[all.one’s.power ACC expend.one’s.power do] ‘totally expend all one’s power’ and
*汚れた顔を洗顔する *yogoreta kao o sengan suru [dirty face ACC wash-face do]
‘wash one’s dirty face’are impossible (see also Shimamura 1985).
In the two-character Sino-Japanese words of (13b), which are composed of two verb-like elements,救助 kyuu-zyo‘save’ contains two near-synonyms, kyuu ‘save’ andzyo‘save’, and開閉kai-hei‘open and close’contains two antonyms,kai‘open’ andhei‘close’. In neitherkyuu-zyonorkai-hei is it possible to say that one or the other of the verb-like elements determines the meaning of the word as a whole, so, from the perspective of identifying the semantic head, they can probably be analyzed as being dual-head two-character S-J words. In contrast to these, since 殴殺oo-satu[beat-kill] means‘kill by beating’, it is the right-hand verb-like element (satu‘kill’) that can be identified as the semantic head of the word.
Finally, in the examples given in (13c) of two-character Sino-Japanese VNs composed of an adjunct-like element and a verb-like element, since病死 byoo-si [illness-die] means‘to die of an illness’, the left-hand adjunct-like elementbyoocan be taken as an adverb-like element giving the cause of death. 銃殺 zyuu-satu [gun-kill] means ‘kill with a gun (kill by shooting with a gun)’ and the left-hand adjunct-like element zyuu ‘gun’ shows the instrument used in the act of killing.
密売 mitu-bai[secretly-sell] means‘sell secretly/illegally’and the left-hand adjunct- like elementmitushows the manner in which the act of selling takes place. All of these two-character S-J words have their semantic heads on the right.
4 Characteristics of three-character Sino-Japanese words
This section examines the structure and meaning of three-character S-J words.
However, as in the case of two-character S-J words, apparent three-character words resulting from clipping of longer expressions, such as 女子大 zyosidai (←女子 大学 zyosi-daigaku ‘women’s college’) and 知事選 tizisen (←知事選挙 tizi-senkyo
‘governor-election’) are excluded from consideration.
Table 5:Classification of three morpheme S-J words
A. Free one-character S-J morpheme
B. Bound one-character S-J morpheme I. One-character S-J
morpheme on right
中国 茶tyuugokutya [China-tea],
修正 案syuuseian [revision-proposal]
問題 視mondaisi [problem-view], 永住 権eizyuuken [permanent.residency-right]
II. One-character S-J morpheme on left
腸 閉塞tyooheisoku [intestine-blockage], 市 役所siyakusyo [city-offices]
最 先端saisentan [most-leading.edge], 重 労働zyuuroodoo [heavy-labor]
Considering the four types given in Table 5 from a semantic perspective, all have their semantic heads on the right. However, they differ in terms of their morphol- ogical structure. For example, the word中国-茶 tyuugoku-tya‘Chinese tea’in IA is composed oftyuugoku‘China’and tya‘tea’, and both are free morphemes that can appear as either the left or the right member of a compound (茶-道具 tya-doogu
‘tea utensils’ is an example of a compound with tya on the left and 古代-中国 kodai-tyuugoku‘ancient China’is an example withtyuugokuon the right). Likewise, the words in IIA are composed of free morphemes, so they are probably best ana- lyzed as compounds. On the other hand, the example問題-視 mondai-si‘regard as a problem’in IB can be separated intomondai‘problem’andsi‘view, see as’, butsi is a bound morpheme and moreover, in the meaning of‘view as’can only appear as the right member of a compound, as in特別-視 tokubetu-si[special-view.as]‘regard as special’or 例外-視 reigai-si[exception-view.as] ‘regard as an exception’. Thus, this視 siin the meaning of‘view, regard as’can be analyzed as a suffix.2The word 最-先端 sai-sentan in IIB is composed of 最 sai ‘most’and 先端 sentan‘leading edge’and, sincesai is a bound morpheme that can only appear at the beginning of a word, it can be analyzed as a prefix. Overall, the words in IA and IIA can be analyzed as compounds, and those in IB and IIB as affixed words. Since a single S- J morpheme may have several different uses, however, the distinction between com- pounding and derivation is often not very clear.
One type of compound in which multiple morphemes co-occur with equal status is coordinate compounds. The examples in (16) are all composed of three Sino- Japanese morphemes that are placed in a coordination relation, but this pattern is not limited to three-character S-J words; there are also cases of two-character S-J words like男女 dan-zyo[man-woman] as seen in Section 3.3 and four-character
2The視siappearing in視界si kai[see limit]‘visibility’or視野 si ya[seefield]‘field of vision’, although written with the samekanji, has the meaning‘see, perceive visually’and is not the same as thesiappearing inmondai si.
S-J words like都道府県 to-doo-hu-ken‘the 47 prefectures (fromtookyoo-to, hokkaidoo, oosaka-hu, kyooto-hu, and 43-ken)].
(16) 松竹梅 syoo-tiku-bai[pine-bamboo-plum]‘high, middle, and low rankings’ 雪月花 setu-getu-ka[snow-moon-flowers]‘the beauty of the four seasons’ 市町村 si-tyoo-son[city-town-village]‘municipalities’
Kageyama (1982) points out a phenomenon notable in Japanese that cannot be properly accounted for within the traditional morphological framework of compound- ing, derivation, and blending. Examples are given in (17).
(17) a. 与野党 yo-ya-too‘(political) parties in and out of power’(cf.与党 yo-too
‘party in power’,野党 ya-too‘party out of power’)
内外科 nai-ge-ka‘internal medicine and surgery’(cf.内科nai-ka‘internal medicine’,外科 ge-ka‘surgery’)
柔剣道 zyuu-ken-doo‘judo and kendo’(cf.柔道 zyuu-doo‘judo’, 剣道 ken-doo‘kendo’)
入退院 nyuu-tai-in‘entering and leaving hospital’(cf.入院 nyuu-in
‘enter the hospital’,退院 tai-in‘leave the hospital’)
出入国 syutu-nyuu-koku‘leaving and entering a country’(cf.出国 syuk-koku‘leave a country’,入国 nyuu-koku‘enter a country’).
b. 輸出入 yu-syutu-nyuu‘export and import’(cf.輸出 yu-syutu‘export’, 輸入 yu-nyuu‘import’)
国内外 koku-nai-gai‘domestic and foreign’(cf.国内 koku-nai [country-inside]‘domestic’,国外koku-gai[country-outside]‘foreign’) 祖父母 so-hu-bo‘grandparents’(cf.祖父so-hu‘grandfather’,祖母 so-bo
‘grandmother’)
In (17a),与野党 yo-ya-too, for example, is created fromyo-too‘party in power’and ya-too ‘party out of power’. Expressed in a more general format, the examples in (17a) are all created from two two-character S-J wordsX-Aand Y-Athat share a common right-hand elementA, giving a three-character S-J word of the formX-Y-A.
In (17b), on the other hand, 輸出入 yu-syutu-nyuu, for example, is created from yu-syutu‘export’ and yu-nyuu‘import’. The examples in (17b) are all created from two two-character S-J wordsX-AandX-Bthat share a common left-hand elementX, giving a three-character S-J word of the formX-A-B.
How are such examples to be analyzed theoretically? One apparent possibility is the idea of“blending”. As shown in Chapter 1 (Kageyama and Saito, this volume), in blending parts of two independent words are extracted and joined to form one word,
Sino-Japanese words 109
as ingozira‘Godzilla’fromgorira‘gorilla’andkuzira‘whale’. Just counting the num- ber of morphemes comprising the three-character Sino-Japanese words in (17), the idea of blending may appear promising, but from a semantic perspective this view cannot be adopted. This is because blending normally has the function of creating a new word different from the two original words. While Godzilla is a beast com- pletely different from either a gorilla or a whale, zyuu-ken-doo ‘judo and kendo’ does not express a new concept but rather simply expresses the conjunction of the original wordszyuu-doo‘judo’andken-doo‘kendo’. It is, therefore, not appropriate to analyzezyuu-ken-dooas a blending ofzyuu-dooandken-doo.
Another way of looking at the expressions in (17) would be to analyze them as a combination of a two-character S-J word and a one-character S-J word, that is treat zyuu-ken-doo‘judo and kendo’as [zyuu-ken] + dooornyuu-tai-inas [nyuu-tai] +in.
However, *zyuu-ken and *nyuu-taido not exist as two-character S-J words. Further- more, in the case of内外科 nai-ge-ka, the morpheme外 ge‘external’is pronounced not as the more commongaibut as a special allomorph gethat only appears in a very few words like外科 geka‘surgery’and外宮 ge-kuu‘the Outer Shrine of Ise’. 内外 nai-gai‘inner and outer’does exist as a two-character S-J word, but if内外科 were analyzed as 内外 with 科 ka ‘department’ attached, the wrong reading of
*nai-gai-kawould be predicted.
Thus, the examples of (17a) are not to be explained with garden-variety word formation processes like simple compounding or blending. There is very little previ- ous research regarding constructions such as these, but Kageyama (1982) analyzes them as coordinate structures incorporating a word-internal empty category. For instance, he argued, by analyzing nai-ge-ka as [nai-Δi] [ge-kai] where the empty categoryΔand theka ofge-kaare coreferential, the correct interpretation of‘naika [internal medicine] andgeka[surgery]’can be obtained. (Note that this coordinate structure incorporating an empty category is the same as the structure proposed for a coordinate structure with predicate deletion, as in Ani wa ringo oΔi, otooto wa banana o tabetai ‘The older brother Δi an apple and the younger brother atei a banana’.) A similar structure incorporating an empty category can also be con- sidered for long compounds like those in (18). The symbol・ (middle dot, centered dot) is a common punctuation mark in Japanese and shows word-internal conjunction (like&). TheΔis part of the analysis and is not realized phonetically or graphically.
(18) a. 格安Δ・激安航空券 kakuyasu-Δ・gekiyasu-kookuuken(=格安航空券 kakuyasu-kookuuken[cheap-airline.ticket]& 激安航空券 gekiyasu- kookuuken[dirt-cheap-airline.ticket])
b. 第2Δ・第3章 dai-ni-Δ・dai-san-syoo(=第2章 dai-ni-syoo[chapter 2]
&第3章 dai-san-syoo[chapter 3])
A problem concerning the analysis of Kageyama (1982) is whether or not hypothe- sizing a word-internal empty category is theoretically permissible or not. Hypothe- sizing deletion instead of an empty category gives the same results, but runs into a similar problem of whether or not word-internal deletion is theoretically appro- priate. We can conclude, then, that the word formation mechanism observed in (17a) is still unresolved theoretically. In contrast, since the right-hand elements of the examples of (17b) all exist as two-character coordinate compounds, they can be given a simple analysis like (19) without any need to posit empty categories or deletion.
(19) yu-[syutu-nyuu](transport-[out-in]),koku-[nai-gai](country-[in-out]), so-[hu-bo](grand-[father-mother]),gi-[hu-bo](in.law-[father-mother])