行政院國家科學委員會專題研究計劃成果報告
*************************************************************** * * * 古漢語“所“字結構的句法分析 * * * **************************************************************** 計劃類別：個別型計劃 整合型計劃 計劃編號：NSC 89-2411-H-003-029 執行期限：88 年 8 月 1 日至 90 年 1 月 31 日 計劃主持人：丁仁 國立台灣師範大學 計劃參與人員：鄭胥德，黃欣然 （碩士生研究助理）國立台灣師範大學 執行單位：國立台灣師範大學中華民國 90 年 4 月 3 日
行政院國家科學委員會專題研究計劃成果報告
古漢語“所“字結構的句法分析 計劃編號：NSC 89-2411-H-003-029 執行期限：88 年 8 月 1 日至 90 年 1 月 31 日 主持人：丁仁 國立台灣師範大學 計劃參與人員：鄭胥德，黃欣然 （碩士生研究助理）國立台灣師範大學 1.中文摘要 本計劃由衍生語法(generative grammar) 的角度，歸納、討論所字結構 的句法特徵，且與現代漢語“所“字的 句法分布（cf. Ting (1996, 1998)）作 比較，藉此對古漢語所字結構提出新的 分析。本研究指出，古漢語的“所“， 如同現代漢語的“所“，應分析為附著 於 I0 上的填充性代詞黏著語 (resumptive clitic)。而“所“字結構 在古漢語與現代漢語中不同的句法表 現，應歸因於漢語的語法演變；例如， 現代漢語的介系詞在古漢語為動詞。本 計劃同時也收集了古漢語所字結構的基 本語料，以作為句法分析的基礎。 關鍵詞：“所“字結構，古漢語語法， 填充性代詞，黏著語 AbstractThis project investigates the suo
construction in classical Chinese and attempts to provide a proper syntactic analysis of it. To achieve this purpose, the basic facts of the suo construction have
been collected as a basis of the analysis. The result of this project shows that the classical suo, like its modern Chinese
counterpart, is a resumptive clitic adjoined to the INFL head. Its contrasting behavior with respect to the modern Chinese suo
construction is argued to result from other properties of classical and modern Chinese, for example, the categorical status of coverbs.
Keywords: the suo construction, classical
Chinese syntax, resumptive, clitic 2. Introduction
The analysis of the suo construction
in classical Chinese has long been a controversial issue. According to X. Zhu (1996) (cf. F. Zhou (1961)), the bone oracle (jiaguwen) and the bronze
inscriptions (jin wen) do not have the word
suo but have you （攸）, which has the
same usage as suo. The occurrence of suo
is seen as late as in jing shu. It may originate at the time around the early Warring States period (chunqiu.) Shuo wen states its meaning as “sound of chopping wood”. Later suo is borrowed to
mean “location.”
(1) 獻於公所 （詩經、鄭風）
Then suo further developed into a different
usage. It may occur before verbs or verb phrases to form a nominal phrase, which most commonly refers to the patient argument of the event. This is generally called the suo construction.
There is no general agreement concerning a proper analysis of the suo
construction and the grammatical status of the particle suo. There are two main lines
of approaches: suo as a pronoun and suo as
a construction particle. The first approach is represented by Ma Jianzhong (1898) according to whom suo is a connective
pronoun (jie du dai ci), with the function of relative pronouns as in languages like English. Wang Li (1976) and Zhou Shumin (1993) also consider suo a
demonstrative pronoun or pronominal of some sort. The latter approach is pursued by Zhu Dexi (1983), Yie Baomin et al. (1992), Yang and He (1992), who all regard suo as a particle of some sort
preceding verbs or verb phrases to form nominal constructions, which in turn refer
to the object of the predicate. In this project, I argue that suo is an X0
element, heading the NP in the corresponding theta-position of the head noun of the relative clause. Due to a morphological
requirement, it undergoes subsequent raising to I0
. This analysis will be shown to provide a proper analysis of suo in
classical Chinese and account for its syntactic behavior which would be otherwise left mysterious in previous analyses.
3. Basic facts of the suo construction
Let’s now turn to the basic facts of the suo
construction. Suo precedes a predicate and
the entire sequence becomes a nominal. The nominal status of the suo construction,
as pointed out by Yi (1989, p. 196) and Qi (1992, p. 6) (among others), is evidenced by its distribution as a regular nominal phrase. (The examples are taken from Yi (1989)) As a grammatical subject: (2)其所厚者薄，而其所薄者厚，未之有 也﹒（禮記、大學）. As a nominal predicate: (3)此六子者，世之所高也﹒（莊子、盜拓）
As a grammatical object of a verb:
(4)行法志堅，好修正其所聞，以矯飾其
情性﹒（荀子、儒效）
As a grammatical object of a preposition:
(5)以其所受，悖其所辭，則能禁之矣﹒（荀
子、正名）
As a nominal modifier introduced by zhi:
(6)凡說者之務，在知所說知所矜，而滅
其所恥﹒（韓非子、說難）
While the nominal introduced by suo most
commonly denotes the patient argument of the event as noted earlier, it can also denotes location, reason, or manner of the event. Patient: (7)吏之所誅，王之所養。（韓非子、五蠹） (8)駕彼四牡，四牡騤騤。君子所依，小 人所騑。（詩經、小雅） Location: (9)其北陵，文王所避風雨也。（左傳、僖 公三十二年） (10)舟車所至，人力所通，天之所覆， 地之所載，日月所照，霜露所墜，凡有 血氣者莫不尊親。（中庸） (11)西方有木焉，名曰射干，莖長四寸， 生於高山上，而臨百仞之淵。木莖非能 長也，所立者然也。（荀子、勸學） Manner: (12)臣不任受怨，君亦不任受德，無怨 無德，不知所報。（左傳、成公三年） (13)他日，子夏、子張、子游以有若似 聖人，欲以所事孔子事之。（孟子、滕文 Reason: (14)所愛其母者，非愛其形也。（莊子、 德充符）
Furthermore, the suo construction may
also denote the object of a preposition. In this case, suo occurs before a preposition
and a verb (phrase). The preposition that most frequently occurs in this pattern is yi (以) (Yi (1989)). Other prepositions in this pattern include you (由), yu (於), wei (為), zi (自), yu (與), hu (乎) and cong (從). (15)聖人非所與熙也，寡人反取病焉。（晏 子春秋、內篇雜下） (16)吾知所以距子矣，吾不言。（表方 式、工具）（墨子、公輸） (17)故之人所以大過仁者，無它焉，善 推其所為而已矣。（表原因）（孟子、梁 惠王） (18)是亂之所由作也。（荀子、正論) (19)梁乃召故所知豪吏，諭以所為起大 事。（史記、項羽本紀） (20)澤被天下而不知其所自始。（呂氏春 秋) (21)則避淫游惰之民無所於食﹒（商君 書、墾令） (22)生有所乎萌，死有所乎歸﹒（莊子、 田子方） (23)是吾劍之所從墜。（呂氏春秋、察今） (24)見漁人，乃大驚，問所從來。（桃花 源記）
mentioning regarding the suo construction.
First, suo does not have to immediately
precede the verb. They can be intervened by negation markers, auxiliary verbs, manner adverbs or preposition phrases. (25)而世之奇偉、瑰怪、非常之觀，常在 於險遠，而人之所罕至焉，故非有志者 不能至也。（王安石、遊褒禪山記） (26)死亦吾所惡，所惡有甚於死者，固 患有所不避。（孟子、告子上） (27)如使置十葉焉，則中不中非臣所能 必矣。（郁離子、求道） (28)和氏璧，天下所共傳寶也。（史記、 廉頗藺相如列傳） (29)法者，天子所與天下公共也。（漢 書、張釋之傳）
Second, the logical subject of the event denoted by the verb (phrase) preceded by
suo can be optionally introduced by the
particle zhi. (30)仲子所居之室，伯夷之所築乎？（孟 子、滕文公下） (31)其北陵，文王所避風雨也。（左傳、 僖公三十二年） (32)舟車所至，人力所通，天之所覆， 地之所載，日月所照，霜露所墜，凡有 血氣者莫不尊親。（中庸）
In this connection, when the logical
subject is a third person pronoun, it is most commonly represented as qi (其), a
possessive form.
(33)奪其所憎而與其所愛﹒（戰國策﹒趙
策）
Third, the suo V(P) sequence can be
optionally followed by the particle zhe
(者). The sequence with or without suo has
the same reference. As Qi (1989, p. 35) notes, the sequence with suo is not
prevalent. In fact, in order to make the reference of the suo sequence explicit, a
noun can be added to follow the suo V
sequence introduced by the particle zhi
(之). This occurrence of the modified NP is rare in the pre-Qin (先秦) period and
has got more popular after the Han period. (Yi (1989)) (34)必欲爭天下，非信無所與計事者﹒（史 記，淮陰侯傳） (35)君大怒曰：“所求者生馬，安事死 馬，而捐五百金？“ (36)仲子所居之室，伯夷之所築乎？（孟 子、滕文公下） (37)未必其身泄之也，而語及所匿之 事﹒（韓非子、說難） (38)所操之術多導故也﹒（王安石、答司 馬諫議書） (39)衡不慕當世，所居之官，輒積年不 徙﹒（後漢書、張衡傳） (40)遠誠畏死，何苦守尺寸之地，食其 所愛之肉，以與賊抗而不降乎？
In this connection, it is necessary to consider the usage of the particle zhe in
classical Chinese. One of the functions of
zhe is, like suo, being added to a predicate
to form a nominal. But its position is following the predicate and the reference of the sequence is the agent of the
predicate.
(41) 新浴者振其衣，新沐者彈其冠﹒（荀
子，不苟）(= Zhu’s (3.19))
(42) 不救火者比降北之罪﹒（韓非子，內
儲說上）(=Zhu’s (3.20))
Given that the particle zhe in general
marks the agent of the event, a question naturally arises why the suo V zhe sequence obligatorily has the same reference of that without zhe and most
commonly marks the patient of the event. Fourth, there is a subject/object asymmetry held with respect to the suo construction.
That is, the suo V sequence can’t refer to the subject, but most commonly the object. Fifth, the occurrence of suo is obligatory
in this construction. For instance, a
sequence S V zhe is not allowed to denote a patient argument (Yiao (1998)).
The pattern of the suo sequence is
summarized as follows: (43)
(agent (zhi)) suo (modifier) V… (zhe) ((zhi) N) The particle suo can be used in modern
Chinese as well. When we compare the distribution of suo in classical and modern
Chinese, more questions arise. The modern Chinese counterpart of 民之所欲 is (44). (44) renmin suo yiao de dongxi
people SUO want DE thing There are a few properties of the suo
construction in modern Chinese. First, like classical suo, there is a subject/object
asymmetry. Second, in contrast to the classical suo, it can refer to patient,
locative, but not manner or reason. Third, unlike the classical suo, it cannot refer to
the object of the preposition. Fourth, it is optional, unlike the classical suo. Fifth,
generally, the agent doesn’t have to be introduced by the particle de, a modern
equivalent of zhi.
4. Analysis Proposed
Ting (1996, 1998) argues that suo in
modern Chinese is a resumptive clitic. I propose that this analysis be extended to the distribution of suo in classical Chinese.
More specifically, the suo sequence in
classical Chinese, like in modern Chinese, involves a relative clause. Suo is an X0 element, heading the NP in the
corresponding theta-position of the head noun of the relative clause. Due to a morphological requirement, it undergoes subsequent raising to I0
.
[IP [Op pro suo shi t ] pro ] (所食)
[IP [Opi NP suoi shi ti ] proi ] (民所食)
[IP [Opi NP suoi shi ti ] zhei] (民所食者)
[CP [Opi NP suoi shi ti ] zhi sui] (民所食之
粟)
minj zhi [CP [Opi proj suoi shi ti ] zhi sui]
(民之所食之粟)
minj zhi [IP [Opi proj suoi shi ti ] zhei] (民之
所食者)
Zhi, like its modern counterpart de, can be
treated as a complementizer. And a null operator is located in [spec/CP]. When there is no zhi, we may assume that there
is no projection of CP, but an IP. The null operator is adjoined to the IP. (cf.
Murasugi (1993) for relative clauses in Japanese) When the head noun does not occur as in suo shi and min suo shi, we
have instances on a par with English free relatives. Let’s assume that the head noun in the free relative is an empty pronoun. As for the instances where the logical subject is introduced by zhi, we suggest
that it is equivalent to the phrase (45) in modern Chinese
(45) renmin de [ xihuan chi de ] dongxi people DE like eat DE thing
The properties of the suo construction in
classical Chinese naturally follow from the proposed analysis. First, given this
resumptive clitic analysis of suo, the
subject/object asymmetry arises because the head movement out of an object position is licensed whereas that out of a subject position is not. It is clear that subject/object asymmetry holds in terms of head movement. This can be evidenced by two facts: first, there is no syntactic
subject clitic in French (Kayne (1975, 1983)) and in Italian (Rizzi (1986)); second, while noun incorporation which Baker (1988) argues undergoes head movement is well-documented from the object position, it is clearly forbidden from the subject position.
Now let’s turn to the question why the classical suo may refer to the object of
a preposition, in contrast to its modern Chinese counterpart. The closed class of morphemes in Chinese, most of which can be translated into English by means of prepositions, were transitive verbs at earlier stages of the language. (Li and Thompson (1974), citing Wang (1958)) (The following examples are quoted from Li and Thompson (1974))
(46) bu cao er zhang yi cong zhi. Must use your stick to follow (it) You must wield your stick to follow it. (Li ji (c. 5th c. B.C.)
(47) dui an bu shi. Face table not eat
(He) sits at the table and refuses to eat. (source unknown; quoted in Wang (1958))
(48) lan jing zi zhao. Hold mirror self reflect
(Shu Jing (c. 5th
century B.C))
There is evidence that coverbs are verbs in classical Chinese. First, in the subject +zhi+predicate construction, the coverb phrase yu (於) NP and yu (與) NP can occur in the predicate position. (Xu (1997, p. 216), Song (1983))
(49)寡人之於國，盡心焉耳矣（孟子、 梁惠王）
(50)今秦之與齊也，猶齊之於魯也﹒ （史記、張儀列傳）
Furthermore the object of coverbs can be dropped (Xu (1997, p. 218)), on a par with verbs. Coverbs in this category include yi (以), wei (為) and yu (與). (51) 小人有母，皆嘗小人之食矣，未嘗 君之羹，請以遺之﹒（左傳、隱公元年） (52) 聞死刑之報，君為流涕.（韓非子、 五蠹） (53) 旦日，客從外來，與座談﹒（戰國 策、齊策）
Coverbs also behave on a par with verbs in that they all require their
interrogative pronominal object to be preposed before them.
(54) 子歸，何以報我？（左傳、成公三 年）
(55) 吾誰與為親？（莊子、齊物論） Given that coverbs belong to the category of verbs in classical Chinese, now it should be clear why suo can stand for the
object of a preposition in classical Chinese. It is because the trace left by the head movement of suo in the object position of
the preposition can be properly governed, and thus licensed. With this in mind, let’s now consider the fact that suo can not only
stand for patient and locative, but also manner and reason in classical Chinese. Yi (1989), following Wang Kezhong (1982), suggests that all the instances of suo
standing for manner, reason or locative can be followed by an appropriate preposition. Based on this observation, I will follow Yafei Li’s (p.c. June 2000) suggestion that there is an empty preposition after suo in
the instances where suo is interpreted as
standing for locative, manner or reason of
the event. These instances are thus analyzed on a par with those where suo
stands for direct object of the predicate. We now turn to the question why
suo is obligatory in classical Chinese.
According to the economy principle (Chomsky (1991), movement is more costly than base-generation. Let’s assume that moving a head is less costly than moving a phrase. In forming a relative clause in classical Chinese, there are two options: base-generation and head movement of suo and null operator
movement. The less costly option blocks the more costly option. Under this analysis, instances like “sha ren zhe” where the subject is relativized involve null operator movement; in instances where suo is
licensed, the null operator movement is blocked. Furthermore, the fact that the suo V zhe sequence does not refer to the agent, but most commonly the patient of the predicate also follows. On the proposed analysis, zhe is bound by the null operator,
which also binds the NP that suo moves
out of. Thus, it is not surprising that the suo V zhe sequence is interpreted as that without the particle zhe.
5. Conclusion and self-evaluation
In this project, we have collected the basic facts of the suo construction in
classical Chinese and provided a proper analysis of it. This analysis has explained the contrasting behavior of suo in classical
and modern Chinese, shedding light on the change of the syntactic structure in
classical and modern Chinese. Given these reasons, we consider the primary goals expected in this project have been achieved.
6. References
Baker, Mark (1988) Incorporation: A Theory of Grammatical Function
Changing, Chicago: University of Chicago
Press.
Chomsky, Noam (1981) Lectures on Government and Binding, Foris,
Dordrecht.
economy of derivation and representation,” in Principles and parameters in comparative grammar,
edited by Robert Freidin, 417-454, Cambridge, MIT Press.
Chomsky, Noam (1995) The Minimalist Program, MIT Press, Cambridge.
Dong, Xiufang (董秀芳)(1998) “重新分 析與“所“字功能的發展，“古漢語研 究 40，50-55。
Guo, Xiliang, et al. (1981) 古代漢語，北 京出版社，北京。
Li, Charles N. and Sandra A. Thompson (1974) “Co-Verbs in Mandarin Chinese: Verbs or Prepositions?”Journal of Chinese Linguistics 2, 257-278.
Li, Charles and Sandra Thompson (1981)
Mandarin Chinese: A Functional Reference Grammar, University of
California Press, Berkeley.
Lu, Shuxiang (呂淑湘) (1974) 中國文法 要略，文史哲出版社，台北。
Ma, Jianzhong (馬建中)(1898) 馬氏文 通，商務印書館，1986。
Murasugi, Keiko (1993) The Generalized Transformation Analysis of Relative Clauses and Island Effects in Japanese,
manuscript, Kinjo Gakuin University. Peyraube, Alain (1996) "Recent Issues in Chinese Historical Syntax", in New
Horizons in Chinese Linguistics, eds. C.-T.
James Huang, and Y.-H. Audrey Li, Kluwer academic publishers.
Pulleyblank, Edwin (1995) Outline of Classical Chinese Grammar, UBC Press,
Vancouver. Qi, Yonghua (祁永華)(1992) 文言句式, 吉林文史出版社。 Song, Shaonian (宋紹年)(1983) “試談 ＜史記＞的幾種句法結構，“語言學論 叢，145-165。
Ting, Jen (1996) "The Resumptive Pronominal Clitic Suo in Mandarin Chinese," Bulletin of National Pingtung Institute of Commerce 4, 309-335,
Pingtung, Taiwan.
Ting, Jen (1998) The Nature of the Particle Suo in Mandarin Chinese, ms,
National Chiayi Teacher’s College.
Wang, Kezhong (王克仲)(1982) “關於先 秦所字詞性的調查報告，”古漢語研究論 文集，北京出版社。
Wang, Li (1958) Han-yu shi-gao (History of the Chinese language,) Vol. 2. Peking: ke-xue chubanshe. Wang, Li (王力)(1976) 古漢語通論，中 外出版社，香港。 Xiao, Taifang (蕭泰芳)(1988) “所﹢介 詞﹢動詞“結構略說，語文研究 26， 52-53。 Xu, Qiting (徐啟庭) (1997) 古今漢語語 法異同，調和文化出版事業公司，高雄。 Yang, Boujun and Dongshi He (楊伯峻, 何東士) (1992) 古漢語語法及其發展， 語文出版社，北京。 Yao, Zhenwu (姚振武)(1998)個別性指稱 與“所“字結構，古漢語研究 40，45-49。 Yi, Menchung (易孟醇)(1989) 先秦語法 (Pre-Qin Grammar,) Hunan educational publisher.
Yie, Baomin, Xiu Yan and Jianqiao Yang (葉保民、嚴修、楊劍橋)(1992) 古代漢
語（上），張世祿主編，洪葉出版社，台
北。
Zhang, Zhong-xing (張中行) (1992) Wen-yan Chang-shi (Common knowledge of the classical Chinese,) Xinwenfeng Publisher, Taipei.
Zhou, Fagao (周法高) (1961) Zhongguo gudai yufa: zaoju bian, Academia Sinica, Taipei. Zhou, Shumin (周淑敏)(1993) 古漢語自 學入門，中國人事出版社，北京。 Zhu, Dexi (朱德熙) (1983) “自指與轉 指，”方言，第一期。 Zhu, Xing (朱星)(1996) 朱星古漢語論 文選集，洪葉出版社，台北。
