CC BY-NC-ND
11
Martine Robbeets University of Leuven
Proto-Japanese: Issues and Prospects.
(Amsterdam Studies in the Theory and History of Linguistic Science. Series IV — Current Issues in Linguistic Theory 249.) Bjarke Frellesvig and John Whitman (eds.).
Amsterdam / Philadelphia: John Benjamins Publishing Company, 2008. VII + 229 pp.
In a recent publication titled "Proto-Japanese: Issues and Prospects", Frellesvig and Whitman revise the reconstruction of proto-Japanese, which they define in the opening line as (1:) "the reconstructed language from which all later varieties of Japanese descend." Reconstructions proposed by various specialists in the field are organized in four subject areas: I phoneme inventory, II use of dialects, III accent and IV morpho-syntax. The methods that are employed are internal reconstruction, especially applied to Old Japanese and dialect comparison, including the varieties of the Ryukyus. Incorporating some references to Korean parallels in passing, external comparison is limited to an absolute minimum. The primary goal of this book is described as (2): "to present new research which advances our understanding of pJ".
In the present review article I will argue that although this volume gives a clear overview of recent research and presents a state-of-the-art analysis of certain important issues in the field, the primary goal could have been reached more successfully. This is due to the fact that the editors have chosen to leave out a valuable source of information that may accelerate our understanding of proto-Japanese, namely the historical comparison of Japanese with the other Altaic languages. In what follows I intend to give some indications of how external comparison can shed light on the reconstruction of proto-Japanese, by creating new insights and by supporting the insights developed in this volume. But the significance of the comparative method in this context also goes the other way around: the reconstruction of proto-Altaic becomes less problematic as our understanding of proto-Japanese advances. When we intend to crack the genetic code of Japanese, we should not only examine its offspring, but also its ancestors.
1. Proto-Japanese issues
The four subject areas into which the book is organized may seem somewhat forced: Whereas seg-
mental and suprasegmental phonology, morphology and syntax refer to the various areas of reconstruction — sound, form and structure, — dialect comparison is one of the methods employed in the reconstruction. Since proto-Japanese is defined as the ancestral language of all Ryukyuan and mainland Japanese varieties, any contribution ideally takes dialect comparison into account.
The reader will notice a relative imbalance when it comes to the various areas of reconstruction: six chapters deal mainly with phonology, three with morphology and one with syntax. Chapter four on the uses of Ryukyuan deals mainly with phonology, but peripherally reviews two issues that bear on the reconstruction of pJ morpho-syntax. Given the emphasis on sound and accent reconstruction, it will come as no surprise that the cutting-edge work is found in this field. Most notable in this respect is chapter one, on the seven vowel system by Frellesvig and Whitman and six, on the accent system for disyllabic nouns by Shimabukuro.
Basing their arguments on their joint research over the past ten years, the results of which were preliminarily published in 2004, Frellesvig and Whitman challenge the widely accepted hypothesis that pJ consisted of only four vowels (*i, *a, *u, *a). They strengthen the case for adding two mid vowels (*e, *o), a reconstruction that is supported by the Ryukyuan evidence reviewed by Serafim in chapter four. They further argue for the reconstruction of an additional high central vowel (*i). Their hypothesis deals only with short vowels, leaving out pJ long vowels, which are commonly thought to be reflected in low pitch in EMJ.
Building on his doctoral dissertation (2002) on suprasegmental reconstruction, Shimabukoro then nuances this widespread idea that low register in EMJ directly corresponds to vowel length in pJ. He argues that the low register forms reflect original long or short vowels in the initial syllable, whereas the high register forms reflect only short vowels. He finds that the original distinction in vowel length is reflected in
Ryukuan, namely in the Shuri and Nakijin accentual subclasses 2.3-5a and 2.3-5b. He convincingly argues against Matsumori's analysis of the pR accent system presented in chapter 5.
In agreement with Shimabukuro, Vovin in chapter 7 supports the correlation between the low registered accentual subclasses 2.3-5 a and 2.3-5b and original vowel length. He proposes that the remaining distinction between low and high register goes back to an original voice distinction at the so-called "pre-proto-Japanese" level. Puzzling is not only Vovin's reference to an undefined stage of pre-pJ (144, 153), presumably in reference to Japanese before it arrived in Japan. The reader may be more perplexed by the question, why the ancestral register distinction is traced back to voice and not some other feature. The answer is simple: external comparison with Altaic (Vovin 1995). However, since Vovin (2005) has recently closed the Altaic debate with a negative answer, voice distinction here appears as a deus ex machina. Although not mentioned by the author, the correlation between pJ pitch and Altaic voicedistinction has been proposed earlier by Kortlandt (1993).
Vovin's reconstruction of voice distinction in pre-pJ is incompatible with Unger's hypothesis in chapter two that pJ had voiced obstruents /*b, *d, *g, *z/ which lenited and are reflected in OJ as /w, y, 0, 0-s/. Challenged by the Chinesecharacter data in Miyake (1999), Unger revises the classical lenition theory, proposed by himself and Ramsey (1972) and adopted by Martin (1987). The lenition of pJ *g > 0 plays an important role in this idea.
The three chapters on morphology all study the form, function and occurrence of verbal morphemes. In chapter three Onishi explains patterns of dialect distribution between eastern and western Japan. What is new is his calculation method for the time at which the radiation of certain morphological elements in western Japanese reached the natural boundary of the North and Central Japan Alps. But, unfortunately, it does not work. The calculation is based upon the assumption that the radial diffusion of linguistic items occurs at a constant speed and that Kinai is — and has always been — the center for radiation. Both assumptions are unlikely and explain the unrealistic outcome of the calculations. For the radiation of negative suffixes, the method calculates that the form reached the boundary of the Alps in 480 BC which implies, carrying Onishi's formula further, that it started to spread from the Kinai center in 761 BC. These dates cannot be reconciled with the chronology proposed at several instances in the volume, for instance (98): "It is widely thought that Japanese languages entered Japan ...
with the carriers of Yayoi culture,.. (...300BC...)" and (156): "If the Japanese proto-language indeed arrived in Japan together with the Yayoi culture,.."
In chapter 8 and 9 Whitman and Frellesvig each give their viewpoint on the origin of the bigrade verb classes. Although Frellesvig proposes some changes to Whitman's hypothesis, both scholars are basically in agreement that the bigrade classes originate in the grammaticalization of an ancestral form of the verb e-'get, obtain, be able to'. This is a "new" idea because it breaks with the more common hypothesis presented by Unger (1977: 131) and adopted by Martin (1987: 672) that a formant with causative-passive connotation -(C)i- is involved, but it is built on an "old" proposal by Yoshida (1973:85-86). Whereas the bigrade of OJ puka- 'be deep', for instance, would be derived in the first way according to Unger and Martin, Whitman would derive it in the second way.
(1) *puka- 'be deep' + -(C)i (consonant deletion) > *pukay (contraction) > puke-'deepen'
(2) *puka- 'be deep'+ e- (vowel raising) > *pukay (contraction) > puke- 'deepen'
Whitman's account of the origin of bigrades is considered to be (178): "One of the most exciting recent advances in our understanding of earlier Japanese" and (179): ".superior to previous reconstructions", but it is not without problems. His analysis contains a formal as well as a functional contradiction. From a phonological perspective the above derivation is in conflict with his own chronology of sound changes proposed in chapter 1 (40), where the contraction of falling diphthongs is thought to precede mid vowel raising. The second obstacle is in the proposed pathway of grammaticalization. Whitman's cross-linguistic insight (165) that "Typological parallels become relevant here: while passives derived from inchoatives and causatives derived from passives are robustly attested across languages inchoatives derived from passives or causatives are not" is in conflict with the results of Haspelmath's (1990: 49) typological research: "And note that there is again unidirectionality: a causative can become a passive, but to my knowledge there is no evidence for a case of a passive becoming a causative" [emphasis added in both citations].
The volume concludes with the sole contribution that attempts to reconstruct aspects of pJ syntax. Building on his doctoral dissertation on complementation, Wrona compares the distribution and function of the OJ nominal form in -aku and the adnominal verb form. He argues that the OJ adnominal form originated as a marker of clausal modifiers of nominal heads, and developed as a marker of complementation in cases where the nominal head was dropped. If
reference was made to Serafim's contribution in chapter 4 where it is shown that kakari-musubi is re-constructable for pJ and not only for "pre-OJ", the conclusion (213): "...it is difficult to say how the ex-clamative [and kakari-musubi] usage of the Adnominal form developed diachronically" could have sounded somewhat less inconclusive.
2 Altaic prospects
The scholarly background of this book is set by two large scale reconstructions of proto-Japanese, one by Martin (1987) and one by Hattori (1978-1979). Martin's "The Japanese language through time" is a milestone in the reconstruction of proto-Japanese segmental and suprasegmental phonology and lexicon. Inflectional morphology and syntax are underrepre-sented vis-à-vis the other areas of reconstruction. Unfortunately the present volume does not succeed in restoring the balance for morpho-syntax. Whereas Martin mainly employs internal reconstruction, except for the accentual data, Hattori's "Nihon sogo ni tsuite" is concerned with Ryukyuan dialectal comparison. However, it can be remarked that these authorities of the reconstruction of proto-Japanese both took an active interest in the external comparison of Japanese with Korean and the Altaic languages during their careers.
Admitting that (123) "When actually working on the reconstruction, we need to work in a bottom-up fashion,.." and that (146) "external evidence should never take precedence in explaining internal data", to give priority to internal evidence is wise, but it is unwise to ban external evidence. External comparison can shed light on the pJ reconstructions. It will help us to see the driving forces behind the changes in a larger perspective. Ideally internal and external evidence should be used in tandem. When we want to unravel the genetic code of Japanese, we should not only examine its offspring, but also its ancestors.
The editors justly emphasize the importance of internal reconstruction and dialect comparison in the reconstruction of proto-Japanese. It is not justifiable, however, to reduce the usage of external evidence to an absolute minimum. Our own contribution to the Workshop on proto-Japanese in Copenhagen in 2003, on which this volume is based, for instance, was declined on the motivation that the methodology had to be limited to internal reconstruction and dialectal comparison only. Although we respect this decision, we also regret it.
In our opinion, it is inappropriate to picture any reliance on the historical comparative method as (146) "falling into a Moscow Nostratic trap". First of all, it is possible to apply the classical historical comparative method on Japanese without any reliance on the techniques of the Moscow school of comparative linguistics and still reach the conlusion that Japanese is an Altaic language (Robbeets 2005). Second, the use of the term "trap" — a trick that is intended to deceive someone — is rancorous. Although one may be sceptical of long-range comparison, it is unprofessional to accuse scholars that take a different approach in their scientific work of deliberately misleading the public. Such emotional attitudes to difference in opinion are counterproductive to progress in our field.
In the remainder of this review article I intend to give some indications of how the historical comparison of Japanese with the Altaic languages can strengthen the case for the majority of proposals made in this volume.
2.1 Seven vowel system
Under the assumption of a four vowel system, some words with OJ -e-, or -ye- with so-assumed diphtongal origins in pJ and some words with OJ -i-show an "irregular" correspondence to pA *-e-. Under the seven vowel system, the correspondences in table 1 can perhaps be explained by mid vowel raising (pJ *e > ye > i). Proto-vowels between rackets are reconstructed on the basis of external comparison only.
Table 1. Correspondence of OJ -e-, -ye- or -i- with pA *-e-
Japanese Korean Tungusic Mongolic Turkic
OJ negwi- 'appease, pray', pJ *nenki- MK neki-, nyeki- 'consider', pK *neki- Evk. neke- 'intend, demand', pTg *neke- WMo. neke- 'demand', pMo. neke-
OJ ki- ~ kyes- '(make) wear', pJ *ke- WMo. kedur- 'wear (clothing)', pMo *kedur- OTk. ked- 'wear (clothing)', pTk *ke:d-
OJ we 'bait', pJ *we Evk. be 'bait', pTg *be OTk mey 'bird-seed', pTk *bey
J ketu 'butt, rump', pJ *ketu Evk. gedimuk 'back of the head', pTg *gedimuk WMo. gede 'back of the neck', pMo *gede OTk. kedin 'behind', pTk *ke:(d)
OJ sir- 'know', pJ *s(e)ra- WMo. sere- 'notice, sense', pMo *sere- OTk. sez- 'perceive', pTk *se:n -
Whereas the majority (42) of words with OJ -u- correspond to MK -wu- and can be traced back to pA *-u-in Robbeets (2005a: 368-370), there are about twenty etymologies where words with non-final OJ -u- correspond to MK -wo- and reflect pA *-o-. There is also one
etymology with final OJ -wo- and one with non-final OJ -wo- reflecting pA *-o- (2005: 370). Correspondences such as those in table 2 can probably be explained by mid vowel raising (pJ *o > wo > u).
Table 2. Correspondence of OJ -u- or -wo- with pA *-o-
Japanese Korean Tungusic Mongolic Turkic
OJ kura 'valley', pJ *k(o)ra MK "kwol(oy) 'valley, deep hole', pK *koln MMo. qol (SH), yol 'river (valley), centre', pMo *gol OTk. qol 'valley', pTk *ko:l
OJ nure- 'get wet', pJ *n(o)ra- WMo. nor- 'soak, be wet', pMo *nor-
OJ yuru 'loose, lax, slow', pJ *y(o)ru WMo. doru 'weak, feeble', pMo *doru Tk. yor- 'tire, tired', pTk *yor-
OJ suk- 'dig up earth', pJ *s(o)ku- Evk. soko- 'scoop', pTg *soko-
OJ ywowa- 'weak', pJ *y(o)wa- Ma. jobo- 'suffer, be poor', pTg *jobo- WMo. joba- 'suffer', pMo *joba-
It can further be noted that the seven vowel system opens new perspectives on Arisaka's law and vowel harmony in Jap., as mentioned by the authors of chap. 1.
2.2 Double origin of OJ /g/
Whereas it is generally agreed that the voiced obstruent OJ /g/ reflects a contraction of a nasal with a
following voiceless velar obstruent (< pJ *nk), Unger argues that it additionally reflects a velar nasal pJ In support of this hypothesis we can refer to a merger of cognates that reflect a velar cluster (table 3) and cognates that reflect a velar nasal (table 4) in the Altaic etymologies for words with OJ -g- (Robbeets 2005a: 349-350, Robbeets 2005b).
Table 3. Correspondence of OJ -g- with pA *-Ck-
Japanese Korean Tungusic Mongolic Turkic
OJ pagi 'lower leg, shin', pJ *panki MK pal 'foot' MK polh'arm', pK *palh 'limb' Evk. xalgan 'foot', pTg *palgan MTk. balaq 'trouser leg', pTk *balak
OJ pag- 'strip off', pJ *panka- Evk. hegde-li:- 'tear off', pTg *pegde-
OJ tagir- 'flow rapidly', pJ *tanki- Ev. jalki- 'be agitated (sea)', pTg *jalki- WMo. dargil 'rapid current', pMo *dargil MTk. talya 'sea undulation', pTk *talga
Table 4. Correspondence of OJ -g- with pA *-ij-
Japanese Korean Tungusic Mongolic Turkic
Mod. J mugo- 'cruel', pJ *munka- MMo. murj (SH) 'distress', pMo *mur OTk. buy], mu] 'suffering', pTk *bu]
OJ toga 'blame', pJ *tanka WMo. dorgud- 'blame', pMo *dorgud- OTk. yo] 'accusation', pTk yoy
J toguro 'coil', pJ *t3nkur3 MK twong-kul- 'be round', pK *tori Evk. toqollo 'rotund', pTg *toqol
2.3 Loss of final *-m in accent class 2.5
Vovin links the origins of accent class 2.5 of disyllabic nouns with a unique low-falling pitch, limited to the Kansai dialects, with the loss of the final consonant pre-pJ *-m. He relates the nominalizer pre-pJ *-m in verbal adjectives denoting colors (such as in OJ awo 2.5 'blue/green' < *awo- 'be blue/green' + *-m) with the Korean nominalizer MK -m (such as in MK chwu-m 'dance (n.)' < MK chwu- 'dance (v.)' + -m). Contrary to his view in 1994, Vovin now believes that (146): "The
limited attestation within Japanese in this case ... points more to a borrowing scenario from some variety of Old Korean to CJ than to common genetic inheritance." Against the limited attestation within Japanese we can remark that color terms in some Ryukyuan languages are nominalized with final nasals, e.g. Yonaguni aun 'blue' and that Ryukyuan preserves a deverbal nominalizer *-m, e.g. in Shodon Yubyum tyi 'I hear that he calls (or will call)' and Yudâm tyi 'I hear that he called' (Martin 1970: 128-131).
Against the borrowing scenario it can be noted that an (ad)nominalizer *-m(A) is attested not only in Korean, but across the Altaic languages. In the Tungusic languages we find evidence for a deverbal (ad)nominal-izer *-mA, frequently lexicalized in color terms such as in Evk. bagdama 'white' from bagda- 'become white' (Benzing 1955: 1038). Mongolian has the nominalizers *-m and -mA such as in MMo. qurim 'feast' from quri-'come together' or in Mgr. gurma 'plaited hair' from
guru- 'plait'. Turkic has the nominalizers *-(X)m and *-mA such as in OTk. yarma 'crack' and OTk. yarim 'half' from yar- 'split (open)' (Erdal 1991: 290-300; 316-320).
2.4 Voice distinction as a source for register
Vovin proposes a correlation between pre-pJ voice distinction and vowel length with pJ register and the MJ accent classes as summarized in table 5 below.
Table 5. Correlation voice, vowel length and register in Japanese
pre-pj voiced initial voiceless initial
short V (CVCV) pJ low register (2.3-5b) pJ high register (2.1-2.2)
long V (CV:CV) pJ low register (2.3-5a) pJ low register (2.3-5a)
Vovin adds an interesting remark (154): "All words belonging to subclasses 2.3a-2.5a are reconstructed with long vowels in the first syllable. Theoretically, they could have voiced onsets as well, but we simply have no evidence for that." This is indeed a good example of a case where internal reconstruction or dialectal comparison cannot provide evidence, but where external evidence holds the key to the solution. The
majority of cognates reflecting voiced initial stops in TEA correspond to Japanese words that belong to the accent classes 2.3, 2.4 and 2.5, thus supporting the correlation between Altaic voice and Japanese register. Whereas table 6 shows the correspondence for bimo-raic nouns, table 7 illustrates the correspondence for verbs and verbal adjectives with initial low pitch (B-type).
Table 6. Correspondence of Japanese accent classes with pA voice distinction for bimoraic nouns
Accent Japanese Korean Tungusic Mongolic Turkic
2.3 (B) OJ padi 'shame', pJ *panti Khal. balci- 'shame (tr.)', pMo *balci-
2.3 (B) OJ taka 'height', pJ *taka Evk. dag- 'cross over', pTg *daga-
2.3 OJ tura 'face', pJ *tura Evk. durun 'pattern, design', pTg *duru-n WMo. düri 'form, outlook', pMo *düri
2.4 OJ kata 'shoulder', pJ *kata WMo. yar 'hand', pMo *gar OTk. qarï 'arm', pTk *kari
2.5 (B) OJ paya 'early, already', pJ *paya MK polo- 'early' MK ■spolo- 'fast', pK *pnln Even baj 'early', pTg *badi OTk baya 'recently', pTk *ba(d)a
2.5 OJ tara 'angelica tree', pJ *tara MK tal 'rush', pK *tal Na. dara: 'reed', pTg *dara
2.5 OJ toga 'blame', pJ *tsnka WMo. doqgud- 'blame', pMo *dorjgud- OTk. yoii 'accusation', pTk y°i
Table 7. Correspondence of Japanese accent classes with pA voice distinction for verbs and adjectives
Accent Japanese Korean Tungusic Mongolic Turkic
B OJ kupasi- 'be beautiful', pJ *k(o)pa- MK kwo po- 'be beautiful', pK *kopn- WMo. yuwa 'beautiful', pMo *goba
B OJ tomo-si 'scarce', pJ *t(i)m(i)- MK ■ tu ■ mul- 'be rare', pK *timili- WMo. dömü- 'be scarce', pMo *dömü-
B OJ pur- 'wave, shake', pJ *puru- Evk. bul- 'caress, stroke', pTg *bul- WMo. büli- 'stir', pMo *büli-
B OJ kir- 'cut', pJ *kira- Ma. giri- 'cut', pTg *giri- OTk. qïr- 'break, scrape', pTk *kir-
B OJ kom- 'be crowded', pJ *kama- Ma. gemu 'all', pTg *gemu
B OJ kop- 'beg', pJ *kop- Ev. goy-, gowjo- 'hunt', pTg *gob- OTk qov- 'hunt', pTk *kob-
2.5 Source of the bigrade verb classes
In addition to the phonological and semantic problems with Whitman and Frellesvig's new hypothesis discussed above, external comparison supports the traditional analysis that the origin of the bigrade verb classes goes back to a formant with causative-passive connotation -(C)i-. The Japanese formant has a parallel not only in the Korean, but across the Altaic languages (Robbeets 2007). Korean has a causative-passive suffix K -ki-, -hi-, -i-, MK - ki-, - Gi-, - hi-, - i- (< pK *-ki-) such as in MK "wolm- 'move (intr.)' -> MK wolm ■ ki- 'move (tr.)' and MK elk- 'tie (tr.)' -> MK el khi- 'get tied (intr.)' (Martin 1992: 221-225, 623). In Tungusic we find a causative pTg *-ki:- with reflexes Ma. -gi-, Evk. -ki:- ~ -gi:-, Even -k(i)- ~ -g(i)-, -yi-, -i-, Ud. -gi-, Na. -(g)i-(Benzing 1955: 1070) such as in Evenki ulap- 'get wet (intr.)' > ulapki:- 'make wet (tr.)' (Vasilevic 1940: 93, Nedjalkov 1997: 230). In Mongolic and Turkic we find verbs meaning 'do, make' with the same shape, such as MMo. ki- / WMo. ki- and OTk. kil- and many contemporary reflexes. Functionally this etymology reflects a universal pathways of grammaticalization, whereby a causative verb 'do, make' develops into a causative formant and finally ends up as a fientive and passive marker. Although we lack conclusive internal evidence for an initial velar in Japanese (pJ *-(k)i-), velar elision before a high front vowel is sporadically attested in other stages of Japanese (e.g. OJ taka-ki for J taka-i 'high').
2.6 Adnominal forms used as finite predicates
Wrona wonders (210): "These usages [exclamative/ kakari-musubi: Adnominal forms as finite predicates] are not directly relatable to any relative construction. The exclamative usage has been argued to be similar to noun-final sentences (Yamada 1990 [1913]), but there is little evidence of a diachronic connection". From a typological perspective, however, insubordination (the conventionalized main clause use of morphologically subordinate forms) is a frequently observed diachronic process (Evans 2008). From a factual perspective, a connection between adnominal forms and finite predicates can be observed in Ryukyan and in the Altaic languages. In Shodon, for instance, adnominal forms derived with -un (e.g. yuby-ún tyu 'the person who does / will call') can also enter sentence-final predication (Yuby-ún 'he calls, he will call') (Martin 1970: 128-131). The same observation goes for the other Altaic languages: the participle or adnominal verb form is primarily used in relative constructions, but secondarily it can take over the function of finite predication. This has been noted before by a.o. Ramstedt 1952: 85-86, Poppe 1955: 557,
Kormusin 1984, Gorelova 2002: 478. Insubordination is a shared structural property of the Altaic languages, which is not necessarily due to genealogical retention. The pathway of grammaticalization can be explained by the fact that participles are verbal adjectives which, like other adjectives in Altaic, readily assume nominal tasks, i.e. they can be heads of noun phrases (e.g. OJ taka- 'high' and OJ taka 2.3 'height'). In addition, there is a tendency to use noun-final sentences across the Altaic languages, for instance when the Korean nomi-nalizer -m is used as a sentence-final suffix in the documentary style of written Korean (e.g. Onul-un swuep-i eps-um [today-Top class-NOM not.exist-NML] 'No class today.' Martin 1992: 887 / 254).
3 Conclusion
The recent publication "Proto-Japanese: Issues and Prospects" by Frellesvig and Whitman is an attempt to reconstruct proto-Japanese, the common ancestor of all living and historically accessible varieties of Japanese. The main emphasis of the reconstructions is on segmental and suprasegmental phonology. Since most of the issues raised in this book have been published elsewhere, the book is more successful in giving an overview on recent scholarly thinking than in presenting new research.
Although we have three tools at our disposal to gain information about the ancestral language, namely internal reconstruction, dialect comparison and external comparison, the editors have chosen to restrict themselves to the first two. The incorporation of Ryukyuan evidence is a noteworthy feature of this volume. However, disregarding external evidence, in our opinion, the editors leave out a valuable source of feedback. The scholarly background of this book is framed by the monumental work on Japanese reconstruction by Martin and Hattori, scholars who — not surprisingly — took an active interest in the external comparison of Japanese with Korean and other Altaic languages. In the present review we have shown how comparison with these languages could shed additional light on the main issues in the book.
In spite of this methodological weakness, the book remains an important contribution to the field with cutting-edge work on the proto-Japanese vowel system and on the accent system for disyllabic nouns. As promised in the title, the volume raises many valuable issues, but, in our opinion, the compatibility of the findings with the Altaic evidence is the most important prospect. One can gain more by applying external comparison than one can lose by denying it.
References
Benzing 1955 — Benzing, Johannes. Die tungusischen Sprachen. Versuch einer vergleichenden Grammatik // Abhandlungen der geistes- und sozialwissenschaftlichen Klasse 11, 949-1099.
Dybo & Starostin 2008 — Dybo, Anna & Starostin, George. In defense of the comparative method, or the end of the Vovin controversy // Aspects of comparative linguistics 3, 119-258.
Erdal 1991 — Erdal, Marcel. Old Turkic word formation. A functional approach to the lexicon. Vol. 1-2. (Turcologica 7). Wiesbaden: Harrassowitz.
Frellesvig & Whitman 2004 — Frellesvig, Bjarke & Whitman, John 2004. The vowels of proto-Japanese // Japanese language and literature 38, 281-299.
Gorelova 2002 — Gorelova, Liliya. Manchu grammar. Leiden: Brill.
Haspelmath 1990 — Haspelmath, Martin. The grammaticization of passive morphology // Studies in Language 14.1, 25-71.
Hattori 1978-1979 — HATTORI, Shirö. Nihon sogo ni tsuite 1-22 // Gekkan gengo 7:1-7:3, 7:6-8:12.
Kormusin 1984 — Kormusin, Igor. Sistemy vremen glagola v altajskich jazykach. Moscow: Nauka.
Kortlandt 1993 — Kortlandt, Frederik. The Origin of the Japanese and Korean Accent Systems // Acta Linguistica Hafniensia 26.57-65.
Martin 1970 — Martin, Samuel. Shodon: a dialect of the northern Ryukyus // Journal of the American Oriental Society 90. 1. 97-139.
Martin 1987 — Martin, Samuel. The Japanese language through time. New Haven: Yale University Press.
Martin 1992 — Martin, Samuel. A Reference Grammar of Korean. Tokyo: Tuttle.
Miyake 1999 — Miyaké, Marc. The phonology of eighth century Japanese revisited: Another reconstruction based upon written records. University of Hawaii dissertation.
Nedjalkov 1997 — Nedjalkov, Igor. Evenki. Descriptive Grammar. London: Routledge.
Poppe 1955 — Poppe, Nicholas. Introduction to Mongolian Comparative Studies (Mémoires de la Société Finno-Ougrienne 110). Helsinki: Suomalais-Ugrilainen Seura.
Ramsey & Unger 1972 — Ramsey, Robert & Unger, Marshall. Evidence of a consonant shift in seventh-century Japanese // Papers in Japanese linguistics 1.2. 270-295.
Ramstedt 1952 — Ramstedt, Gustaf. Einführung in die altaische Sprachwissenschaft. II: Formenlehre. MSFO 104.2.
Robbeets 2005a — Robbeets, Martine. Is Japanese Related to Korean, Tungusic, Mongolic and Turkic? (Turcologica 64). Wiesbaden: Harrassowitz.
Robbeets 2005b — Robbeets, Martine. Does Doerfer's Zufall mean 'cognate'? The case of the initial velar correspondence in Altaic // Turkic Languages 8, 146-178.
Robbeets 2007 — Robbeets, Martine. The causative-passive in the Trans-Eurasian languages // Turkic Languages 11.2.235-278.
Shimabukoro 2002 — Shimabukoro, Moriyo. A reconstruction of the accentual history of the Japanese and Ryukyuan languages. University of Hawaii dissertation.
Starostin et al. 2003 — Starostin, Sergei, Dybo, Anna & Mudrak, Oleg. Etymological Dictionary of the Altaic Languages. Leiden: Brill.
Unger 1977 — Unger, Marshall. Studies in Early Japanese Morphophonemics. Yale University Dissertation. Indiana: Indiana University Linguistics Club.
Vasilevic 1940 — Vasilevic, Glafira. Ocerk grammatiki Evenkijskogo (tungusskogo) jazyka. Leningrad: Ucpedgiz.
Vovin 1994 — Vovin, Alexander. Genetic Affiliation of Japanese and Methodology of Linguistic Comparison // Journal de la Société Finno-Ougrienne 85.241-256.
Vovin 1995 — Vovin, Alexander. Origin of register in Japanese and the Altaic theory // Japanese/Korean Linguistics 6, 113-133.
Vovin 2005 — Vovin, Alexander. The end of the Altaic controversy. Central Asiatic Journal 49.1, 71-132.
Yamada 1990 — Yamada, Yoshio 1990 [1913]. Naracho bunposhi. Tokyo: Höbunkan.
