Александр Владимирович Вовин - On the Etymology of the Name of Mt. Fuji

modern Kyoto accent is HL, which can reflect either 2.3 (LL) or 2.2 (HL). Since the only overlap of accent patterns is between Kyoto and Kagoshima for 2.2, one might think that we should reconstruct PJ *punsi as 2.2 (HL). But such a mechanical reconstruction ignores both the history of the language and geographic realities. First we must explore the historical records of the accent for OJ puⁿzi > MJ fuⁿzi > MdJ fuzi. Table 9.2 above traces this accentual history.

TABLE 9.2 Accentual history of Fuji^[6]

Segmental form Accent Textual source of accentual marks Date fuⁿzi LL various KKWKS manuscripts late 11th c.–1333 AD fuⁿzi LL SCS 1185–1190 AD fuⁿzi LL Jōben-bon of SIWKS 1185–1333 AD fuzi HL CMJF end of 17th c. fuzi HL HKMB 1776 AD fuzi HL GK mid-Edo period fuzi HL WTS first half of the 18th c. fuzi HL modern Kyoto 20th c.
One can see from this table that all early attestations in late Heian and Kamakura periods have LL accent, while all late attestations starting from Edo period have HL accent. This is to be expected, because it is well known that the merger between 2.2 (HL) and 2.3 (LL) accent classes as a single 2.2 (HL) class did not happen in the Kyoto dialect before the fifteenth century. Therefore, the HL accent in modern Kyoto cannot reflect the original Kyoto 2.2 (HL) class, but must be a reflex of 2.3 (LL) before it merged with 2.2 (HL). As far as we rely on the historical data and the modern Kyoto dialect alone, Martin’s reconstruction of pJN accent as 2.3 is justified. But how do we explain the discrepancies with accentuation in Tokyo and Kagoshima?

The difference with Kagoshima might seem quite bothersome, because Kagoshima HL clearly points not to a LOW, but to a HIGH register. This is where the geographical and historical realities should be taken into serious consideration. Mt. Fuji is quite close to Tokyo, somewhat far from Kyoto, but very far from Kagoshima. How many people from the uneducated classes in Kagoshima had ever seen Mt. Fuji let alone even known of its existence prior to the twentieth century? In sum, this place name in Kagoshima is unlikely to be native, going back to pJN. On the contrary, it is more likely to be a loanword from the Kyoto dialect after the accentual shift LL > HL had occurred in the latter. Thus, the Kagoshima accent can be excluded from further consideration.

As far as Tokyo is concerned, we still have the same LOW register, because 2.3 (LL), 2.4 (LH), and 2.5 (LF) classes are all LOW-initial. For the purposes of this etymology, the LOW register is all that really matters as the reader will see below shortly. However, once the etymology has been established, I will return again to the problem of reconstructing the pJN accent for *punsi.

After all these preliminaries, I am finally ready to offer my own etymology for Mt. Fuji. As I have briefly mentioned above, this etymology was already obscure in the Heian period. It is quite possible that the same was already true in the Nara period: the complete absence of logographic spellings seems to offer at least partial circumstantial evidence for this point of view. Nevertheless, I am going to demonstrate below that this place name has a perfect Japanese etymology, and that the reason that it became obscure so early is that the language that underlies is not the WOJ of Yamatö, but EOJ of Azuma.

Let me start from the second syllable ⁿzi of puⁿzi. I believe that it can be explained as ‑ⁿzi, a contraction of OJ nusi ‘master, owner’, which also occurs in OJ muraⁿzi ‘kabane title’ < *mura-nusi ‘village-master’, OJ tôⁿzi ‘mistress of the house’ < *tô-nusi ‘gate mistress’, OJ arôⁿzi (MJ aruⁿzi) ‘master’ < *ar-ô-nusi ‘exist-ATTR-master’. There is also MJ miyaⁿzi ‘majordomo’ (not attested phonographically in OJ texts) < *miya-nusi ‘palace-master’.

Then what is the first syllable pu‑? I believe it is EOJ pu ‘fire’^[7] that corresponds to WOJ pï (pô‑ in compounds). EOJ pu ‘fire’ is a hapax legomenon attested only in MYS 20.4419, a sakîmôri poem:^[8]

(1) 伊波呂尔波 (2) 安之布多氣騰母 (3) 須美与氣乎 (4) 都久之尔伊多里弖 (5) 古布志氣毛波母

(1) ipa-rö-ni pa (2) asi pu tak-ë-ⁿdömö (3) sum-î yö-kë-wo (4) Tukusi-ni itar-i-te (5) kôpusi-kë-mô pa mö

(1) house-DIM-LOC top (2) reed fire burn-EV-CONC (3) live-NML good-ATTR-ACC (4) Tukusi-LOC reach-INF-SUB (5) be.longing-ATTR-EXCL TOP PT

(2) Although [we] make a fire out of reeds (1) at [my] house, (3) the living [there] is good, so (5) [I] will be longing for it (4) when I reach Tukusi!

Note that pJ *poy ‘fire’ belongs to 1.3 (L) accentual class. The well-known rule of thumb for accentuation of Japanese nominal compounds is that the first element defines HIGH or LOW register, while the second element defines locus, if there is any. The LOW initial register of puⁿzi and its actual accentual class further confirm the possibility that the first syllable can be explained as EOJ pu ‘fire’. Thus, EOJ puⁿzi < pJN *po-nusi ‘fire master,’ is quite a fitting name for an active volcano. For the usage of nusi ‘master’ in volcano deities names, cf. the OJ name of the deity Opo-ana-nusi ‘big-hole-master’, which, as J. Marshall Unger once suggested (p. c.), could be the name of a volcano deity.

In conclusion, two issues need to be addressed. First, if my solution of this etymology is correct, then EOJ pu ‘fire’ is no longer a hapax Legomenon, and we have the second independent evidence that the EOJ word for ‘fire’ was indeed pu. It is almost customary now to reconstruct the PJ form of WOJ pï as *pəy (Martin 1987: 405) rather than *poy. However, I know of no examples where pJ *ə raises to EOJ u, because only pJ *o > EOJ u. Recently, both Pellard (on the basis of the Ryūkyūan evidence) (2011: 10), and myself on the basis of the philological evidence in WOJ (pace Mabuchi 1972: 88) and EOJ phonological evidence just mentioned above (Vovin 2011: 222) argued for the reconstruction of PJ *poy rather than *pəy. Another attestation of EOJ pu ‘fire’ gives further credence to the EOJ side of argument in favor of *poy.^[9]

Second, the reconstruction of the pJN accent of *punsi as 2.3 (LL) may not really be as certain as Martin has presented it (1987: 420). In addition to the Tokyo accent pointing to 2.4 (LH) or 2.5 (LF), there is another problem which concerns the accentuation of ‑ⁿzi < nusi ‘master, owner’ in the compounds listed above. The MJ accent of nusi ‘master, owner’ itself is 2.4 (LH). Consequently, we would expect -ⁿzi to have a HIGH pitch. And indeed, this high pitch is regularly reflected in MJ muraⁿzi HHH ‘kabane title’ < *mura‑nusi ‘village master’ and MJ aruⁿzi XXH ‘master’.^[10] The possible explanation accounting for the minor accentual discrepancy between MJ fuⁿzi LL vs. the expected *fuⁿzi LH is that WOJ puⁿzi and MJ fuⁿzi must be loanwords from EOJ puⁿzi due to their EOJ peculiar phonetic shape with ‑u‑ rather than with ‑ô‑: we would expect WOJ *pôⁿzi and MJ *foⁿzi, if they were really cognates. Under this scenario, a minor accentual discrepancy that affects locus but not register is not totally unexpected, as such discrepancies occur even within the same language, cf. WOJ isô (MJ accent HH) ‘rock, rocky shore’ and WOJ isi (MJ accent HL) ‘rock, stone’.

Abbreviations

Grammatical Terms

ACC Accusative

ATTR Attributive

CONC Concessive

DIM Diminutive

EV Evidential

EXCL Exclamative

INF Infinitive

LOC --">