings. It only occurs in the Man’yōshū, and is only spelled using 由 (yu). Once again, these are found in Table . For example:
() 久左麻久良 / 多妣由久 比等乎 / 伊波比之麻伊久 / 与布流末弖 / 伊 波比伎尓家牟 kusa makura taNpî yuk- -u -N- pîtö -wo ipapîsima grass pillow journey go- - -- person - Ipapîsima iku yö p- -uru -maNte ipap- -î k- -î -n- how_many generation pass- - - pray- come- - -- -i -k- -êm- -u - -.- -- - “For how many generations has Ipapîsima prayed for people going on a journey [using] grass [for their] pillow?” ( : )
. Meter Going ba to Example (), we can see that yuk- occurs in a line with normal meter, while ik- occurs in a hypermetric line:
() 和我勢故波 / 多麻尓母我毛奈 / 手尓麻伎鵜 / 見都追由可牟乎 / 於 吉鵜伊加婆 乎思 wa -Nka se -kô pa tama n- -i möNkamô na TE -ni I - beloved - jewel - - arm - mak- -î -te MÎ -tutu yuk- -am- -u -wo ok- wrap- - - look()- - go- -- - - leave- -î -te ik- -aNpa wosi - - go- - be.regreable “[I] want my beloved to be a jewel! Although [I] would go wrapping [him] around my arm and looking at [him], [it] is regreable if [I] go leaving [him].” ( : ) (Vovin , p. )
e last line of the poem is hypermetric, having eight syllables rather than seven. In this environment, though, we would expect any adjacent vowels to contract (Vovin , pp. -). erefore, if the reading of 伊加婆 really is ikaNpa, we would then expect a contraction of okîte ikaNpa to okîtikaNpa, whi would resolve the hypermetric line. If, however, this verb is yik-, then we are le with a slightly unsatifying answer: okîte yikaNpa should not contract, and we would end up with a truly hypermetric line. Compare this mu more likely contraction:
() 草枕 / 多妣伊尓之 伎美我 / 可敝里許牟 / 月日乎之良牟 / 須邊能思 良難久 KUSA MAKURA taNpî (i)n- -i -si kîmî -Nka kapêr- grass pillow journey go_away- - -. lord - return- -i -kö- -m- -u TUKÏ PÎ -wo sir- -am- -u suNpe - -come- -- - month day - know- -- - way -nö sir- -an- -aku - know- -- - “e fact is that I don’t know the way that I will know the month [and the] day when my lord who went away on a journey [using] grass [for his] pillow will come ba.” ( : )
e Nihonshoki Kayō
As mentioned earlier, the Nihonshoki Kayō would provide the best evidence internal to Old Japanese about a possible yi, as its orthography did not go through an interme- diary from Middle Chinese (Miyake , pp. -; Vovin , p. ).
As stated above, all examples of the verb yuk-/ik- ( , , , , and ) are wrien with 喩 , whi Baxter reconstructs as *yuH (Baxter , p. ). ere are no examples of the word yumë/imë in the Nihonshoki Kayō. ere is one example of the noun yu/i (“sacred”) is aested once in the Nihonshoki Kayō ( ) and is spelled as i using the man’yōgana 伊 , whi Baxter reconstructs as *ʔjij (Baxter , p. ). e placename Yukî/Ikî appears once ( ), as already noted in Vovin (, pp. -), and is wrien as using the man’yōgana 以 , whi Baxter reconstructs as *yiX (Baxter , p. ). 以 (i) is also important here because it is a relatively uncommon aracter in the Nihonshoki Kayō. While these results are conflicting, they do tend towards using aracters that are reconstructed to have had *y as an onset rather than *ʔ. e relatively small size of the Nihonshoki Kayō unfortunately limits any stronger conclusions.
Conclusions
“Real” support for a syllable yi would come from triangulating evidence from Middle Chinese or from Ainu, or perhaps a “missing link” in terms of Japonic internal recon- struction. is kind of support is not forthcoming. While nothing in this paper cries out for an explanation, I have tried to show here that there are some oddities that fit fairly well into the hypothesis that there was indeed a syllable yi in Old Japanese, as well as one possible counter-example.
References
Baxter, William H. (). A Handbook of Old Chinese Phonology. Trends in Linguistics: Studies and Monographs . New York: Mouton de Gruyter. — (). An Etymological Dictionary of Common Chinese Characters. Dra. : htt p://www-personal.umich.edu/~wbaxter/pdf/d001-020.pdf. Miyake, Marc Hideo (). “e Phonology of Eighth Century Japanese Revisited. An- other Reconstruction Based Upon Wrien Records”. Ph.D. thesis. University of Hawai‘i at Mānoa. Takagi, Iinosuke, Tomohide Gomi, and Susumu Ōno, eds. (). Man’yōshū . Japanese. th ed. Nihon Koten Bungaku Taikei [Japanese Classical Literature Compendium] . Tōkyō: Iwanami Shoten. — eds. (). Man’yōshū . Japanese. th ed. Nihon Koten Bungaku Taikei [Japanese Classical Literature Compendium] . Tōkyō: Iwanami Shoten. — eds. (). Man’yōshū . Japanese. th ed. Nihon Koten Bungaku Taikei [Japanese Classical Literature Compendium] . Tōkyō: Iwanami Shoten. — eds. (). Man’yōshū . Japanese. rd ed. Nihon Koten Bungaku Taikei [Japanese Classical Literature Compendium] . Tōkyō: Iwanami Shoten. Tsuihashi, Yutaka and Jin’ii Konishi, eds. (). Kodai Kayōshū [Compliation of Ancient Songs]. Japanese. th ed. Nihon Koten Bungaku Taikei [Japanese Classical Literature Compendium] . Tōkyō: Iwanami Shoten.
Vovin, Alexander (). A Descriptive and Comparative Grammar of Western Old Japanese. Volume One: Sources, Script and Phonology, Lexicon, Nominals. Folkestone: Global Oriental. — (). A Descriptive and Comparative Grammar of Western Old Japanese. Part Two: Adjectives, Verbs, Adverbs, Conjunctions, Particles, Postpositions. Folkestone: Global Oriental.