Prosodic Morphology in
Mandarin Chinese
It is not entirely clear if Modern Chinese is a monosyllabic or disyllabic
language. Although a disyllabic prosodic unit of some sort has long been
considered by many to be at play in Chinese grammar, the intuition is not
always rigidly fleshed out theoretically in the area of Chinese morphology. In
this book, Shengli Feng applies the theoretical model of prosodic morphology
to Chinese morphology to provide the theoretical clarity regarding how and
why Mandarin Chinese words are structured in a particular way. All of the
facts generated by the system of prosodic morphology in Chinese provide new
perspectives for linguistic theory, as well as insights for teaching Chinese and
studying of Chinese poetic prosody.
Shengli Feng is Professor of Chinese Linguistics at the Chinese University of
Hong Kong, Director of the CUHK-BLCU Joint Research Center for Chinese
Linguistics and Applied Linguistics and Yangtze Scholar Chair Professor at the
Beijing Language and Culture University.
Prosodic Morphology in
Mandarin Chinese
Shengli Feng
First published 2018
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
and by Routledge
711 Third Avenue, New York, NY 10017
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2018 Shengli Feng
The right of Shengli Feng to be identified as author of this work has been
asserted by him in accordance with sections 77 and 78 of the Copyright,
Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or reproduced or
utilised in any form or by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying and recording, or in
any information storage or retrieval system, without permission in writing
from the publishers.
Trademark notice: Product or corporate names may be trademarks or
registered trademarks, and are used only for identification and explanation
without intent to infringe.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
A catalog record for this book has been requested
ISBN: 978-1-138-22835-1 (hbk)
ISBN: 978-1-315-39278-3 (ebk)
Typeset in Bembo
by Apex CoVantage, LLC
Contents
1
Introduction
1
2
Prosodic word as an origin of compounds in Classical Chinese
14
3
Monosyllabicity and disyllabicity
68
4
Prosodically constrained compound formation
90
5
Minimal and maximal word effects
115
6
Prosodic register grammar
136
7
Compound prosodic word – Sizige 四字格164
8
Conclusion and final remarks
185
Index192
1Introduction
This book studies prosodic morphology in Chinese.
Prosodic morphology concerns the shapes and sizes of canonical words,
affixation as well as word formation in a language. Although being a new area
in general linguistics and especially in Chinese linguistics, many prosodic morphological phenomena have been recognized over the past half-century. The
early Republican philologist Huang (1932/1983:99) 黃侃 first pointed out the
following fact:1
In Chinese, the sound (refers to morpheme, by Feng) is monosyllabic while the tone
(refers to prosodic unit, by Feng) is disyllabic, and thus monosyllabic words often
become disyllabic ones, giving rise to the fact of one meaning with two words: one is a
monosyllabic while the other is disyllabic, such as tia-n 天 ‘sky’ and huángtia-n皇天
‘sky’ or hàotia-n 昊天 ‘sky’. The disyllabic words are redundant in plain speech
(zhìyán質言), but useful in literary language. Because of the properties given above,
Chinese poems and parallel proses are created accordingly. Other types of versification of the language are also formed by these two features.
In Chinese, the morphemes are monosyllabic while the rhythmic units are
disyllabic. It is also recognized that monosyllabic words often become disyllabic
ones in literary texts, as pointed out by Huang (1932/1983). In the late 1930s,
Guo 郭紹虞 (1938) also pointed out the syllabic flexibility of Chinese vocabularies in terms of elasticity (elastic words 彈性詞), and in the early 1960s, Lü
(1963) recognized the morphosyntactic preferences between [2+1] and [1+2]
syllabic patterns (namely, forms with [σσ+σ] and [σ+σσ] syllable structure) in
Modern Chinese. For example:
1. a. 鞋店
xié diàn
shoes shop
‘shoes shop’
鞋帽店
xié mào diàn
shoes hat shop
‘shoes & hat shop’
*鞋商店
*xié sha-ngdiàn
shoes shop
‘shoes shop’
鞋帽商店
xié mào sha-ngdiàn
shoes hat shop
‘shoes & hat shop’
2 Introduction
b.種樹
zhòng shù
plant tree
‘to plant trees’
*種植樹
*zhòngzhí shù
plant tree
‘to plant trees’
種果樹
zhòng guǒshù
plant fruit tree
‘to plant fruit trees’
種植樹木
zhòngzhí shùmù
plant tree
‘to plant trees’
The most recent and important works on prosodic morphology were initiated by Lu and Duanmu in 1991 and Feng in 1995.The former proposed a stress
theory which successfully captured the difference between the [2+1] (nominal)
and [1+2] (verbal) patterns in Chinese grammar, while the latter introduced
the theory of prosodic morphology (McCarthy and Prince 1993) into Chinese
linguistics and developed a subsystem of Chinese prosodic morphology.
This book is mainly based on the newly developed theory of prosodic morphology in Chinese and argues that the crucial point in prosodic morphology,
as defined by McCarthy and Prince (1993:79–153), is as follows:
“The right/left edge of some grammatical constituent coincides with the corresponding edge of some phonological constituents.”
As we will see in the following chapters, the sizes of Chinese morphological
categories, namely, morphemes and words, would coincide with the prosodic
categories of mora and foot respectively in the language (Feng 1995, 2009).
According to the theory of Alignment: [M] = [σ] (namely, a morpheme coincides with a syllable), the notion of morphosyllabicity, first coined and defined
by DeFrancis (1986), is formulated as a prosodic constraint in the following:
2. Morphosyllabic Constraint (MC)
[M] = [σ]
ALIGN: M-Edge, σ-Edge = Left, Right
In (2), “M” stands for morpheme and “σ” for syllable. The operation system
of ALIGN (alignment) requires the left and right edges of a morpheme to be
coincided with the left and right edges of a syllable respectively. Given the Morphosyllabic Constraint in (2), this book demonstrates that a syllable in Modern Chinese corresponds to a morpheme, which demands that the indigenous
morphemes in Chinese are monosyllabic. The MC in (2) successfully captures
the essential characteristics of the Chinese languages among the Sino-Tibetan
family as what Li 李方桂 (1973:2) has pointed out:
One of the characteristics of this family (Sino‑Tibetan) is the tendency toward Monosyllabism. By Monosyllabism we do not mean that all the words in these languages
consist of single syllables, but that a single syllable is an important phonological unit
and often is a morphemic unit, the structure of which is rigidly determined by the
Introduction
3
phonological rules of the language, and serves as the basis for the formation of words,
phrases, and sentences.
The correctness of the prediction by (2) is also evidenced by the statistics
shown in Shen (2007). Briefly, there is a total of 41,915 words in the Fifth Edition of Xiàndài Hànyǔ Cídiǎn 現代漢語詞典 Modern Chinese Dictionary (2005),
but only 849 (3%) of these words are polysyllabic, which are arguably all nonindigenous in nature in the sense that they are either loan words from other
languages, or words passed down from Classical Chinese thousands of years ago.
Despite the complexity of the origins of polysyllabic words, they are neither
indigenous morphemes, nor root morphemes in the morphology of Mandarin
Chinese (see Sproat and Shih 1996, Feng 2009).
Phonologically, a well-known phenomenon in Chinese is the fact that there is
no resyllabification process in the language. For example, the process of CVC|VC
→ *(CV (CVC) lín-a-n 林庵 → *lí-nán 黎楠 is not allowed. The lack of a resyllabicification process in Chinese phonology is arguably a consequence of the
Morphosyllabic Constraint, that is, the morpheme-final consonant or vowel
must occupy the final position in the corresponding syllable, and the morphemeinitial consonant or vowel must occupy the initial position in that syllable. Consequently, a “morpheme mid-syllable/consonant” will not de-align a morpheme
(see McCarthy and Prince 1993:38).This may be the reason why there is no such
‘de-alignment’ operation (resyllabification) in Mandarin Chinese.
As we will see in Chapter 3, the Morphosyllabic Constraint in (2) also leads
to a phonological reduction when lexical morphemes become functional ones.
Kratochvil (1977) has observed: “under some conditions it (i.e., “the leftward
movement of stress” in a disyllabic word – Feng, 1995:107) causes atonicity,
reduction in the segmental structure, and ultimately the loss of syllable status of
B altogether, and the fusion of B with A (in an [A+B] construction)” and thus,
“Modern Peking Dialect shows signs of a process involving syllable fusion as
its ultimate result” (Kratochvil 1977:26–27). Note that this process exclusively
happens to functional elements and no root morphemes have undergone a
phonological reduction in the language. This shows a great possibility that the
phonological reduction of the second syllable in disyllabic words may be a
result of the Morphosyllabic Constraint.That is, all root morphemes follow the
MC in (2) and only (semi-)functional elements (root morphemes that have lost
their lexical meaning in a disyllabic form) are exceptional. For example,
3. Monosyllabic Word
600 AD
孩
hái
Disyllabic Word
11th century AD
孩-兒
hái-ér
Monosyllabic Word
Mandarin Chinese
孩兒
háir
Hái-ér 孩兒 ‘child-son’ is a disyllabic word formed by hái 孩 ‘child’ plus a
monomorphemic nominal suffix -ér 兒 which etymologically means ‘son’ or
‘child’ in Classical Chinese and begins to be used as a diminutive suffix around
4 Introduction
the Tang Dynasty (618–907 AD) (Norman 1988:114). However, the syllable ér
in almost all nouns of Mandarin Chinese has been reduced to a /r/ feature only,
and fused with the proceeding syllable, thus yielding what Kratochvil (1977)
called a fusion syllable.
Are there disyllabic or polysyllabic words in Chinese? The answer is yes, but
they are overwhelmingly formed by a compounding of monomorphemes in
prosodic morphology. While the morphemes in Chinese coincide with syllables, the combination of morphemes coincides with a prosodic category bigger
than the syllable in the Prosodic Hierarchy given below:
4. Prosodic Hierarchy
Prosodic Word (PrWd = Compound)
|
Foot
|
Syllable
|
Mora
According to the Prosodic Hierarchy proposed in McCarthy and Prince (1993),
the next hierarchical category above the syllable is the foot, and again the next
prosodic category above the foot is the Prosodic Word (PrWd, for short). In
fact, a PrWd is realized by (or mapped from, in technical terms) a foot. As a
result, the foot is essential in determining the morphological category PrWd.
What is a foot in Chinese? In the first place, what is a foot cross-linguistically?
Foot is known as a kind of rhythmic unit.What is rhythm then? In English, rhythm
has long been recognized by scholars like Lanier, who argues in his The Science of
English Verse (1880) that if equal or proportionate intervals of time are marked off
to any of our senses by recurrent stresses of similar events, we may be said to perceive a primary rhythm through that sense. In other words, regular recurrence of
all kinds of objects perceived by our physical sensors from hearing, seeing, moving
and so on. What is important for rhythm is the fact that we can never perceive
rhythm without the fundamental alternation between strong and weak elements,
and this is what Liberman (1975) has discovered and defined, namely, the Relative Prominence Principle, for all rhythm-related prosodic phenomena, which
brings forward the concept that a minimal rhythmic unit is a binary formed by
two elements, that is, a stronger one with a weaker one, because there would have
been no stronger elements without others being weaker.The relative prominence
principle is well-tested by our perception of the following fact:
A clock which ticks seconds may be said to set up a primary rhythm for the ear which
hears each recurrent tick. These ticks are exactly alike: they fulfill the definition of
Introduction
5
primary rhythm, which describes it as a conception resulting from a similar event
recurring at equal (or simply-proportionate) periods of time. But everyone who has
been in a room alone with a ticking clock must have observed that every other
tick seems to be different, somehow, from its fellow, as if it said, “Tick-tack, ticktack . . .”; and the effect of this difference is to arrange the whole series into groups,
of two ticks in each group.
(Martin 2012:125)2
The “tick-tack” seems to be universal, but the stronger-weaker rhythm does
not seem to be the same among all languages, because the phonological systems
of these languages are different from one another. What would the foot unit in
Chinese then?
The foot formation in Chinese can be successfully tested by using nonsense
syllable strings (onomatopoeic words, sound translation of foreign names, etc.)
or syntactically non-structured words (a string of identical numbers) or syntactically equal-structured words (coordination of monosyllabic words), examples
of which are listed as follows (‘ ( ) ’ represents a rhythmic group):
5. a. (hong long), (hong long), (hong long) . . . (the sound of a thunderstorm)
(guangdang, guangdang), (guangdang, guangdang) . . . (the sound of a train)
(GAda gada GAda gada) (GAda gada GAda gada) . . . (the sound of a
horse running)
b. 柴 米 油 鹽 醬 醋 茶
(chái mǐ ) (yóu yán) (( jiàng cù) chá )
firewood rice oil salt sauce vinegar tea
c. 加 利 弗 尼 亞
(jia- lì ) (( fúní ) yà)
California
d. (99) (99) ((99)9)
The rhythmic groups in (5) are rightwardedly organized into disyllabic units
(foot) with the stray syllable being attached to the last foot when the syllable
string contains an odd number. This is called Natural Foot Formation (Feng
1998), which is formulated as follows:
6.
f
σ
σ
Natural Foot Formation (NFF)
A natural foot in Mandarin Chinese is grouped by two syllables from left to right
and the stray syllable is attached to the neighboring foot when the number of syllables is odd.
6 Introduction
The generalization of NFF has a number of implications in Chinese prosodic
morphology.
First, as seen in the Prosodic Hierarchy stated in (4), the prosodic foot will
interact with (to impose on) the morphological category, resulting in what is
called the Prosodic Word (PrWd for short) in a language. If a standard foot
in Chinese is disyllabic, the standard PrWd in Chinese will also be disyllabic,
which directly dominates the morphological processes of the language, as will
be seen in Chapter 4.
A remarkable effect of PrWd’s constraint on Chinese morphology is
shown in the reduplication process of the language. For example, the outcome of noun reduplication (meaning ‘every noun’) in Chinese must be a
PrWd, and thus, jia--jia- 家家 ‘family-family, every family’, nián-nián 年年
‘year-year, every year’ are acceptable but *xı-ngqı--xı-ngqı- 星期-星期 ‘weekweek, every week’ is not, because the latter is bigger than the size of a
PrWd.
The theory of prosodic word explains properly what Chinese compounds
come about, as first proposed in Feng (1997) (where ‘M’ stands for morpheme,
‘σ’ for syllable):
7.
Compound
prosodic morphology
PrWd
f
σ
M
σ
M
prosodic phonology
According to (7), “a compound in Chinese must first be a prosodic word,
though a prosodic word is not, by necessity, a compound.” This generalization
captures the facts that (i) word formation in Chinese is overwhelmingly (if not
exclusively) a compound formation, that is, a process of combining a monosyllabic morpheme/word with another monosyllabic morpheme/word, and thus,
(ii) the majority of Chinese compounds are disyllabic.That is to say, new words,
rather than the old and most commonly used ones like shǒu 手 ‘hand’, tóu 頭
‘head’, niú 牛 ‘cow’, yáng 羊 ‘sheep’, etc., that were passed down from Classical
Chinese thousands of years ago and are thus exceptional to the modern prosodic constraint (that is, applications of prosodic morphological rules are sensitive to different classes of morphemes and morphological (sub-)categories), are
formed almost exclusively by no fewer than two syllables in Mandarin Chinese.
Thus the standard size of all new (compound) words is overwhelmingly disyllabic, which is borne out as predicated by the following statistics (Zhang 1997):
disyllabic words make up 49,641 (70.6%) of the total 70,343 words in Mandarin Chinese.
Introduction
7
Second, the NFF also entails that monosyllabic forms cannot stand alone
where an independent prosodic unit is required. This is evidenced by the following example:
8. A: 你幾歲？
nǐ jǐ suì ?
2SG how many year
‘How old are you?’
B: 十五。
shíwǔ.
Ten five, fifteen
‘I am fifteen years old.’
C:a. *五
*wǔ
five
‘I am five years old.’
C: b. 五歲。
wǔ suì。
five year
‘I am five years old.’
In Chinese you may answer this question about the age by mentioning any
polysyllabic number, but if you want to specify a monosyllabic number, you
have to add the syllable suì 歲 ‘year old’ for otherwise the sentence is unacceptable. The same is true for monosyllabic place names and dates in Chinese, as
generally observed (see also examples given in Chapter 2).
Given the unacceptable monosyllabic forms in prosodic parsing, it follows
that trisyllabic units should be allowed by the grammar even if they are highly
conditioned. In other words, when a monosyllabic morpheme or a word is
used, it must attach to a neighboring foot in order not to be ruled out by the
NFF. This has in fact resulted in what is called a Super Foot Formation (SFF
for short), which has given rise to trisyllabic compounds in the language.
Given the NFF and the SFF, the sizes of Chinese wordhood produced in
morphology will be at minimum two and at maximum three syllables long
under the prosodic morphological system outlined above. With the exception
of loan words and phrasalized expressions, the [2 ≮ wordhood ≯3] generalization for word size is true for 82.4% of compound words produced by the
prosodic word formation (Zhang 1997, Zhou 1998).
The third important implication of the NFF is the grammatical function of
the foot directionality: left-footing (i.e., 1+2, thereafter) is preferred by word
formation while right-footing (i.e., 2+1, thereafter) is favored by phrasal prosody in Chinese prosodic morphology. Compare:
9. 2+2
進口商品
1+2
進商品
2+1
進口商
8 Introduction
jìnkǒu sha-ngpǐn
enter mouth
economic goods
‘imported product’
‘to import product’
複印文件
fùyìn wénjiàn
copy document
‘a copied document’
‘to copy document’
jìn sha-ngpǐn
enter economic goods
‘to import product’
‘*imported product’
jìnkǒu sha-ng
enter mouth
economic
‘importer’
印文件
yìn wénjiàn
copy document
‘*a copied document’
‘to copy document’
複印件
fùyìn jiàn
copy document
‘a copied document’
‘*to copy document’
Fùyìn wénjiàn 複印文件 ‘to copy document or a copied document’ is a [2+2]
syllable pattern and hence the footing directions make no difference whether it
is leftwarded (i.e., [2+2]) or rightwarded (also [2+2]) and as result, the outcome
of the [2+2] word strings can be either a word or a phrase:
10. 兩份複印文件
liǎng fèn fùyìn-wénjiàn
two CL. copied-document
‘two copied documents’
複印了兩份文件
fùyìn le liǎng fèn wénjiàn.
copy Asp. two CL. document
‘to have copied two documents.’
However, the double-directional property of [2+2] will not be shared by the
[1+2] and [2+1] rhythmic structures, because [1+2] is rightwarded and [2+1]
is leftwarded. As a result, the [1+2] and [2+1] rhythmic structures cannot have
the dual properties (i.e., being either a phrasal or a word category) as the [2+2]
forms do, which is shown by the following facts.
11. 皮鞋工廠
pí-xié
go-ng-chǎng
leather-shoe
worker-mill
‘leather shoe
factory’
皮鞋工
pí-xié go-ng
皮廠
pí chǎng
皮工
pí go-ng
鞋工
xié go-ng
鞋廠
xié chǎng
leather mill
leather worker
shoe worker
shoe mill
‘leather
factory’
‘leather worker’ ‘shoemaker’
皮鞋廠
pí-xié chǎng
*鞋工廠
*xié go-ng-chǎng
‘shoefactory’
*皮工廠
*pí go-ng-chǎng
Introduction
leather-shoe
worker
‘leather shoe
factory’
leather-shoe mill
shoe worker-mill
‘leather shoe
factory’
‘shoe factory’
大皮鞋
dà pí-xié
big leather-shoe
‘a big leather shoe’
9
leather
worker-mill
‘leather
factory’
小工廠
xiǎo go-ng-chǎng
small worker-mill
‘a small factory’
The interesting footing-effect lies in this: noun+noun compounds favor the
[2+1] rhythmic pattern while the adjective+noun phrases prefer the [1+2]. It
has been commonly assumed that the combination of Noun+Noun (like píxié
chǎng 皮鞋廠 ‘leather shoe factory’) creates compound words, while that of
Adjective+Noun (dà go-ngchǎng 大工廠 ‘big factory’) produces phrases in Chinese (Duanmu 1990).
Given this, it is expected that the [1+2] pattern is not acceptable for [N+N]
compounds but perfect for [A+N] phrases because it belongs to phrasal prosody, as seen in (12). A corpus-analysis (Duanmu 2011) shows that only 1% of
[N+N] compounds in Chinese are formed by the [1+2] syllable pattern such
as jı-n xiàngliàn 金項鏈 ‘gold necklace’, zhǐ lǎohǔ 紙老虎 ‘paper tiger’, etc. However, even if the jı-n xiàngliàn and zhǐ lǎohǔ exist in Mandarin Chinese, it does
not mean that jı-n 金 ‘gold’ and zhǐ 紙 ‘paper’ can be freely used to create [1+2]
nominal compounds, as the following examples show:
12. a.
*金工廠
*jı-n go-ngchǎng
gold factory
‘a gold factory’
b.
*紙工廠
*zhǐ go-ngchǎng
paper factory
‘a paper factory’
Note that jı-n xiàngliàn 金項鏈 ‘gold necklace’ and jı-n go-ngchǎng 金工廠 ‘gold
factory’ are different.The former means that ‘the necklace that is made of gold’,
while the latter refers to ‘the factory that produces gold’. “Being made of gold”
and “producing gold” have two different internal relationships between elements within the nouns (i.e., necklace and factory). When jı-n 金 ‘gold’ is used
with a meaning of “made of gold” as in jı-n xiàngliàn 金項鏈 ‘gold necklace’,
it functions as a property classifying the head “necklace”, which is the reason
why it uses the phrase prosody of [1+2] to describe the head, the result of
which is acceptable. However, when jı-n 金 ‘gold’ is used with to mean a “product” as in jı-n go-ngchǎng 金工廠 ‘gold factory’, it occurs in a position generated
by compound formation and hence it cannot use the [1+2] phrasal prosody,
and thus the result is unacceptable (12). Interestingly, if jı-n go-ngchǎng 金工廠
is understood, even if the semantics is unrealistic, as ‘a factory that is made
of gold’, then the result is acceptable exactly like ‘gold necklace’ (the same is
10 Introduction
true with zhǐ go-ngchǎng 紙工廠 if it is understood as ‘a factory that is made of
paper’). Apparently, the prosodic system recognizes the phrasal semantics and
compound semantics by allowing the former with [1+2] pattern and later with
[2+1] pattern, which shows the grammatical function of foot directionality.
Finally, the MC, NFF and SFF combine to derive a notion of the minimal
word in Chinese. For example, only by conforming to the size of a minimal
word, (i) can a [VO] be formed to take an outer object as seen in (13a), (ii) can
an [Auxiliary+V] form become an adjective as in (13b), (iii) can a [VO] form
be used as an adverb as seen in (13c), and finally, (v) can a [size Adjective+N] be
modified by color adjectives as seen in (13d).
13. a. *開玩笑他
*ka-i wánxiào tamake joke him
‘make fun of him’
a’
取笑他
*qǔ xiào tatake joke him
‘make fun of him’
b. *非常可懷疑
*fe-icháng kě huáiyí
extremely can suspicious
‘extremely suspicious’
b’
非常可疑
fe-icháng kěyí
extremely can suspicious
‘extremely suspicious’
c’ *並肩膀戰鬥
bìng jia-n zhàndòu
juxtapose shoulder fight
‘fight side-by-side’
c.
並肩戰鬥
*bìng jia-nbǎng zhàndòu
juxtapose shoulder fight
‘fight side-by-side’
d. *黑大汽車
*he-i dà qìcheblack big vehicle
‘a big black vehicle’
d’
黑大雁
he-i dà yàn
black big goose
‘a big black goose’
As seen above (and the similar examples in following chapters), prosodic
morphology in Chinese has its unique characteristics and this book will make
a deeper investigation into these issues. It will be seen, first, that, unlike the
situation in other languages where prosody only dominates affixation, prosody
in Chinese morphology mainly affects compound word formation due to the
morphological processes that took place in the Han dynasty, as discussed in
Chapters 2 and 3. Second, prosodic morphology in Chinese directly interacts with syntax, which will be shown in discussions of Chapter 4 about foot
formation on syntactic trees resulting in compounds, and in Chapter 5 about
Minimal Word effects on syntax. Finally, prosody not only constrains morphology but also functions as (part of) morphology, which may better be considered
as morphological prosody serving for Register Grammar in Chinese, as seen in
the investigation of Chapters 6 and 7 on monosyllabic (elevated) words used
in disyllabic template (i.e., Qiàn-ǒu Cí 嵌偶詞) and disyllabic (formal) words
Introduction
11
used in quadri-syllabic template (Hé-ǒu Cí 合偶詞) and compound prosodic
word (i.e., Sìzì Gé 四字格 ‘the Four-Syllable Expressions’). All of these forms
produced in Chinese language are morphological results uniquely motivated
by prosody.
We are fully aware of the fact that morphology is a complex system: there
are not only rules but also exceptions to whatever generalizations that are come
up with, and these exceptions are actually caused by different factors. This is
so because the notion of wordhood can be derived from morphology, syntax
and phonology and can also be created by specialists in special areas for special
purposes (such as cuisine terminologies). Facing the complicated situations of
word-forming capacity, of human linguistic competence and of the linguistic
recourses for making words at different components of the grammar (phonology, prosody, semantics, syntax, register, etc.), it is easy to doubt the validity
of rules that produce morphological results without exceptions, and turn to
believe tendencies that tolerate arbitrary exceptions. However, instead of just
showing tendency patterns, we have tried very hard in this study to follow the
Galileo method:
Galileo plowed his way through this, putting much of the data aside, redefining
what was relevant and what was not relevant, formulating questions in such a way
that what appeared to be refuting data were no longer so, and in fact very often just
disregarded data that would have refuted the system.This was done not simply with
reckless abandon, but out of a recognition that explanatory principles were being
discovered that gave insight into at least some of the phenomena.3
This is exactly what Noam Chomsky and Morris Halle (1968) did in their
book The Sound Pattern of English, as they explained (1968:172):
It is quite obvious that many of the phonological rules of the language will have
certain exceptions which, from the point of view of the synchronic description, will be
quite arbitrary. This is no more surprising than the fact that there exist strong verbs
or irregular plurals. Phonology, being essentially a finite system, can tolerate some
lack of regularity (exceptions can be memorized); being a highly intricate system,
resulting (very strikingly, in a language like English) from diverse and interwoven
historical processes, it is to be expected that a margin of irregularity will persist in
almost every aspect of the phonological description. Clearly, we must design our
linguistic theory in such a way that the existence of exceptions does not prevent the
systematic formulation of those regularities that remain.
Following the steps of previous scholars and based on their studies of prosodic morphology and Chinese linguistics, I initiated the present research with
some new theories on Chinese prosodic morphology which tackled some
of the traditional and enigmatic problems, providing some new perspectives
with new results in this book. Of course, it does not mean that the problems
raised in Chinese morphology are all settled after this study, but, following
12 Introduction
what McCawley has taught us: “what is important in scholarly work is not
the analytical framework used but the insight reached” (Mufwene et al. 2005),
I hope this book will bring some new insights and new ways of dealing with
the old problems such as the monosyllabic myth, the origin of disyllabification,
the suprasegmental morphology (after segmental morphology), etc. Whether
there are true insights or not, they are what I have always been trying to reach
in the past, for the present and also the future, quoting from the Analects of
Confucius, “I am not saying that I am capable of doing it, I am learning toward
it 非曰能之，願學焉!”
Notes
1 The original Chinese reads: “中國語言音單調複，故往往變單字為雙字。而每一名
有單名雙名二者。如‘天’稱‘皇天’、‘昊天’是也。雙名在質言為贅疣，而于文言則
需用之。由於音單調複，單名多變為雙名，因之製成詩歌、駢文等文體。他如有
節奏之語言，也皆由此形成也。
” —— 《文字聲韻訓詁筆記》99頁。
2 Martin, William (2012) Joyce and the Science of Rhythm. New York: Palgrave Macmillan.
3 Taken from “An Interview With Noam Chomsky.” Linguistic Analysis. 1978. 4:301–319.
References
Chomsky, Noam and Halle, Morris. 1968. The Sound Pattern of English. New York: Harper
and Row.
DeFrancis, John. 1986. The Chinese Language: Fact and Fantasy. Honolulu: University of
Hawaii Press.
Duanmu, San. 1990. A Formal Study of Syllable, Tone, Stress and Domain in Chinese Languages.
Doctoral dissertation. MIT Press, Cambridge, MA.
Duanmu, San. 2011 [2016]. “Wordhood in Chinese”. In Wolfgang Behr, Gu Yueguo, Zev
Handel, C.-T. James Huang, and Rint Sybesma (eds.) Encyclopedia of Chinese Language and
Linguistics. Leiden: Brill.
Feng, Shengli. 1995. Prosodic Structure and Prosodically Constrained Syntax in Chinese. Ph.D. dissertation. Philadelphia: UPENN.
Feng, Shengli. 1997. “Prosodic Structure and Compound Words in Classical Chinese”.
In Jerome Packard (eds.) New Approaches to Chinese Word Formation. Berlin: Mouton de
Gruyter, 197–260.
Feng, Shengli. 1998.“On ‘natural feet’ in Chinese”. Chinese Philology 中國語文 1.262: 40–47.
Feng, Shengli. 2009. “Monosyllabicity and Disyllabicity in Chinese Prosodic Morphology”.
Macao Journal of Linguistics, 1: 4–19.
Guo, Shaoyu. 1985 [1938]. “The Elastic Property of Chinese Word Length中國語詞之彈
性作用”. In Shaoyu Guo (eds.) Zhao Yu Shi Yuyan Wenzi Lunji 照隅室論言文字語集.
Shanghai Guji Press. Original: Collection of Linguistic and Philological Works Yenching Journal
of Chinese Studies 燕京學報 24, 1938.
Huang, C.-T. James. 1982. Logical Relations in Chinese and the Theory of Grammar. Ph.D. dissertation. Department of Linguistics and Philosophy, MIT, Cambridge, MA.
Huang, Zhuo (eds.) 1983. Huang Kan's 1932 Work: Notes on Paleography, Phonology and Exegesis in Classical Chinese 文字聲韻訓詁筆記, Wuhan: Wuhan University Press.
Kratochvil, Paul. 1977. “Traditions in Chinese Linguistics: Fact or Fiction?” Cahiers de Linguistique Asie Orientale 1: 17–30.
Introduction
13
Li, Fang-Kuei. 1973. “Language and Dialects of China”. Journal of Chinese Linguistics 1(1): 1–13.
Liberman, Mark. 1975. The Intonation System of English. Ph.D. dissertation, MIT, Cambridge, MA.
Lu, Bingfu and San, Duanmu. 1991. “A Case Study of the Rhythm Between Rhythm and
Syntax in Chinese”. Paper presented on the Third American Conference on Chinese
Linguistics.
Lü, Shuxiang. 1962. “On ‘free’ and ‘bound’”. 說‘自由’與‘粘著’ Chinese Philology 中國語文
1: 1–6.
Martin, William. (2012) Joyce and the Science of Rhythm. New York: Palgrave Macmillan.
McCarthy, John and Alan, Prince. 1993. “Generalized Alignment”. In Geert Booij and J. van
Marle (eds.) Yearbook of Morphology 1993. Dordrecht: Kluwer, 79–153.
Mufwene, Salikoko S., Francis, Elaine, and Wheeler, Rebecca S. 2005. Polymorphous Linguistics: Jim McCawley’s Legacy. Cambridge, MA: MIT Press.
Norman, Jerry. 1988. Chinese. Cambridge: Cambridge University Press.
Shen Huaixing. 2007. “On Entries of Binome Dictionary and Related Questions
《聯綿字典》的收詞及相關問題”, Lexicographical Studies Císhu- Yánjiu- 辭書研究
4: 85–92.
Sproat, Richard and Chilin, Shih. 1996. “A Corpus-Based Analysis of Mandarin Nominal
Root Compound”. Journal of East Asian Linguistics 5.1: 49–71.
Zhang, Kai. 1997. “A Statistic Analysis of Basic Characters in Chinese Word Formation
漢語構詞基本字的統計分析”. Language Teaching and Linguistic Studies 語言教學與研究
1: 43–52.
Zhou, Jian. 1998. “The Identity of Lexical Entries in Dictionaries: From a Perspective of Word Length 從詞長看詞典語彙單位的確定”. Essays on Chinese Lexicography
中國辭書學文集 2: 34–39.
2Prosodic word as an origin
of compounds in Classical
Chinese
Based on previous investigations (Dobson 1959, Chou 1962, Feng 1995, 1997a),
this chapter discusses the nature of compound words in Classical Chinese1 and
claims, following Pulleyblank (2000), Pan (2000), Behr (2004), Zhengzhang
(2003, 2017), Zhao (2014), Ho (2016) etc., that the prosodic phonological system changed from Archaic Chinese to Medieval and Modern Chinese, which
gave birth to disyllabic foot formation and further motivated compound formation to satisfy the prosodic word requirements. This investigation mainly
concentrates on the periods from the Pre-Qin period (221 BC) to the Han
dynasties (202 BC–220 AD). This is because compound words in Classical
Chinese, as we will see below, developed to a large extent during the Han
dynasties and as argued by Huang 2005, Xu 2006 and Feng 2005, a typological change from syntheticity of Archaic Chinese to analyticity of Medieval/
Modern Chinese occurred between the periods of Western and Eastern Han
dynasties.
There are three crucial points being discussed in this chapter. First, it is
shown that disyllabic compound formation was rarely productive in Archaic
Chinese and the number of compound words in Classical Chinese increased
sharply during the Han dynasties. Second, such a development of disyllabic
compounding in Classical Chinese is chiefly due to disyllabic foot formation,
which was newly established around the Han dynasties as a result of the loss of
bimoraic feet in Old Chinese (c. 1000 BC). Third, it is argued that compounds
in Classical Chinese are not only syntactic words, but also prosodic words. The
former is shown by syntactic relations among each part of the compounds,
and the latter is derived from the Prosodic Hierarchy and Foot Binarity in the
theory of prosodic morphology.
This chapter is organized as follows: section 2.1 examines the criteria for
identifying compounds in Classical Chinese; section 2.2 presents a comparative
study of Mencius (c. 372–289 BC) and the commentary on Mencius by Zhao
Qi (c. 107–201 AD); section 2.3 discusses previous accounts on development
of compounding; section 2.4 investigates the development of disyllabicity and
proposes that the development of disyllabicity is independent of compounding; section 2.5 discusses the phonological changes of Old Chinese (OC) and
Prosodic word in Classical Chinese
15
proposes that the change of CVC basic (minimal) syllable structure of Old
Chinese to a CV basic (minimal) syllable structure of Medieval Chinese (MC),
alongside of the development of tone system, inevitably results in a loss of
bimoraic foot formation.
In this section, it is shown that the loss of bimoraic feet was compensated for
by the introduction of disyllabic feet, and disyllabic combinations are therefore
produced in sharply increased quantity during or after the phonological change
took place. Given this historical development and the monosyllabic nature of
the language, I further propose a Word Formation Rule, incorporated with a
Foot Formation Rule based on the recent theories of prosodic morphology.
Section 2.6 discusses some theoretical implications and empirical consequences of the theory developed in this chapter. Section 2.7 provides a summary of this study.
2.1. Criteria for identifying compounds in classical
Chinese
Before we discuss compound words in Classical Chinese, we must first answer
the question of what a compound word in Classical Chinese is. For example,
the combination of two words tia-n-zı̆ 天子 ‘The Son of the Heaven, Emperor’
in Classical Chinese is generally considered a compound, while Ju-n-chén 君臣
‘monarch and official’ is not.2 What is the difference between these two? Are
they differentiated syntactically, morphologically or semantically? Obviously,
we need a set of criteria to identify what can be called a compound in Classical
Chinese.
However, the problem with criteria proposed to date is that they are not
entirely satisfactory for use with Classical Chinese compounds. For example, let
us look at the criteria given by Chao (1968):3
1. a.
b.
c.
d.
e.
Part of the item is neutral-toned.
Part of the item is a bound form.
The parts are inseparable from each other.
The internal structure is exocentric.
The meaning of the whole is not compositional of its parts.
If a combination of two morphemes meets one of these criteria, according to
Chao, it is considered a compound in Modern Chinese.
For Classical Chinese, the “neutral-tone” test is not valid simply because
Classical Chinese is an extinct literary language. Therefore, we do not know
whether any part of the two combined forms is neutral-toned or not. Therefore, criterion (1a) cannot be considered as a criterion for Classical Chinese
compounds.
According to (1b), if part of the item is a bound form, this item is a compound. However, it is well known that morphemes in Classical Chinese are
16 Prosodic word in Classical Chinese
nearly always free forms, since each part of a compound can be used independently. For instance:
2. a. 小人少而君子多 . . .
Xiăorén shăo ér ju-nzi duo- . . .
villain few but gentleman more
‘If there are fewer villains but more
gentlemen. . .
國家久安。
《韓非子·安危》
guó-jia- jiŭ a-n.
country-family long save
the country will be safe forever.’
(Hanfeizi. Anwei)
b. 晉國貳則子之家壞。
《左傳·襄公二十四年》
Jìn guó èr zé zı̆ zhı- jia- huài.
Jin country two then you ’s family break
‘If Jin country is broken up, your family will be destroyed as well.’
(Zuozhuan. Xiang. 24)
In these examples, although guó 國 ‘country’ and jia- 家 ‘family’ can form a
compound in (2a), they can also be used independently in other sentences as
in (2b). This shows that although sometimes two elements are closely knitted
together to be used as a compound, there is hardly any evidence to show that
one of the parts is a bound form in Classical Chinese.4 As a result, the criterion
(1b) would not work for Classical Chinese compounding either.
Let us consider the criterion (1c), that is, the inseparability of the parts from
each other in a compound. (3) is an example showing a fairly well-known
compound in Chu-nqiu- Fánlù (《春秋繁露》) by Dong Zhongshu (董仲舒
179–104 BC) (“prt” refers to “particle”, henceforth):
3. 天子，天之子也。
Tia-n-zı̆, tia-n zhı- zı̆ yě.
heaven-son, heaven’s son particle
‘An Emperor is the Son of Heaven.’
Tia-n-zı̆ 天子 ‘emperor’ is a compound but this does not mean that the two
parts cannot be separated. Since Classical compounds are usually composed
of free forms, even if the two forms are bound together to form a compound
under one circumstance, they may also be used as a phrase with two single
words separately in other contexts. In other words, the inseparability criterion
cannot apply without regard to specific contexts as examples (2a) and (2b) show.
Therefore, criterion (1c) may not be ideal for use with the Classical Chinese.
The two remaining criteria for determining compound-hood are (1d) and
(1e). These two criteria seem to work for identifying compounds in Classical
Chinese. For example:
4. a. 妻子
qı--zı̆
wife-children
Prosodic word in Classical Chinese
‘wife’
妻子好合《詩經·常棣》
qı--zı̆ hăo hé
wife good marry
‘good marriage with a nice wife’
17
(Shijing.Tangdi)
b. 動靜
dòng-jìng
active quiescent
‘activity’
察其動靜《漢書·金日磾傳》
chá qí dòng-jìng
scout his activity
‘to scout his activity’ (Hanshu. Jinmidizhuan)
c. 車馬
ju--mă
carriage horse
‘carriage’
大夫不得造車馬。
《禮記·玉藻》
dàfu- bù dé zào ju--mă.
Officialdom not can
make carriage
‘The officialdom cannot make carriages themselves.’ (Liji. Yuzao)
d. 司馬
sı--mă
charge military
‘General (a title in army)’
Let us consider the criterion of exocentricity indicated in (1d) first: the internal
structure (of a compound) is exocentric; that is, the syntactic form class of the
head of the compound is not the same as that of a phrase in which the compound occurs. In other words, syntactic phrase structure rules cannot apply to
the internal structure of a compound, which has been considered as a corollary
of the Lexical Integrity Hypothesis (LIH, Huang 1984). According to the criterion of exocentricity, the example given in (4d) must be a compound, because
the verb sı- (to control) cannot serve as a head of phrase when sı--mă is used as a
compound (since sı--mă is a noun). However, tia-n-zı̆ ‘Emperor’, as we have seen
before, should be considered a compound, since it has become a proper noun.
Yet, in (3) tia-n zhı- zı̆ yě ‘The Son of Heaven’, the zhı-, a possessive marker in
Classical Chinese can be inserted into it, which means that a phrasal rule can
actually apply to it. Is tia-n-zı̆ a compound? By (1d) it should not be, but in fact
it is. Obviously, (1d) is not a sufficient criterion.
Consider next (1e), i.e., the criterion of semantic noncompositionality. This
criterion can be rendered as the following equation (“‖ . . . ‖” indicates the
meaning of “. . . . . .”.):
5. ‖AB‖ ≠ a+b
18 Prosodic word in Classical Chinese
Let AB be a combination of two forms A and B, and let the meaning of A be
“a” and that of B be “b”. If the meaning of AB is compositional, i.e., “a+b”,
then AB must be a phrase, rather than a compound, given the criterion that
the meaning of the whole is not merely a composition of its parts. On the
other hand, if the meaning of AB is not “a+b”, we will have the following
possibilities:
6. a. ‖AB‖ = a (left part of AB)
b. ‖AB‖ = b (right part of AB)
c. ‖AB‖ = c (other)
Accordingly, if a combination of two forms meets one of the three possibilities in (6), it will be considered as a compound. Based on the extended formula
given in (6), examples described in (4a–4c) must all be analyzed as compound
words. This is because in all of these examples, the meaning of the whole (i.e.,
AB) is not simply a composition of its parts (i.e., AB≠a+b).
While the semantic criterion seems to work for identifying compounds
in Classical Chinese, it is not perfect. For example, in (4c), ju--mă (carriage­
horse) meets the condition of the semantic criterion: ‖AB‖= a. That is,
ju--mă means only “carriage”, and another part of the combination mă (horse)
has no semantic value at all, hence it is considered as a compound. (4c) represents a special type of compound traditionally known as “pia-nyì fùcí”
(偏義副詞) – a combination using only one meaning of the two.5 At first
glance, this type of combination would make perfect sense to be identified
as a compound, because if one part of the combination has no meaning, the
combination would be more like a word, rather than a phrase. However, the
problem with this treatment is that, without the sentence given in (4c), ju--mă
will not mean “carriage” but “carriages and horses”; that is, the meaning of
“carriage” in ju--mă is totally dependent on the verb zào (to build/make), and
there is no evidence to show that ju--mă (carriage) has been used anywhere else.
If ju--mă does not occur freely as a compound, it is difficult to consider it as an
independent lexicon entry.
There is an additional problem. If we treat ju--mă as a compound, what
is the function of mă in ju--mă? Although the semantic criterion has identified ju--mă to be a compound, it creates a problem for further analysis of
the internal structure of the compound. If ju--mă is formed by a syntactic
coordination rule, that is, the structure of ju--mă is syntactically “carriage and
horse”, how do we explain the fact that half of the structure has no semantic
value? As we know that mă is a noun proper, and is not a functional element
or a suffix, if ju--mă is a compound, how can mă be ignored totally within
the structure?
As we have seen above, none of these five criteria would work completely
for Classical Chinese compounds. However, each of them, except for the neutralization of tones, works to a certain degree for certain types of compounds.
Prosodic word in Classical Chinese
19
For example, compounds created by what is known as the reduplication process
(Dobson 1959) are to be easily identified by criteria given in (1):
7. 匍匐
pú-fú
crawl
‘to creep, to crawl, to toddle’
赤子匍匐將入井。
《孟子·滕文公上》
chìzı̆ pú-fú jia-ng rù jı̆ng.
baby crawling will enter well
‘A baby crawling is about to fall into a well.’
(Mencius. tengwen gong shang)
It has been observed (Dobson 1959) that compounds which are derived
by reduplication may have the meaning “actions or states in a repetitive pattern, succeeding each other”. Obviously, this type of compound can easily be
identified by either (1b) “part of the item is a bound form” or (1c) “the parts
are inseparable from each other” or even (1e) “the meaning of the whole is not
compositional of its parts”. However, the easiest cases, such as reduplicatives,
are in the minority, while the most difficult cases, those that have been called
syntactic words (Chao 1968),6 are in the majority, such as the examples given
in (2a) and (4).The following statistics (taken from Cheng 1981) show the proportion between these two categories (“Der” refers to Derivative compounds
and “SynW” refers to syntactic words):
Table 2.1 Proportion between derivative and syntactic compounds in Confucius
and Mencius
Chronology
Texts
Total
Der
%
SynW
%
c.550 BC
c.300 BC
Confucius
Mencius
180
333
24
44
13.3
13.2
138
249
76.7
74.8
There are only 13.3% derivatives in Confucius, and 13.2% in Mencius, but
76.7% syntactic words in Confucius and 74.8% in Mencius. If a criterion can
only handle 13% of the data in the language, it should not be considered valid.
If we consider the development of compound formation through time, what
we can see from Cheng’s statistical data is that by Han times (c. 100 AD) the
derivatives have decreased to only 8.22% among all the compounds as shown
in the following table:
Table 2.2 Proportion between derivative and syntactic compounds in Lunheng
(c. 100 AD)
Confucius
Total
Total
%
SynW
%
462
38
8.22
424
91.78
20 Prosodic word in Classical Chinese
Given that 91.78% of the compounds in the language are “syntactic”, we
conclude that, in practice, the most effective criterion for identifying compounds in Classical Chinese is the semantic one, that is, the one given by Chao
in (1e), formulated in (6) and modified here as (8):
8. Semantic Criterion:
If A and B are two independent forms, and the semantic interpretation of A is
“a” and that of B is “b”; and if in context X, either
a. ‖AB‖= a (left part of AB)
b. ‖AB‖= b (right part of AB)
c. ‖AB‖= c (other)7
then the combination of AB is a compound in context X.
In what follows we will adopt the semantic principle as a working criterion to embark upon the following study of Classical Chinese compounds.
However, a more theoretical and formal constraint, i.e., the Word Formation Rule (WFR), and the notion of Idiomatized Prosodic Word defined
by the Foot Formation Rule, as developed in section six, will be taken
to characterize the idiomatic property of compound words in Classical
Chinese.
2.2. Compounding in Zhao Qi’s Mencius Zhangju
In order to examine the development of Classical Chinese compounds, I have
compared Mencius (c. 372–289 BC) with Zhao Qi’s commentary on Mencius, i.e.,
the Mencius Zhangju (c. 200 AD). The reasons for selecting Zhao Qi’s work as a
body of comparative data are the following: First, the Han dynasty (206 BC to 220
AD) in which Zhao Qi lived (107–201 AD), was an important transition period
from Old Chinese (c. 1000 B.C., i.e., the Shijing [The Book of Poetry] period)
to Medieval Chinese (seventh century AD, i.e., the Qieyun [rhyme dictionary]
period). It is well known that from Old Chinese to Medieval Chinese, the language changed typologically with respect not only to its phonology and morphology, but also to its syntax (Chou 1962;Wang 1980; Mei 1980; Norman 1988;
Huang 2005; Xu 2006, Feng 2014; and many others).Therefore, Zhao Qi’s work is
an ideal one to look at the development of Classical Chinese compounds. Second,
we can deduce that the a language used by Zhao Qi is close to the vernacular of
that time.This can be seen from Zhao Qi’s preface to the Mencius Zhangju, which
I translate as follows:
. . . When I took refuge in Hăidài (i.e., Shandong province), I had nothing
to do but read classical books. Often, I gained new insights from reviewing
classics. During this period, a noble man (i.e., Sun Song) admired my hard
work and old age. He often came to me and discussed classical texts with
interpretations of those texts. . . . Under these circumstances, I narrated
what I know, and wrote this book . . .
(Preface to Mencius Zhangju)
Prosodic word in Classical Chinese
21
From this, we know that (i) Mencius Zhangju was written during the special time
that Zhao Qi had discussions with (or probably gave lectures to) Sun Song, and
that (ii) the language used in Mencius Zhangju was based on those discussions or
lectures.Thus, we might conclude that Mencius Zhangju is closer to the Han vernacular than most other documents found in this period (see Harbsmair 2013).
Third and most importantly, in Mencius Zhangju, probably because it is close
to the vernacular language, Zhao Qi often used two-character combinations
to interpret one-character words in Mencius. I will call this the “one-to-two”
interpretation in the following discussions. The “one-to-two” annotations
allow us to determine when one character has been replaced by two between
the Warring States and the Han periods (300 BC – 200 AD)7.
The procedures of the investigation for compounding of Classical Chinese
in Mencius Zhangju are as follows. First, all the tokens are listed, which consist
of two characters in Zhao Qi’s work but are one-character words in Mencius.
For example:
9. Mencius: 聖人且有過。
Shèngrén qiě yŏu guò.
sage-person also have mistake
‘Even sages make mistakes.’
Zhao Qi: 聖人且有謬誤。
shèngrén qiě yŏu miù-wù.
sage-person also have false-mistake
‘Even sages make mistakes.’
In Mencius, the one-character monosyllabic word guò was used for the concept
“mistake”. In Zhao Qi’s exegesis, the two characters miù and wù are combined
to gloss the one character guò. In addition to all of the instances of one-to-two
translations, annotational materials which contain two-character combinations
in Zhao Qi are also listed. For example:
10. Mencius: 棺槨無度。
gua-n
guŏ
wú dù.
inner-coffin outer-coffin no rule
‘The inner and outer coffins have no rules.’
Zhao Qi: 棺槨厚薄無尺寸之度。
gua-n guŏ hòu-báo wú chı̆-cùn zhı- dù.
inner-coffin outer-coffin thin-thick no meter-inch’s rule
‘The thickness of inner and outer coffins have no rules for
their size.’
In (10) there are three combinations in Zhao Qi: gua-n-guŏ, which is repeated from
Mencius, and hòu-báo and chı̆-cun, which have no corresponding words in Mencius,
but are used by Zhao Qi. We will call this type of annotation “none-to-two”.
Although this type of data is not a word-to-word annotation like the ones given
22 Prosodic word in Classical Chinese
above, nevertheless they are annotations of meanings implied in that sentence.
These two-character combinations provide us an opportunity to see how meanings are expressed by the two-character combinations in the Han period language.
Therefore, such examples are also included in my percentage study of this section.
As indicated in (10), we also take into consideration the two-character combinations that Zhao Qi repeated from the Mencius, such as gua-n-guŏ. This type
of annotation will be named “two-to-two”.
11. Mencius: gua-n guŏ wú dù.
inner-coffin outer-coffin no rule
‘The inner and outer coffins have no rules.’
Zhao Qi: gua-n guŏ hòu-báo wú chı̆-cùn zhı- dù.
inner-coffin outer-coffin thin-thick no meter-inch’s rule
‘The thickness of inner and outer coffins has no rules for their
size.’
Thus we have three types of combinations that we will examine in this study.
These are: (a) combinations used to gloss a monosyllabic word (l-to-2), (b)
combinations used to explain the meanings or implications of the sentences
(0-to-2) and (c) combinations repeated from the original text (2-to-2). Putting
all these combined forms together, we then evaluate them according to the
semantic criteria for compounding given in the previous section. Since the use
of two characters by Zhao Qi to gloss the one character given in Mencius provides an excellent illustration for the study of the development of compounding, we are able to see where and how a monosyllabic word was replaced by a
disyllabic compound. The questions we seek to answer are:
How many two-syllable combinations used by Zhao Qi can be identified
as compounds?
ii) How many one-character words in Mencius have been glossed by compounds in Zhao Qi’s annotation?
iii) How many compounds have been used by Zhao Qi in his explanations of
meanings and ideas within sentences?
iv) How many compounds used by Zhao Qi have survived in present-day
Mandarin Chinese?
i)
As we can see from Table 2.3, there are a total of 169 two-character combinations in my data: in the Liang Huiwang Shang section of the Mencius Zhangju,
there are 113 tokens; in the Gongsun Chou Xia section of Mencius Zhangju, there
are 56. Among these 169 cases, there are 73 cases that belong to the “one-totwo” category, 60 cases are “none-to­two” and 36 cases are “two-to-two”.
From the data given in Table 2.3, we can see that 73 monosyllabic words
in Mencius have been replaced by two-syllable combinations in Zhao Qi, and
among the 73 two-syllable combinations used by Zhao Qi, 47% of them are
compounds. In addition to the replacement of one by two, there are 60 cases of
“none-to-two”. Among these 60 cases, 65% are compounds. From these data we
Prosodic word in Classical Chinese
23
may conclude the following: first, an ever-greater number of compound words
were formed during this period.This can be seen clearly from Table 2.3. Among
all of the 169 cases, only 21% of the tokens were disyllabic combinations in the
Warring States Period, while 79% of them occurred in the Han dynasties.
Table 2.3 Combinations of two characters in Zhao Qi’s Mengzi Zhangju and Mencius
Total
%
Han
%
Compound
1‑to‑2
0‑to‑2
2‑to‑2
Total
73
60
36
169
43%
36%
21%
100%
34
39
29
102
Modern
%
Compound
47%
65%
80%
60%
31
25
18
74
42%
42%
50%
44%
Although the use of compounds can be traced back to the Shang dynasty
(sixth to eleventh centuries BC, see Cheng 1981) and a further development
can be found during the Warring States Period, it is evident that a sharp increase
in compounding occurred during the Han periods. The 73 one-to-two cases
show that 43% of the time the Han people used disyllabic forms, whereas the
people who lived in the Warring States Period used monosyllabic forms in the
same linguistic contexts. And this is even clear if we look at only the compounds words in Mencius and Zhao Qi’s Mengzi Zhangju:
Table 2.4 Compounds in Mencius and Zhao Qi’s Mengzi Zhangju
1-to-2
2-to-2
Total
Mencius
Zhao Qi
0
29
29
34
29
63
The total number of compounds used by Zhao Qi are 63. Among the 63
cases, 29 of them are shared by Zhao Qi and Mencius. The other 34 cases are
monosyllabic forms in Mencius but compounds in Zhao Qi’s texts. That is, in
the same linguistic environment, Zhao Qi added 34 more compounds onto the
29 compounds used by Mencius.This shows that the compounds used by Zhao
Qi has increased about 54% comparing with the number of compounds used
by Mencius, i.e., the number of compound words has doubled during the Han
dynasties.This strongly suggests that a sharp increase in compounding occurred
during the Han dynasties.
All of these pieces of evidence suggest that the people of the Han periods use
more two-syllable combinations or compounds to express the same concepts
which are expressed through using monosyllabic words during the Warring
States Period.
Secondly, the data also suggest that the development of compounds correlates with the appearance of disyllabic combinations. There are 169 disyllabic
24 Prosodic word in Classical Chinese
combinations in Zhao Qi, and by the semantic criterion, only 43% of compounds have appeared. The fact that there are more two­syllable combinations
appearing in the language but that fewer compounds can be identified indicates
that the appearance of two-syllable combinations may be the fundamental basis
of the development of compounds in Classical Chinese.
The data from Zhao Qi comport with the general observation that Classical
Chinese compounds are structurally formed using rules from syntax. The following syntactic relations between the two parts of compounds are observed
in our data:
12. Coordinative Compounds
a． NN
尺寸 chı̆-cùn
衣食 yı--shí
b.
c.
meter-inch, ‘size’
cloth-food, ‘daily use’
VV
忖度 cŭn-duó
贈饋 zèng-kuì
think-measure, ‘ponder’
send-give, ‘make a present of ’
AA
險阻 xiăn-zŭ
純粹 chún-cuì
dangerous-blocking, ‘difficult’
pure-best, ‘unadulterated’
Subordinative Compounds
d. AN
寡人 guă-rén
e.
NN
國人
single-person, ‘I’
Emperor
guó-rén country-person, ‘aristocrat’
There are no S(ubject)-P(redicate), V(erb)-R (esultative complement) and
V(erb)-O(bject) compounds found in the data presented here, indicating that
coordinative and subordinative relations are the most favored structures for
compound formations, and that VR-structures,VO-structures and SP-structures
are disfavored structures.
The comparison between Mencius and Zhao Qi shows a number of
generalizations:
i) an ever-greater number of compounds developed during the Han dynasty;
ii) the development of compounds is based on the development of disyllabic
combinations;
iii) compounds must be formed structurally from syntax; and
iv) coordination and subordination are favored structures for compounding while Verb-Object, Subject-Predicate and Verb-Resultative are
disfavored.
Prosodic word in Classical Chinese
25
All these facts about Classical Chinese compounds call for a theory of morphology to explain why they exhibit such properties during the course of their
development.
2.3. Questions regarding the development of compounds
This section deals with the questions raised from the study of Zhao Qi’s data
given in last section and also from the studies of compounds in general. First, if,
as indicated in Zhao Qi’s data, compounds are derived from two-syllable combinations or phrases, then why are coordinative and subordinative compounds
very common, but Verb-Object, Verb-Resultative and Subject-Predicate compounds extremely rare?
Second, if coordinative structures such as căo-mù 草木 ‘grass-tree’ and lín-mù
林木 ‘woods-tree’ can develop into compounds by specializations of meaning
(‘grass and trees’ → ‘vegetation’,‘woods and trees’ → ‘woods’, respectively), then
why do we not find three-character coordinative structures such as căo-mù-shù
草木樹 ‘grass-woods-tree’ as a result of the same process: ‘grass, woods, and
trees’ → ‘vegetation’?
Third, if syntax determines the internal structure of compounds, then coordinative compounds such as dòng-jìng (active-quiescent, ‘activity’), ju--mă
(carriage-horse, ‘carriage’) must be structurally interpreted as ‘active and quiescent’ and ‘carriages and horses’, respectively. However, the semantics of these
compounds does not allow us to give a full interpretation of the meanings
conveyed by each part of the compounds in these structures: ju--mă is not interpreted as ‘carriage and horse’, but as ‘carriage’; dòng-jìng is not interpreted as
‘active and quiescent’, but as ‘active’. The interpretation requires the other part
of the compound to be semantically empty, and the syntax of such coordinating
structures must thus be interpreted as: ‘dòng and Ø’, ‘ju- and Ø’. How can a syntactic rule allow a coordinative structure with the second part semantically empty?
If the second part of a coordinative structure has no semantic value, what does
‘coordination’ mean structurally, and semantically?
Fourth, why there is an ever-greater number of compounds produced specifically around the period of the Han dynasties? Why does the Chinese language
suddenly have such a strong tendency toward the formation of compounds?
Finally, if both coordinative and subordinative structures are the most productive types of compounding, why at the beginning of their development (The
Spring and Autumn Period, c. 550 BC) were there more subordinative compounds than coordinated words (see Cheng, 1981)? Also, why after the Warring
States Period (c. 221 BC) did coordinative compounds become more and more
dominant while the number of subordinative compounds declined (Cheng 1981)?
As a response to these questions, it is argued that the sharp increase in the
number of compound words is a consequence of the development of disyllabic
feet resulting from syllable-structure simplification that occurred from Old Chinese to Medieval Chinese and activated a new system of prosodic morphology.
26 Prosodic word in Classical Chinese
In what follows, we will first review some previous accounts for the development of compounds in section 2.3, and then propose that the development of
disyllabicity is independent of compounding in section 2.4. Section 2.5 shows
how the syllable-structure simplification resulted in disyllabic feet. Some theoretical and empirical implications will be discussed in section 2.6.
2.3.1. Loss of phonological contrast
There have been a number of hypotheses to explain the development of compounding in Chinese. To date, the answers that have been provided are mostly
function-oriented.
Norman (1988), for example, has suggested that it was “. . . chiefly due to
phonological attrition, which greatly decreased the number of phonologically distinct syllables in the language” (1988:86). If phonologically distinct
syllables were merged into phonologically non-distinct syllables, it would
result in a great increase in the number of homophones in the language. It
seems quite reasonable to assume that the increase in the number of compounds around the Han dynasty is a result of the phonological changes in
the language.
Let us first consider the argument that compounding was caused by phonological attrition. Among the facts known about phonological attrition from
Old Chinese to Medieval Chinese, two changes have been posited in SinoTibetan studies: consonant cluster simplification and the loss of morphological
affixation.
Haudricourt (1954 [1972]) proposed that the departing tone in Medieval
Chinese originated from a suffix *-s in Old Chinese, a hypothesis that has been
widely accepted in the literature (Mei 1994; Baxter 1992; Zhengzhang 2003,
Pan 2000, and many others). Following this hypothesis, all departing tones of
Medieval Chinese originally ended in *-s in Old Chinese, from which we may
infer a final consonant cluster in CVC8 roots: *CVC-s. These clusters were lost
in the transition to Medieval Chinese.
Not only does the suffix *-s allow us to reconstruct final consonant clusters,
but sets of characters which shared a common phonetic element (Xiéshe-ng
諧聲) also lead to the reconstruction of initial consonant clusters in Old Chinese. For example, a cluster *sm- has been reconstructed for Old Chinese in
examples such as the following (see Baxter, 1992:175):
13. Modern Chinese
喪 sa-ng
亡 wáng
Medieval Chinese Old Chinese
< sang
< *sm-ang ‘mourning, burial’
< mjang
< *m-jang ‘not have, not exist, die’
It has also been posited (Benedict 1972; Bodman 1980; Mei 1994 and
many others), that Old Chinese and Old Tibetan had a prefix *s- which had a
Prosodic word in Classical Chinese
27
causative or denominative function in Proto-Sino-Tibetan as seen in example
(14b) (taken from Mei 1994):
14. 林 lín < ljem
森 se-n < sjem
< *rjem
< *s-rjem
N:
SV:
‘forest, woods’
‘woodsy, well-wooded’
Mei (1994) posits that lín 林 and se-n 森 in Old Chinese differ by the presence or absence of the *s- prefix. Lín 林 ‘woods’ was a noun whereas se-n 森
was ‘woodsy’, i.e., a stative verb. The function of *s- in the lín /se-n pair is
denominative, turning a noun into something other than a noun, in this case,
a stative verb. If this is so, the prefix *s- would create another type of cluster
in every word: *s-CVC. However, the *s- also did not survive the transition
to MC.
This can also be seen from a comparison between the final consonants of
Old Chinese and those of Medieval Chinese. Based on Li’s reconstruction of
Old Chinese (1980:33), a great change in syllable structure occurs from Old
Chinese to Medieval Chinese (Li 1980:8) shown as follows:
15. Old Chinese Final Consonants
Tone
Level Tone
Rising Tone
Departing Tone
Nasal
-m -n -ng -ngw
-mx -nx -ngx -ng
-mh -nh -ngh -ngwh
Stop
(-b) -d -g -gw
vx (-bx) -dx -gx -gwx
-bh -dh -gh -hwh
16. Medieval Chinese Final Consonants
Nasal
-m -n -ng
Stop
-p -t -k
According to the inventory of final consonants reconstructed by Li, there
were no open syllables in Old Chinese.9
There is no doubt, according to reconstruction of Proto-Chinese phonology
(Ting 1979; Li 1980;Yu 1981; Baxter 1992; and many others), that the change
from Old Chinese to Medieval Chinese resulted in a simplification in syllable
structure.10 This is true if we assume (i) the simplification of consonant clusters,
(ii) the loss of morphological affixation and (iii) the reduction of the inventory
of final consonants. Nevertheless, before the consonant clusters were simplified
and affixes such as *-s were lost, the possible syllable structure of Old Chinese
was *CCVCC. However, in Medieval Chinese the syllable structure changed to
CV(-C) where -C consisted of only -m, -n, -ng, -p, -t, -k as final consonants. In
other words, the CCVCC syllable structure simplified to CV(C). What is crucial to observe here is that consonant clusters were no longer permitted either
before or after the main vowel in Medieval Chinese.
28 Prosodic word in Classical Chinese
As a result of consonant-cluster simplification, the number of phonologically
distinct syllables in the language decreased dramatically.11 If both morphological
affixes were lost, and the number of phonologically distinct syllables decreased,
a consequence of this change would be a great increase in homophones, inevitably increasing the functional load of syllables in the language.
Given the simplification of syllable structure and the decrease in the number
of phonologically distinct syllables, it is reasonable to expect the language to
develop other means, e.g., compounding, to reduce the functional overload on
the syllable. This is the ‘functional explanation’ for the development of compounding in Chinese.
Note that although the functional approach seems a reasonable explanation
for compound formation, there are difficulties with this account. The functional approach is based primarily on the assumption that the information carried by the merged syllables results in a functional overload. However, there
were some newly developed phonological elements that could have supplanted
the loss of contrast. For example, the departing tone replaced the *-s (see Baxter 1992:135), and the loss of the final *-ʔ (a glottal stop, Baxter 1992:320) is
thought to have been replaced by the high-rising tone (Pulleyblank 1962:225–
227; Mei 1970). Given these developments and also the general hypothesis that
“the tones of Middle Chinese are developed from Old Chinese codas and postcodas”12 (Baxter 1992:7), it is clear that the functional load was at least partly
reduced by remaining contrasts.13 Given this fact, the functional approach loses
some of its appeal as an explanation for the rapid expansion of compounding
(see also Labov 1987).
2.3.2. Semantic disambiguation of monosyllables
It is possible that, during the later development of compounding (the Han
dynasty), one may find some supporting evidence for the functional explanation from examples such as given in (17).
17. Mencius:
Zhao Qi:
(王良)天下之賤工也。
(Wáng Liáng) tia-nxià zhı- jiàn go-ng yě.
Wang Liang world’s lousy artisan particle.
‘Wang Liang is the lousiest artisan in the whole world.’
(王良)天下鄙賤之工師也。
(Wáng Liáng) tia-nxià bı̆-jiàn zhı- go-ng-shı- yě.
Wang Liang world clumsy-lousy’s artisan-artisan prt.
‘Wang Liang is the lousiest artisan in the whole world.
In (17), Zhao Qi uses two characters bı̆ and jiàn to annotate the single character:
jiàn. Both bı̆ and jiàn had the meaning “workers who lack skill”, but jiàn could
also mean “cheap, lowly, underestimated”, etc. As Jiao Xun (1763–1820 AD, see
Mencius Zhushu) has pointed out: Wang Liang was by no means a lowly (jiàn)
person, because he was a senior official. Jiàn in (17) means only “an artisan who
lacks skill”; this is why Zhao Qi uses bı̆ with jiàn in order to disambiguate jiàn.
Prosodic word in Classical Chinese
29
When jiàn is used alone, its meaning can be vague. Therefore, jiàn is combined
with bı̆ to select the meaning of jiàn ‘lacking skill’. This kind of semantic disambiguation was undoubtedly a contributing factor in the development of
compounding.
However, the disambiguating function of compounding does not necessarily
lead to a conclusion that disambiguation is the key factor in the development
of compounding. In fact, empirical evidence argues against such a conclusion. If
disambiguation of words via compounding is the major source of compounds
as suggested by the functional account, then we would expect coordinate compounds such as bı̆-jiàn to be in the majority, especially at the beginning of their
development. This is because unlike other types of compounds, (for example
the Modifier Head compounds tia-n-zı̆, ‘Heaven’s son, Emperor’, which lacks
the function to disambiguate a monosyllabic word), coordinate compounds
can be readily formed in order to fulfill the disambiguative function discussed
above. Therefore, they would be a trigger for compounding according to the
functional account. However, statistical data given by Cheng (1981) show that
coordinate compounds were in the minority as shown in Table 2.4 (“Total
Comp” stands for Total compound words, “CC” stands for Coordinating compound words; “MH” stands for Modifier Head compound words):
Table 2.4 Percentage of CC and MH compounds in Confucius (c. 550 BC)
Total
Total
Total
Comp
CC
%
MH
%
180
48
26.7
67
37.2
Some 37.6% of compounds were MH compounds but only 26.7% were
CC compounds in Confucius’ Analects. One may still argue that the numerous
MH compounds may also result from disambiguation such as plant names that
are combined with a word indicating its category: pine-TREE, willow-TREE,
oak-TREE, etc. However, if this is so, why were there more CC compounds
than MH later on as shown in Table 2.4? A function-based argument cannot
win on both cases. The fact that there were fewer CC compounds but more
MH compounds argues against the functional solution and calls for a different
explanation.
Another problem with the functional explanation is the development of
counter-functional compounds during this period. By “counter-functional
compounds” I mean that the meaning of a compound cannot be inferred from
the meanings of the individual parts, and it must be learned independently. For
example:
18. a. 動靜
dòng-jìng
active quiescent
‘activity’
30 Prosodic word in Classical Chinese
察其動靜《漢書·金日磾傳》
chá qí
dòng-jìng
scout
his activity
‘to scout his activity’
(Hanshu. Jinmidizhuan)
b. 車馬
ju--mă
carriage horse
‘carriage’
大夫不得造車馬。
《禮記·玉藻》
dàfu- bù dé zào ju--mă.
officialdom not can make carriage
‘The officialdom cannot make carriages themselves.’
c. 市朝
shì-cháo
market-(imperial)court
‘market’
肆諸市朝《論語·憲問》
sì zhu- shì-cháo14
kill him-at market
‘to kill him at the market’
(Liji. Yuzao)
(Confucius. Xianwen)
Under the functional account, the combination of two forms to form a word
must be semantically distinctive and must functionally contribute to the meaning. In most disyllabic combinations, the meanings of the two forms can either
be consistent or cooperative. For example, in zhàn-dòu 戰鬥 ‘war-tussle, fight’,
the meaning of zhàn ‘war’ and dòu ‘tussle’ are consistent, and the combination
zhàn-dòu is more distinctive than either zhàn or dòu. In tia-n-zı̆ 天子‘Heavenson, Emperor’, shì-fe-i 是非 ‘right-wrong, right and wrong’ and yı--.cháng 衣裳
‘shirt-skirt, clothes’, the meanings of the two forms are cooperative, i.e., the
concept of ‘Emperor’ is realized by participation of two meanings ‘heaven’ and
‘son’; the expression ‘right and wrong’ is realized by combination of the meaning ‘right’ with ‘wrong’; and the concept ‘clothes’ is achieved by a process of
abstraction from two meanings ‘shirt’ and ‘skirt’. All these examples show that
in order to make a combination functionally contributive and semantically
distinctive, each of the two forms in the combination (syntactic compound)
must have an independent semantic value. In other words, the meaning of
each compound includes, at least to some extent, the meanings of the individual parts. If the meaning of one part of a combination is originally ‘zero’, it
makes no contribution to the meaning of the compound under the functional
approach.
However, what we found is what is not expected by the functional account
in the cases given in (18). As we have seen in (4), this type of compound is traditionally known as “pia-nyì fùcí 偏義複詞” – combinations using only one meaning of the two. In other words, the meaning of the other part of the compound
Prosodic word in Classical Chinese
31
must be “zero”, or suppressed, in these combinations. It can be seen clearly
from the examples above. Ju--mă 車馬 means only ju- (carriage); chéng-bài 成敗
‘success-failure’ conveys only “failure” (see endnote 14).15 The shì-cháo 市朝
case (18c) is even more convincing: it is a rule in old times that killing must be
done at a market, not at the imperial court. It is clear from historical context
that the combined form shì-cha-o must be interpreted as shì (the market), rather
than shì-cha-o ‘the market and the imperial court’.
Notice that, the ju--mă–type examples are different from examples such as
shì-fe-i (right-wrong, ‘right and wrong’, Zhuangzi c. 300 BC) and yı-.cháng ‘shirtskirt, clothes’, (Shı-jı-ng 1000 BC). The semantic interpretation for ju--mă types
is this:
19.
Combined forms
Surface Meaning
Actual Meaning
A+B
a b
a
ju--mă
carriage-horse
carriage
The semantic interpretation for shì-fe-i is this:
20.
Combined forms
Surface Meaning
Actual Meaning
A+B
a
b
a
b
shì-fe-i
right-wrong
right and wrong
The semantic interpretation for yı-.shang is this:
21.
Combined forms
Surface Meaning
Actual Meaning
A +B
a b
c
yı-.cháng
shirt-skirt
clothes
As we can see from the above illustrations, the actual meaning of shì-fe-i is
the same as its surface meaning, and the actual meaning of yı-.cháng is different from its surface meaning. Yet, the actual meaning of ju--mă only takes one
part of its surface meaning: carriages. Functionally speaking, there would be
no confusion in shì-fe-i and yı-.cháng, because their meanings can be identified
or inferred from their surface meanings. However, combinations of the ju--mă
type are very different. Since no abstract meaning is available for ju--mă, users
of the language must rely on surface meaning to determine the communicative function. Yet, the surface meaning of ju--mă (carriage and horse) contains
an element that is not what the speaker has actually meant (viz., horse) in
(18b). Therefore, how are users of the language to infer that part of ju--mă is
different from what the speaker actually meant? This is why many traditional
philologists particularly named them the pia-nyì fùcí ‘combinations using one
meaning of the two’.l5
Obviously, the development of this type of compound is not explained by a
functional account, because they increase rather than eliminate the functional
32 Prosodic word in Classical Chinese
load in communication. If, as the functional account presumes, the development of compounds is caused by the need to reduce confusion created by phonological attrition, there was no reason for this type of compound to appear in
the language. The existence of this type of compound shows that two-syllable
combinations were highly preferred in the language, regardless of whether the
combinations created more functional load or not. In other words, the pressure
to create two­syllable forms overrode the communicative function.
These examples not only argue against a purely functional explanation,
but also raise the following question: why could Classical Chinese, during the
period of phonological change, tolerate a linguistic process which might cause
communication problems? I suggest that the reason is phonological, namely, the
advent of a two-syllable unit. This possibility is explored below in Section 2.4.
2.3.3. Vocabulary expansion
Cheng (1981), and many others, has suggested that the developing complexity of society required a greater number of vocabulary items and that social
requirements motivated the increase in the number of compounds. Of course,
by the Han dynasties, China had just been united and become a huge country.
Long-term peace encouraged trade, and contacts with foreign countries, especially with India, became much more frequent.The influence of Buddhism also
began to be felt in daily life. It was true that the society was much more complicated and there must have been some pressure to develop a lingua franca for the
newly formed country (see, Zhu 1992). However, if this is true, the linguistic
question that remains to be answered is why derivational morphological affixes
such as *-s and *s- shown in 4.1, could be lost in Old Chinese in the first place.
These losses reduce the size of lexicon. Furthermore, if compounding were
just caused by social development, why were compound words formed overwhelmingly by two, rather than by three or more syllables? More importantly,
we know that there are various ways for morphological processes to reduce the
overload on syllables, for example, by creating polysyllabic words or producing
new words by affixation, etc. If new vocabulary items were needed, why were
the new forms not created mostly by a morphological process of affixation and
why did compounding become the most important device in Classical Chinese
morphology? Social explanations do not account for the structural mechanism
of this language change, and the direction of the morphological development
of compounding in Classical Chinese remains a mystery according to the functional account.
2.3.4. Aesthetic factors
Cheng (1981) and Hong (1999:160–172) have also argued that the simplification of the Old Chinese phonological system was caused by the development
of disyllabic compounds, and that the development of compounds is due to
extra-linguistic factors. In Chinese tradition, people conceptually prefer a pair
Prosodic word in Classical Chinese
33
of two things, therefore the paired-syllable words (compounds) developed. This
solution is theoretically unattractive, and empirically problematic. If this is true,
for example, it remains to be explained why the Chinese could develop a fivesyllable pattern poetry around the late Han dynasty, and a seven-syllable poetry
before the Tang dynasty, instead of keeping the perfectly symmetrical foursyllable pattern poetry of Old Chinese.16
2.4. The development of disyllabicity
2.4.1. Chronology
It is well known that two-syllable combinations can be found in the earliest documents such as Shangshu (Archaic History, c. 1000 BC) and Shi-jing (The Book
of Odes, c. 1000 BC; see Cheng 1981). However, if we compare the Analects of
Confucius (551–479 BC, the Spring and Autumn Period) and Mencius (372–289
BC, the Warring States Period) with Lunheng (written by Wang Chong of the
Han dynasty, 27–97 AD), we clearly see a sharp increase in two-syllable combinations. For example, Table 2.5 (taken from Cheng 1985) shows that:
Table 2.5 Compounds in Confucius, Mencius and Lunheng
Chronology
Texts
No. of characters
Compounds
%
c. 550 BC
c. 100 AD
c. 300 BC
c. 100 AD
Confucius
Lunheng
Mencius
Lunheng
15,883
15,553
35,402
35,221
183
462
336
794
1.15
2.97
0.94
2.25
About 1% of the words in both Confucius and Mencius are compounds.
However, the 1% increased to nearly 3% in the Lunheng, suggesting that the
greatest increase in compounds occurred during the Han dynasty. Our study
comparing Mencius and Mencius Zhangju (Zhao Qi’s commentary on Mencius)
given in section 3 confirms these observations: of the total of 169 disyllabic
tokens used by Zhao Qi, 43% correspond to monosyllabic words used in Mencius, and 36% of them did not appear in Mencius at all. In other words, 79%
of the disyllabic combinations used by Zhao Qi of the Han dynasty to gloss
Mencius did not appear in Mencius. Furthermore, according to Xu De-an’s study
(1981), in the Classical Dictionaries Er Ya (c. 200 BC) and Fang Yan (c. 50 BC),
839 monosyllabic words were replaced by disyllabic forms by Guo Pu (276–
324 AD) of the Wei Jin Period (220–420 AD) in his Commentary on Er Ya and
Commentary on Fang Yan.
2.4.2. Textual evidence
Why did the two-syllable unit become preferred during the Han dynasty
(206 BC–220 AD)? The answer is partly given by the philologist Kong Yingda
34 Prosodic word in Classical Chinese
(574–648 AD) of the Tang dynasty in his Wu Jing Zhengyi (Commentary on the
Five Classics) (“NOM” refers to “nominalizer”,henceforth.):
22. a. 視民如禽獸。詩經·小雅·何草不黃·序》
Shì
mín
rú qín shòu.
See
people like bird-beast
‘Treat people like animal.’ (Shijing.Xiaoya.He cao bu huang.Xu)
Kong: 經言“虎”， “兕”及“狐”，只有獸耳，言“禽”以足句。
.
Jı-ng yán hŭ, sì jí hú， zhı̆ yŏu shòu ěr，yán qín yı̆ zú jù
classic say tiger, rhinoceros and fox, only have beast prt, say birds
for fulfill sentence.
‘The classical text only mentioned beasts: tiger, rhinoceros, and
fox, (Mao Heng) adds birds to explain the text in order to fulfill
the sentence.’
b. 芟夷我農功，
虔劉我邊陲。
《左傳·成公十三年》
Sha-n-yí wŏ nóng go-ng,
Qián-liú wŏ bia-n-chuí.
mow-weed my farm harvest, kill-slay my edge-frontier
‘(You) looted my farm harvest and killed my frontier people.’ (Zuozhuan.
gong.l3)
Kong: 重言“殺”者， 以圓文也。
Zhòng yán sha- zhě， yı̆ yuán wén yě
Repeat say kill Nominalizer. For round writing particle.
‘The reason of repeating “kill” is to perfect the language.’
c. 羔羊之皮《詩經·召南·羔羊》
ga-o yáng zhı- pí
lamb-sheep’s skin
‘The skin of lambs’ (Shijing. Shaonan. Gaoyang)
Kong: 兼言羊者，
以羔亦是羊， 故兩言以協句。
Jia-n yán yáng zhě， yı̆ ga-o yì shì yáng， gù liăng yán yı̆ xié jù.
Also say sheep prt. Because lamb also is sheep, so twice say for balance sentence.
‘The text says lamb with sheep, because lamb is also a kind of sheep,
in order to balance the sentence, it mentions both lamb and sheep.’
In the above examples, according to Kong, the reason for using two­-syllable
expressions is to zú-jù (to fulfill the sentence), to yuán-wén (to round off the sentence) or to xié-jù (to balance the sentence). The terms zú-jù, yuán-jù and xié-jù
used by Kong do not refer to the syntax of the sentence, because (22a), for example, would be perfectly grammatical without qín (bird): shì mín rú qín shòu (to treat
people like animals).The terms also do not refer to the semantics of the sentences,
because, as Kong points out, in (22a) the classical text says only tigers, rhinoceros
Prosodic word in Classical Chinese
35
and wolves, that is, only beasts and no birds.Therefore, adding qín (bird) to explain
the text is not semantically motivated. If these terms refer neither to the syntax
nor the semantics of these sentences, what do they refer to? In fact, the term xiéjù
(to balance the sentence) in (22c) suggests clearly yáng zhı- pí (sheep’s skin) would
not be “balanced”. When another syllable ga-o (lamb) is added to yáng (sheep),
it becomes a four-syllable phrase and is then “balanced”. As we shall see below,
four-syllable units naturally fit into the foot structure, while three-syllable units
are highly marked (see, for example, (43) below). Given the contrast between a
balanced four-syllable unit and an unbalanced three-syllable unit, it is clear that
the term xiéjù as well as yuánwén and zújù, all refer to the prosody.
Following Kong’s intuition, and based on the analysis below, Feng (1997a)
proposed that the increase in disyllabicity during the Han dynasty was triggered
by a new prosodic structure which occurred as a result of a new, simplified syllable structure in Old Chinese.
2.4.3. The independence of disyllabicity and compounding
It is a truism in Chinese linguistics that the change from monosyllabic to disyllabic words has been a strong tendency throughout Chinese history (Wang
1980). In this section, I would like to point out that, although the statement is
generally correct, it is misleading. This is because the statement is usually interpreted as meaning that two-syllable combinations were constructed in order to
satisfy the need for new disyllabic words (compounds). However, the opposite
may be what actually happened: disyllabic words were created to satisfy the need
for two-syllable prosodic units. More specifically, we will show that the disyllabicity was a fundamental requirement of the language, and its development
had inherently nothing to do with compounding, i.e., the language required
disyllabic units regardless of whether they resulted in compounds or not. The
reasons for the tendency toward disyllabicity will be given in section 2.5.3 and
the reasons supporting the development of disyllabic compounding will be
given in section 2.5.5.
The hypothesis that disyllabicity is independent of compounding is supported by the following facts. First, most early documents show that twosyllable combinations originally were not compound words, but two­syllable
phrases. For example: guó-jia- 國家 ‘country’ became a compound word, but was
originally a phrase:
23. 人有恆言， 皆曰天下國家。
《孟子·離婁上》
Rén yŏu héng yán， jie- yue- tia-nxià guó jia-.
people have consistent word all say world country family.
‘People who have constant behavior all consider the world (of the Emperor),
the country (of the feudal princes) and the family (of the high officials
of the state).’ (Mencius. Lilou. Shang)
36 Prosodic word in Classical Chinese
國之本在家， 家之本在身。
《孟子·離婁上》
Guó zhı- běn zài jia-, jia- zhı- běn zài she-n.
country’s base at family, family’s base at body
‘The base of the fiefdom is the family; the base of the family is the person.’
(Mencius. Lilou.Shang)
It is well known that in the Zhou dynasty (c. 1000 BC), guó referred to the fief
granted to feudal princes enfeoffed by the Emperor, and jia- was the fief of high
officials of state who were enfeoffed by the feudal prince. The combination of
guó – jia- conveys the meaning of both “the fiefdoms of feudal princes” and “the
fiefdoms of high officials”. It became a compound word (guó-jia- means only
guó ‘country’, not jia- ‘family’) only during the Warring States Period (500–200
BC), when the social and political system changed.This example indicates how
disyllabicity was independent of compounding, because many two-syllable
units were originally phrases, not compounds.
Secondly, in order to become a compound, according to the semantic criterion (8), a disyllabic phrase must undergo a process of lexicalization through
specialization of sense. For example (taken from Dobson 1959): tia-n-xià ‘skybelow’ and zhǎo-shàng ‘pool-above’ are both used in Mencius. Literally, tia-n-xià
can be interpreted as ‘of the sky’, ‘the below part’; that is, ‘below the sky’ and
zhǎo-shàng would be ‘of the pool’, ‘the above part’, that is, ‘above the pool’.
Zhǎo-shàng in natural language means ‘above the pool’, while tia-n-xià has a
specialized meaning and refers to ‘all below the skies’ → ‘the world of men’ →
‘society’→ ‘the domain of the Emperor’. Tia-n-xià is thus a lexicalized compound, while zhǎo-shàng is still a phrase. This is to say that without lexicalization, a two-syllable combination will remain a two-syllable phrase and cannot
be considered as a compound according to the criterion in (8).
Thirdly, under the pressure to form two-syllable units, some combinations
look like compounds, even at the beginning of their formation. For example:
24. a. 衣裳 yı--cháng
b. 家室 jia--shì
c. 圖書 tú-shu-
shirt-skirt, ‘clothes’
family-bedroom, ‘family’
picture-book, ‘publication’
However, the two parts are interchangeable:
25. a. cháng-yı- ‘clothes’ (Shijing) → yı--cháng ‘clothes’ (Shijing)
b. shì-jia- ‘family’ (Confucius) → jia--shì ‘family’ (Shijing)
c. shu--tú ‘publications’ (Hanfeizi ) → tú-shu- ‘publications’ (Hanfeizi )
The variable order of the forms AB and BA makes it unlikely that the formed
units are words, since the interchangeable order between A and B is a strong
feature of coordinated phrases. The free order suggests that as long as twosyllable forms are established, it does not matter whether the coordinated form
is AB or BA, because the purpose here is to form disyllabic units, not to form
compound words in the first place. The above argumentation suggests that the
Prosodic word in Classical Chinese
37
function of two-syllable units was a fundamental need of the language, regardless of whether the outcome was a word or a phrase.
2.5. A metrical approach to classical Chinese compounds
2.5.1. Syllable structure simplification in Old Chinese
As is mentioned in 2.3.1, phonological changes from Old Chinese to Medieval
Chinese resulted in a simplification of syllable structure. According to Ting
(1979:717–736) and Yu (1985:290), the maximal and minimal syllable structures in Old Chinese and Medieval Chinese are as follows (C = consonant;
M = Medial;V = Vowel, S = Semivowel).
26. Chronology
Maximal
Old Chinese (c. 1000 BC)
CCCMVCCC
Medieval Chinese (800 AD) {C, S} V {C, S}
Minimal
CVC
CV
The crucial point here is that final consonants in Medieval Chinese could
not occur in clusters, and were limited only to two types: nasal -m, -n, -ng
and oral -p, -t, -k stops. If we compare Medieval Chinese with Old Mandarin
(OM) (Dong 1954,The reconstruction of Zho-ngyuán Yı-nyùn ([The Zhongyuan
Rhyme Dictionary] 1324 AD), we will see that the process of simplification of
syllable structure was still active at that time.
27. Medieval Chinese Syllable Endings: CV (-m, -n, -ng, -p, -t, -k)
Old Mandarin Syllable Endings: CV (-m, -n, -ng)
The process of final consonant attrition continues in Modern Mandarin
(MM) as seen in (28) and has been traced by Chen (1975) as given in (29).
He reconstructs a reduction schema that applies to the history of Chinese and
many other languages.
28. Old Mandarin Syllable Endings: CV (-m, -n, -ng)
Modern Mandarin Syllable Endings: CV (-n, -ng)
29. Vn > Ṽn > Ṽ > V
According to a sociolinguistic study of the Beijing dialect by Barale (1982),
the final-nasal consonant attrition noted by Chen follows a process of nasalization of the preceding vowel as seen in (29). Furthermore, Wang (1993) suggests
that Mandarin Chinese syllables can all be analyzed as open syllables, that is, the
maximal syllable structure in Mandarin Chinese is arguably CV.17
Juxtaposing the syllable endings of different periods gives us a clear picture of the process of syllable-structure simplification throughout Chinese
history:
38 Prosodic word in Classical Chinese
30.
OC
MC:
OM:
MM:
Ongoing
MM:
CCVC(C)(C)
-m
-n
CV ( -ng )
-p
-t
-k
-m
CV ( -n
)
-ng
-n
)
CV (
-ng
CV
The first step of the process is to drop the ‘postcoda’ (Baxter, 1992), and the
second step is to drop the coda. There is clearly a strong tendency to simplify
Chinese syllable structures by dropping final consonants.
Taking all the considerations into account, the internal mechanism hypothesis offered here seems quite plausible as an explanation for the rise of disyllabification during the Han dynasties before the completion of the four-tone
system by the Wei Jin Period – another factor that correlates the disyllabification process. This is so because it offers a principled reason for the change: the
newly developed tone-bearing syllables can no longer naturally and systematically differentiate the weight between syllables in terms of moras. The syllable
simplification inevitably weakened the moraic foot system of Archaic Chinese
(see below), thus giving rise to the new disyllabic foot structure in Medieval
Chinese.
2.5.2. Metrical theory and Old Chinese syllable structure
In metrical theory, syllables with a CVCC structure are heavier than syllables
with a CVC structure, and CVC syllables are heavier than CV syllables (see
Goldsmith 1990, and references cited there). The process of syllable simplification outlined above clearly shows that syllable weight has continuously decreased
throughout Chinese history. Given this fact, the phonological change from
Old Chinese to Medieval Chinese can be characterized in terms of syllableweight reduction. An important consequence of the syllable-weight reduction
within the new system is that a single syllable was not heavy enough to form
a minimal independent prosodic unit – a foot. In other words, the new system
requires the minimal prosodic unit (the foot) to be formed not by one, but by
two syllables.
This hypothesis implies that a one-syllable foot was permissible before
the final clusters disappeared, but not afterward. Theoretically, this may be
Prosodic word in Classical Chinese
39
justified as follows. In prosodic phonology, in general the structure of a foot
can be characterized as consisting of one relatively strong and any number
of relatively weak syllables dominated by a single node, according to the
Relative Prominence Rule (Liberman and Prince 1977; Kiparsky 1979;
Nespor and Vogel 1986, and references cited there). Therefore, the structure of a binary foot would be as follows (‘f ’ stands for a foot, and ‘σ’ for a
syllable):
31.
f
σ
{s
σ
w}
However, based on an analysis of a large number of languages, Hayes (1980)
concludes that there are fairly strong restrictions on the grouping of syllables
into feet in any given language. That is, a language may have either binary feet,
consisting of two syllables each, or unbounded feet, consisting of (theoretically)
any number of syllables. In addition to these types of foot, one-syllable feet are
also found, although they are highly marked.
Since Old Chinese was basically a monosyllabic language (observing the
Morphosyllabic Rule to be discussed in next section), it is reasonable to
assume that while a foot in Old Chinese may have consisted of more than
one syllable, a one-syllable foot would also have been allowed, because, theoretically, the maximal syllable structure in Old Chinese was CCCMVCCC
(Ting 1979; Yu 1985), which is, in prosodic terms, not only a heavy, but a
“super-heavy” syllable structure (see McCarthy and Prince 1993). Empirically, as argued in Feng (2013) and Shi (2014), there were different types
of syllables that are contrastive between short and long syllable (e.g., the
contrast between weak and strong pronouns such as 吾 ŋa vs. 我 ŋai) and
the alternation between weak and strong interrogative particles such as 邪
laa vs. 也 laalʔ > laiʔ), etc.). In other words, the moraic foot has played an
important role in Archaic Chinese grammar and sound evolution (Zhengzhang 2017). Thus, in Archaic Chinese, heavy syllables with complex structures may form feet independently, while light or weak syllables with simple
structures may require another syllable to form a foot. Since the Chinese
syllable has changed from being heavy to being weak, it lost the ability to
independently form a foot.
2.5.3. Syllable structure simplification as a cause of possible
disyllabic foot
Within the framework of prosodic phonology, whether a syllable is heavy or
not depends on whether the rhyme constituent of the syllable is geometrically
branching. A heavy syllable is defined as one which has a branching rhyme, and
40 Prosodic word in Classical Chinese
a light syllable is defined as one without a branching rhyme.The “weak-nodesdon’t-branch” principle of metrical theory would allow a CVC syllable to have
the following structure:18
32.
σ
Onset
rhyme
nucleus
coda
V
C
C
Thus, aside from the obvious increase in length, a CVC syllable structure
must also be considered theoretically “heavier” than a CV structure, because
it has a branching rhyme. Note that this is exactly the difference between Old
Chinese and Medieval Chinese with respect to their basic (minimal) syllable
structures, as proposed by Ting (1979), Li (1980) and Yu (1985). Furthermore,
the syllable structure of Old Chinese is not only minimally CVC, but also
maximally CCCMVCCC, i.e., a consonant cluster is allowed in word-final
position.The hypothesis that final clusters created super-heavy syllables in Old
Chinese is supported by looking at other languages where final clusters also
form super-heavy syllables. For example, in Arabic, word-final consonant clusters are permitted and syllables that contain such final clusters are super-heavy.
The interesting fact about Arabic is that it is the syllable-final consonant that
“creates” the super-heavy syllables. McCarthy (1979 and elsewhere) suggests
a structure like (33) for super-heavy syllables (see also Goldsmith 1990:198):
33.
σ
σ
σ
Onset
C
rhyme
V
C C
What is crucial to note here is that the final CC cluster in a syllable, metrically speaking, behaves differently from a single consonant. This is not to say, of
course, that Old Chinese was necessarily exactly like Arabic in terms of prosodic
structure.19 Nevertheless, the Arabic case provides us with evidence that final
consonant clusters may create super-heavy syllables, allowing such syllables to
independently form a foot. Thus the complex syllabic structure in Old Chinese
may hypothetically be organized in terms of “foot”, according to (33), as follows:
Prosodic word in Classical Chinese
34.
41
f
σ
σ
Onset
C
rhyme
V
C C
The hypothesis that Old Chinese had a heavy syllable structure and hence
permits one-syllable feet is also supported by the moraic theory of syllable
structure, in which a mora (μ) is dominated by a syllable node σ and syllables are dominated by feet (f ). The syllable node σ may dominate one or
two moraic nodes, with each mora dominating at most one segmental element. Consequently, consonants are daughters of σ (see McCarthy and Prince
1993:21). The following structures illustrate this analysis:
35.
a.
σ
b.
σ
μ μ
μ
C
V
C
V V
A foot must be either bimoraic or disyllabic, a stipulation required by the
Foot Binarity Principle:
Foot binarity (McCarthy and Prince, 1993:43)
Feet must be binary under syllabic or moraic analysis.
Based on the moraic theory of syllable structure and the Foot Binarity Principle, the structure (35a) cannot form a foot because there is only one mora,
which violates the Foot Binarity Principle. Structure (35b), however, will form
a perfect foot because there are two moras, thus meeting the Foot Binarity
Principle requirement that a foot must be at least bimoraic. Based on this theory, we may reasonably propose that the basic syllable structure of CVC is able
to serve as an independent foot in Old Chinese,20 as shown in (36).
36.
f
σ
C
μ
V
μ
V
42 Prosodic word in Classical Chinese
Note that this theory also predicts that if final consonant clusters are dropped
from the language, we will have a structure of the following, as illustrated in
(37) below:
37.
a.
f
b.
σ
σ
onset
C
f
|
σ
rhyme
μ
|
V
onset
μ
|
C
C
C
rhyme
μ
|
V
μ
|
C
The prosodic weight of the CVCC foot is reduced. If we assume that the loss
of coda reduces the minimal syllable structure to CV in Medieval Chinese, we
lose the phonological basis for bimoraic feet:
38.
C
μ
V
f
f
σ
σ
μ
V
C
μ
V
The loss of the post-coda results in a loss of the super-heavy syllable structure, and the loss of the coda results in a loss of the moraic branching structure.
Since both apparently occurred in the language, the resulting structure would
no longer be able to serve as an independent foot.21 Furthermore, if the language changed its syllable structure systematically from (a) to (b) in (37) and
(38), the result would have been that one-syllable words (since Old Chinese is
basically a monosyllabic language) would no longer constitute the independent feet. If this is so, two-syllable combinations will come to play a major role
in foot formation in the language. To restate, the bimoraic foot disappeared in
Old Chinese due to the loss of final consonants and consonant clusters. This,
in turn, leads to the loss of heavy and super-heavy syllables. Since the foot is an
obligatory level of prosodic structure, according to the theory presented above
(see also Selkirk 1980b, McCarthy and Prince 1991, 1993; Kager 1992, and
many others), the language made up for the loss of bimoraic feet by replacing
them with disyllabic feet. As a result, the change of syllable structure from Old
Chinese to Medieval Chinese can be prosodically characterized as a change
from bimoraic to disyllabic feet, resulting in the tendency to form two-syllable
combinations.22
Prosodic word in Classical Chinese
43
2.5.4. Evidence for bimoraic foot changing into syllabic foot
Recent studies (Feng 2012, 2013, 2015; Shi 2014; Zhengzhang 2017) have
shown that Archaic (or Proto-) Chinese had a moraic foot structure that differed from Medieval Chinese, which had a disyllabic foot structure (Pulleyblank
1962, 1977–1978; Pan 2000; Zhengzhang 2003). For example,
39. a.
b.
吾喪我。
《莊子·齊物論》
*ŋra sàng *ŋajʔ.
1SG lost 1SG
‘I lost myself.’
Zhuangzi: Qiwulun (c. 369–286 BCE)
《孟子·滕文公上》
彼丈夫也，我丈夫也，吾何畏彼哉？
*prjajʔ zhàngfu yě, *ŋajʔ zhàngfu yě, *ŋra hé wèi prjajʔ za-i?
3SG man
PRT 1SG man PRT 1SG why afraid 3SG PRT
‘HE is a man, I am a man, how come I am afraid of HIM?’ Mencius:
Tengwengong Shang (c. 372–289 BCE)
As seen above, the emphatic (that is, heavier) forms, as contrasted with nonemphatic (weaker) counterparts indicate an important fact that the mora, rather
than the syllable, was the basis for determining the prosodic weight in Archaic
Chinese. The “吾/我” contrast makes perfect sense in terms of Focus Prosody Correspondence Principle (Zubizarreta 1998:88): the focused element is
stressed (or heavier). In other words, when pronouns are used in stressed positions (the object position for example as in (39a)) or contrastively (as in (39 b)),
heavier forms such as *ŋajʔ (containing more than one mora) are favored over
their lighter counterparts such as *ŋra (containing only one mora) for the firstperson pronoun.
In terms of pronoun in Archaic Chinese, Pan(2001) provides the following
demonstrative distinction encoded through vowel alternations between stressed
(or emphatic) and unstressed (weakened) forms in Old Chinese:
40. 茲
此
之
是
*ts
*tsh
*tj
*dj
-ə
-eʔ
-ə
-eʔ
‘this’
‘this’
‘this’
‘this’
Cı̆ 此 and shì 是 are among the stressed (or emphatic) forms and the contracts between 此/是 and 兹/之 are distinguished by a heavy syllable (which
is more sonorant or/and has more moras) with a light syllable (which is less
sonorant or/and has fewer moras).Thus, we may conclude that Archaic Chinese
may indeed be a quantity-sensitive language.
There is another type of evidence supporting the hypothesis that moraic feet
in Archaic Chinese function actively, that is, the two-syllable per line structure
of the earliest poems, as seen below (from Wúyuè Chu-nqiu- 吳越春秋).
44 Prosodic word in Classical Chinese
41. 斷竹， 續竹； 飛土， 逐肉。
*tonʔ *trjuk, *ljoks *trjuk; *pjəj *hlaʔ, *drjiwk *njuk.
cut bamboo connect bamboo fly
earth chase flesh
‘Cut a bamboo (and) string it (into a bow); fly the pellet (and) hunt animals.’
This type of the oldest poem demonstrates that one syllable could form an
independent foot (a poetic meter) because no poetic lines are in general formed by
fewer than two feet, and if *tonʔ *trjuk 斷竹 ‘cut bamboo’ is a poetic line, then it
must consist of two prosodic units (or two feet). If, furthermore, *tonʔ *trjuk forms
two prosodic units (feet), then *tonʔ and *trjuk must each be a prosodic unit. Given
this, if *tonʔ 斷 (or *trjuk 竹) is a syllable as is generally assumed in the literature
and if it is also a prosodic unit (foot) as shown above, then one syllable must be a
prosodic unit. Since there is no prosodic unit (foot) without a branching structure
(by the Relative Prominence Principle), the syllable *tonʔ and *trjuk must be
analyzed as a branching prosodic structure. Since a syllable branching structure is
analyzed in terms of moras in metrical theory, the archaic syllables of *tonʔ and
*trjuk in the disyllabic poetic line are inevitably analyzed in terms of moras. This
entails further that each syllable has at least two moras (or two moraic positions) in
a poetic line formed by two syllables, giving rise to a moraic foot structure, which
is, again, a corollary that Archaic Chinese is a weight-sensitive language.
This type of moraic foot structure was replaced by a syllabic foot structure
later in the language and the following evidence shows that it is the case: a
monosyllabic word was unable to form an independent foot during the Warring States Period (475–221 BC), indicating that a two-syllable unit was functioning as a standard foot in positions where an independent foot was required,
as shown in the following analysis:
42. a.
b.
《尚書·益稷》
子何言？
Zı̆ hé yán?
You what say
‘What do you say?’
(Shangshu.Yiji, 1000 BC)
《論衡·禍虛》
是獨遵何哉？
Shì dú zu-n hé za-i?
It only follow what prt.
‘What does it expressly follow?’
(Lunheng. Huoxu, 100 AD)
Although Classical Chinese of the Pre-Qin period (221 BC) is basically an
SVO language, different types of SOV order are clearly observed. For example,
if the object of a verb is a wh-expression, it must occur directly to the left of the
verb as shown in (42a).This type of SOV word order (i.e., wh-V) changed after
the Han dynasty. In example (42b) a wh-object would follow the verb in the
Han text Lunheng. However, when an object wh-expression is formed by two
constituents, e.g., hé zuì ‘what guilt’, it does not appear to the left of the verb
before the change from [wh-V] to [V-wh]: *[What-N V].
Prosodic word in Classical Chinese
43.
45
*宋何罪有？
*Sòng hé zuì yŏu?
Song what-guilt have
‘What guilt does Song have?’
Rather, the structures that are allowed are [what-N pro-V] or [V what N].
For example:
44. a.
《墨子·公輸》
宋何罪之有？
Sòng hé zuì zhı- yŏu?
Song what-guilt it have
‘What guilt does Song have?’ (Mozi Gongshu)
《晉語》韋昭注
b. . . . 有何舊怨？
. . . yŏu hé jiù yuàn?
. . . have what old complain
‘What old grievance do (you) have?’ ( Jinyu 4,Wei Zhao Zhu)
Either a pronoun zhı- ‘it’ is inserted between the wh-expression and the verb
in earlier documents, or the wh-expression appears to the right of the verb after
the Han dynasty.
The question, then, is why *[hé zuì yŏu] (what-guilt have) is not well formed
while [hé yŏu] (what have) is. In Feng (1994), it is proposed that (i) ProtoArchaic Chinese was an SOV language and it changed into an SVO language,
which is what we see as Old Chinese (1000 BC). Based on this, the SOV orders,
such as the [wh-V] structure, are considered as remnants of the change from
SOV to SVO. In order to account for the survival of SOV phenomena, Feng
(1994) proposed that the primary sentential stress (i.e., the NS, as seen in (45))
falls on the right side of the sentence,23 given that Classical Chinese was basically an SVO language.The Nuclear Stress Rule is formulated as follows (taken
from Liberman 1975):
45. Nuclear Stress Rule (NSR)
For [X Y]P , if X and Y are constituents of P, and if P is the last phrase of a
sentence, then Y must be stressed.
According to the Nuclear Stress Rule, a sentence is acceptable if the last
element of the last phrase is properly assigned a stress, otherwise it will be
ill formed prosodically. Following this analysis, the non-existence of (40)
is accounted for by saying that yŏu is the last element of the sentence, and
the last phrase that contains yŏu is the VP structure *[hé zuì yŏu], therefore,
hé zuì will be the X and yŏu is the Y of the Sentential Prosodic Rule. However, the monosyllabic word yŏu ‘have’ is not heavy enough to act as an
independent foot to realize the primary stress in the following structure:
46.
46 Prosodic word in Classical Chinese
f
[X
W
σ
he
σ
zui
*Y]
S
|
σ
you
That is, within the prosodic domain of [X Y]VP, X consists of a branching
node, while Y consists of only a non-branching node, therefore,Y cannot realize the NS at the end of a sentence. Technically speaking, according to Liberman and Prince’s Relative Prominence Principle (1977), a strong node must
be licensed by a weak node. This implies that the stress cannot be realized on
Y itself, because it is a single node, and as we have seen before (see 2.5.3), one
syllable cannot serve as a branching node in a prosodic structure. In the branching node VP,Y still cannot realize the stress, because X, the sister node of Y, is a
branching node, and is prosodically stronger than a single node, i.e., than Y. As
a result, (46) must be ruled out. The implication of this analysis predicts that if
another syllable is attached to the node Yin (46), or the elements under the X
node are reduced to a monosyllabic wh-expression, then the NS can be realized
(on a disyllabic foot), and the sentence will be grammatical.This is exactly what
happened, as we will see below.
Hé yŏu ‘what have’ is grammatical, because hé yŏu is not only the last phrase
but also a minimal prosodic unit, namely, a foot. Therefore, the primary stress
can be assigned to the right element yŏu.
47.
VP
Y]
S
you
have
σ
[X
W
he
what
σ
f
Yŏu hé zuì ‘have what guilt’ is also grammatical because hé-zuì is the last
phrase (NP) and these two words form an independent foot with the stress on
the right, satisfying the requirement of NSR.
48.
NP
Y]
S
zui
guilt
σ
[X
W
he
what
σ
f
Prosodic word in Classical Chinese
47
The structure of hé-zuì zhı- yŏu ‘what guilt it-have’ is also acceptable, because
hé zuì zhı- yŏu forms the last phrase (VP) in the sentence, where hé-zuì is still
the object of the verb, zhı- is cliticized onto the verb yŏu24 (forming a prosodic
foot with you), hence zhı--yŏu would be interpreted as the Y and hé-zuì as the
X, illustrated in (49).
49.
VP
[X
f
σ
σ
he zui
Y]
f
σ
zhi
σ
you
Since there are two syllables under the Y position, they can form a standard
foot so that the stress can be assigned to it, satisfying the NSR. Note that if hé
yŏu is grammatical, there is no reason to rule out hé-zuì yŏu either syntactically
or semantically. The only difference between these two structures is their prosodic structure.Thus the best way to explain the non-existence of *hé-zuì you is
to assume that yŏu is a monosyllabic word and one syllable is not heavy enough
to act as a standard foot.25
This is further confirmed by examples of the following kind in which an
extra meaningless syllable is used in order to form a disyllabic foot.
50. 夥頤， 涉之為王沉沉者！
《史記·陳涉世家》
Huŏ-yí, shè zhıwéi
wáng chénchén
zhě.
Great-yi, She Nom.prt being King magnificent prt.
‘Great, the way that She became a king is magnificent!’ (Shiji. Chenshe
Shijia)
The sentence is traditionally taken to be closest to the vernacular given by
Sima Qian (145–? BC). Probably because the word huŏ-yí used in the Chu dialect is relatively uncommon, Fu Qian (c. 184?–? AD) glosses it as in (51):
51. 楚人謂“多”為“夥”，又言“頤”者，助聲之詞也。服虔，《史記索隱》
Chŭ rén wèi duo- wéi huŏ, yòu yán yí zhě, zhù she-ng zhı- cí yě.
Chu people call great is huŏ, again say yi N.prt. support sound ’s word prt.
‘In Chu dialect, the word for “great” is huŏ. However, “yí ” is added to make
the sound better.’ (Fu Qian, Shiji Suoyin)
According to Fu Qian, the meaning of the exclamation expression Huŏ-yí in
(50) is interpreted as the same as the monosyllabic left-hand constituent huŏ, thus
making yí semantically empty. Here, the addition of yí to huŏ occurs to lend metrical support to huŏ as Fu Qian notes. The fact that a monosyllable needs extra
“metrical support”, while a disyllabic unit does not (see Guo 1985) indicates that
a monosyllable is not heavy enough prosodically to act as an independent foot
needed to realize the stress on an exclamation or a focus expression. Therefore,
48 Prosodic word in Classical Chinese
the use of “metrical support” on a monosyllable provides further evidence for
the argument that a disyllabic unit constitutes a standard foot.
2.5.5. Origin of foot formation and prosodic words
Given the prosodic analysis in section 2.5.3 and the textual evidence in 2.5.4, a
Foot Formation Rule for Classical Chinese is therefore formulated as follows:
52. Foot Formation Rule in Classical Chinse
f
σ
σ
A standared foot must be formed by at least two syllables.
As we have seen, disyllabic feet resulted from syllable reduction, therefore
the Foot Formation Rule must apply chronologically after the loss of final
consonant clusters in Old Chinese. As shown before, there was a sharp increase
in disyllabicity during the Han dynasty, and it is well known that by the Han
dynasty, final consonant and consonant clusters had almost disappeared completely (Mei 1980, Baxter 1992). The fact that the development of disyllabicity
followed the loss of the final consonants and consonant clusters is chronological
evidence corroborating the Foot Formation Rule given in (52).
If the above analyses are correct, we have answered the question of where
the tendency to disyllabicity originated. Recall that we have also argued (e.g.,
2.4.3) that the development of disyllabicity was theoretically independent of
compounding. The Foot Formation Rule in (52) serves as a system-internal
evidence supporting this hypothesis.
Now, if disyllabicity did not directly result in compounding, why did the
morphology in Classical Chinese proceed in the direction of compounding
and what is the relationship between the development of disyllabicity and that
of compounding?
Feng (1995) argued that although the development of disyllabicity is inherently independent of the development of compounding, the Foot Formation
Rule played a crucial role in word formation in Classical Chinese. This is not
because disyllabicity is inherently related to compounding as a means of word
formation, but because of the fact that Classical Chinese was basically a monosyllabic language (see Chapter 3 for a precise definition of the notion of monosyllabicity). Once the monosyllabic nature of the language is assumed, disyllabicity
can then be considered a “cause” for the development of compounding.
2.5.5.1. Prosodic word and compound word
The relationship between disyllabicity and compounds can be naturally derived
from the recent developments in the theory of prosodic morphology (see
McCarthy and Prince 1993). In prosodic morphology, prosodic restrictions are
Prosodic word in Classical Chinese
49
defined in terms of prosodic categories such as mora, syllable, foot and prosodic
word (PrWd), which are hierarchically organized (see Selkirk 1980 a, 1980 b;
McCarthy and Prince 1993):
53.
Prosodic Hierarchy
PrWd
|
Foot
|
Syllable
|
Mora
In this theory, any instance of the category Prosodic Word (PrWd) must contain at
least one foot.According to Foot Binarity Principle, every foot must be bimoraic or
disyllabic.Thus a PrWd must contain at least two moras or syllables.The “at least”
requirement automatically leads to a notion about what would be the smallest
Prosodic Word: a minimal Prosodic Word is a metrical foot. As argued by McCarthy and Prince (1993), the Prosodic Hierarchy and Foot Binarity, taken together,
derive a notion “Minimal Word”. We shall see below how the notion “minimal
word” interacts with the disyllabic foot and phrase structure rules (a detailed discussion on minimal word in Mandarin Chinese will be given in Chapter 5).
2.5.5.2. Phrase structure correspondence and Idiomatized PrWd
As mentioned before, Classical Chinese was a monosyllabic language. If a foot
must be formed by two syllables, and each syllable is a word in the language, the
only way to make a disyllabic foot in the language would have been to group
two words together as shown below (“wd” stands for Word):
54.
f
σ
wd
σ
wd
In other words, a disyllabic foot inevitably results in a two-word prosodic
combination. That is,
55. Foot = σ+σ = wd+wd
A disyllabic foot is construed on basis of two-word combinations in the
monosyllabic system.
However, such combinations are also constrained by phrase-structure rules
in the language.26 It follows that feet that are realized on two syllables/words
would often happen to correspond to phrases (XP), that is:
50 Prosodic word in Classical Chinese
56. Foot = σ+σ = wd+wd = XP
That is, the equation “σ = wd” inevitably leads to the equation “F = XP”.
Once a foot corresponds to a phrase, the prosodic foot will merge with the
phrase, due to structural isomorphism. When this happens, the following situation arises:
57.
F
σ=Wd
wd
σ=Wd
wd
XP
(57) illustrates that a correspondence between a prosodic foot and a syntactic phrase will eventually lead to a merging of these two structures. Since, by
the Prosodic Hierarchy in prosodic morphology, a foot is dominated directly
by the Prosodic Word, and the minimal prosodic requirement for a word is
the presence of one foot, the merging of a prosodic category (a foot – the
minimal prosodic requirement for “word”) with a syntactic category (a phrase)
would readily satisfy the Prosodic Word requirement. Therefore, the merged
structures all have the potential to form PrWds in the prosodic morphological system. Note that the prosodic integrity of foot always forces two elements
in a phrase to be closely knitted together or prosodicazed; hence one element
cannot occur without another, otherwise it will violate the minimality requirement for being a prosodic word.
Under this analysis, however, when a prosodic word is repetitively used in
the language, the two elements in that phrase will be fixed, resulting in what is
called an Idiomatized Prosodic Word (Feng 2015). This analysis proceeds from
the assumption that all of idioms are phrasal categories. Note that by only one
step further the Idiomatized PrWds can be lexicalized as compounds. That is,
compounds are lexicalized idiomatic phrases.
Given the above analysis and based on the Prosodic Hierarchy in (53) and the
structure (57), which is derived from the Foot Formation Rule, Feng (1997b)
proposed Word Formation Rule for Classical Chinese as in (58):
58. Word Formation Rule in Classical Chinese
PrWd
F
X
Y
XP
Prosodic word in Classical Chinese
51
X and Y form a prosodic word, if the combination of X with Y simultaneously satisfies the syntactic and prosodic conditions of being a phrase and a foot,
respectively.
(58) entails that compound words in Classical Chinese are syntactic words
because they historically originated from disyllabic phrases, and compound
words are also prosodic words because they are Lexicalized Idiomatic PrWds.
It follows that not every phrase can develop into a compound, but only those
which meet the prosodic requirements. No foot can be identified as a compound, except for those that represent an independent syntactic unit, i.e., a
phrase. By prosody, only phrases that fit the description of being one foot are
eligible to become compounds. By syntax, only feet that represent independent
phrases are qualified to be compounds.
Given all the analyses above, the origin of compounding can now be understood as follows: the phonological change of Old Chinese resulted in a disyllabic
foot; the disyllabic foot, in turn, resulted in disyllabic PrWds; disyllabic PrWds
are formed by two-syllable phrases given the monosyllabic property of the language; and the two-syllable phrases are idiomatized to usage, becoming Idiomatized PrWds. When Idiomatized PrWds are lexicalized, they become an X0
level category item, i.e., a compound word in the lexicon, as illustrated in (59):
59.
f
A
PrWd
B
XP
A
B
XP
Idiomatized
PrWd
A
B
XP
Compound
Word
A
B
X°
This is how disyllabic phrases, compounds and the prosodic morphological
system came about.
Since the disyllabic foot became standard, and a foot is the minimal unit for
a PrWd, forming a standard foot in the language will eventually lead to Idiomatized PrWds and the ensuing compounds in the language. Compounds
are therefore the result of foot formation. This is why disyllabic compounds
increased in number after the establishment of foot formation. Given the theory presented here, the semantic criterion in (8) can therefore be replaced27 and
most separable disyllabic combinations will all be treated as Idiomatized PrWds
listed in the dictionary. Compounds are only those that have clearly undergone
a process of lexicalization (or a category-changing rule, see Feng 1995:141)
such as sı--mă ‘general’ of (4d).
The theory presented here requires that the prosodic argument of being one
foot and the syntactic relation of being a phrase interact to determine PrWds
and compounds in Classical Chinese: the syntax determines the structural relation between each element of a compound; the prosodic template of a foot
determines the metrical shape of that compound. Compounds are identified
52 Prosodic word in Classical Chinese
only by a process of lexicalization. Any two-syllable combination that is closely
knitted together and listed in the dictionary, but exhibits some phrasal properties, will belong to the category of Idiomatized PrWds. Under the treatment of
Idiomatized PrWd, there is no surprise why the two forms A and B in coordinating structures, such as tú-shu- ‘picture and book’ given in (25) can be formed
as either tú-shu- 圖書 or shu--tú 書圖 – because they are idiomatized phrases, and
both orders AB and BA satisfy the requirements for an Idiomatized PrWd in a
coordinating structure.This also explains why ju--mă ‘carriage and horses’ can be
formed by two words, but without the surface meaning of the second word, as
seen in (18). Because the Foot Formation Rule demands that a minimal prosodic word be formed by at least two syllables, ju- must take another word (here,
mă, in the same semantic field with ju-) to meet this requirement. The PrWd
licenses ju--mă to function as an independent prosodic unit, even though the
actual meaning of ju--mă is focused on only ju-.
2.6. Empirical consequences and theoretical implications
If, as shown above, the bimoraic foot lost its phonological basis, and the twosyllable unit came to constitute the standard foot in Classical Chinese, what we
would expect empirically is for two-syllable combinations to become more
and more common during the course of the change. Given the fact that Classical Chinese was basically a monosyllabic language (following the Morphosyllabic Rule to be discussed in Chapter 3), and given the Foot Formation Rule
requirement that a standard foot must be formed by a unit at least two syllables
long, the only way to make a disyllabic foot in the language would have been to
group two words together. As shown above, a disyllabic foot would often result
in two-word prosodic combinations and such combinations would also be constrained by phrase-structure rules in the language. It follows that the prosodic
foot would, in turn, often result in idiomatized Prosodic Word. If two-word
combinations were the only way to realize disyllabic feet, we would expect
that, in the early stages, naturally occurring syntactic two­word phrases would
be highly preferred candidates to act as two-syllable feet. More explicitly, it is
more likely that naturally occurring phrases would bear two-syllable feet than it
is that entities (two-syllable words) would be created expressly for that purpose.
If disyllabic feet are originally realized on naturally occurring phrases, the
result of these developments would be the following: (“→” means “result(s) in”)
60. phonological change → disyllabic feet → disyllabic phrase →
123
idiomatized PrWd → compounds
4 5
Since disyllabic feet are mostly realized on syntactic two-word phrases, it is
likely that naturally occurring two-word phrases would be the first candidates
Prosodic word in Classical Chinese
53
for disyllabic feet at the beginning of the development of disyllabicity. Also
compounds would originate from these naturally occurring disyllabic phrases.
This hypothesis receives support if we find that disyllabic combinations
(phrases or compounds) in Classical Chinese did originate from naturally
occurring disyllabic phrases, rather than from those expressly created for the
prosodic requirement.
How can we distinguish the naturally occurring phrases from those created expressly for prosodic requirements? Furthermore, how do we distinguish
disyllabic combinations that originated from naturally occurring phrases from
those that were created expressly for the prosody?
Considering the first question, we have seen that there are two structures
which are very productive for compounding, namely, coordinating and subordinating structures (see section 2.3). We also know that each of these structures
can be formed by different types of syntactic relations among the two elements
they contain. For example, the coordinating structure can be formed by a noun
plus a noun, or a verb plus a verb, etc., and the subordinating structure can be
constructed by a noun modifying a noun, or an adjective modifying a noun, etc.
According to Cheng (1981), there are six types of coordinating and nine types
of subordinating structures as shown in Table 2.6 below (N = noun, A = adjective,V = verb, P = pronoun, Num = number).
Table 2.6 Types of coordinating vs. subordinating structures
I. Coordinate structures
Type
Example
Gloss
1. NN>N
2. VV>V
3. AA>V
4. AA>N
5. VV>N
6. Num+Num >A
jia-bing
gong-ji
kong-ju
xian-liang
xue-wen
san-wu
armor-weapon, war, military
attack-assault, to attack
fear-dread, frightened
able-virtuous, worth man
study-inquire, knowledge
three-five, a few
Type
Example
Gloss
1. NN > N
2. AN > N
tian-zi
xiao-ren
3. VN > N
4. VV > N
5. NV > V
6. AV > V
7. AV > N
8. PN > P
9. Num +N > N
qi-ren
fu-xing
cao-chuang
yan-ju
xian-sheng
wu-zi
bai-xing
heaven-son, Emperor,
small-person, a person of low
position
beg-person, beggar
help-travel, entourage
grass-crate, to initiate
confortable-live, to relax
early-born, sir, teacher
my-sir, you
hundred-names, people
II. Subordinate structures
54 Prosodic word in Classical Chinese
Given the different coordinating and subordinating structure types, the argument for naturally occurring phrases can be tested by assuming that if there are
more types of subordinating structures, there would be more occurrences of
subordinating compounds, and if there are fewer types of coordinating structures, there would be fewer occurrences of coordinating compounds. This is
because everything else being equal, more structure types will produce more
total occurrences of that structure, and vice versa. This prediction is borne out
as seen in Cheng’s (1981:112) statistical data given in Table 2.4, repeated here
as Table 2.7. (“Total Comp” words, “CC” stands for Coordinating Compound
words, “MH” stands for Modifier Head compound words):
Table 2.7 Percentage of CC and MH compounds in Confucius (c. 550 BC)
Comp
180
Total
CC
%
Total
MH
%
48
26.7
67
37.2
In Table 2.6, we have seen that there were more structure types of the subordinating than of the coordinating variety. From Table 2.7, we see that there
are more instances of subordinating than of coordinating structures. The correlation between the number of structure types and the number of instances of
that structure can be seen clearly in Table 2.8. (“CC” stands for Coordinating
structure and “MH” for Modifier Head structure.)
Table 2.8 Number of structure types vs. number of structures
Instances for CC and MH compounds
Structure types
CC
MH
6
9
Structure instances
40%
60%
48
67
42%
58%
The 40% versus 60% of structure types closely correlates with the 42% versus
58% of instances of coordinating and subordinating structures, respectively. The
correlation supports our contention that if there are more types of disyllabic
forms (phrases or compounds) formed by subordinating structures, and vice
versa. Given this, a reasonable explanation for the correlation is to assume that
disyllabic feet originated by making use of naturally occurring phrases, and that
compounding started from natural phrases as well.
Although the correlation between the number of structure types and the
frequency or their occurrence supports the claim that disyllabic feet in Classical Chinese were realized on naturally occurring phrases, this does not necessarily mean that disyllabic forms did not originate from phrases that were
created for prosodic purposes, because it is not yet clear what the structure of
phrases created expressly for the prosody would be. Since both structures, the
subordinating and the coordinating, can form two-word phrases equally well,
Prosodic word in Classical Chinese
55
both structures can serve the need for disyllabicity. As a result, if the notion of
“phrases created for prosody” is not specified, there would be no judgment
on the second part of the hypothesis that disyllabic combinations (phrases or
compounds) in Classical Chinese originated from naturally occurring disyllabic
phrases, rather than from those expressly created for the prosodic requirement.
To resolve this difficulty, Feng (1997a) suggested that coordination structures
can be considered structures which are created expressly for the purposes of
prosody. This is because the coordinating structure exhibits special syntactic
and semantic properties which the subordinating structure lacks; that is, with
or without part B, the semantic interpretation of A in a [A+B] coordinating
structure would always be approximately the same. Compare:
61. a. Subordinating
Tia-n-zı̆ 天子(‘Heaven’s son’) ≠zı̆ (‘son’)
Qı̆ rén 乞人(‘beg-person, beggar’) ≠rén (‘person’)
b. Coordinating
Kŏng-jù 恐懼 (‘fear-dread; frightened’) = Kŏng (‘fear, frightened’)
Kŏng-jù 恐懼 (‘fear-dread; frightened’) = Jù (‘dread, frightened’)
Go-ng-jı- 攻擊 (‘attack-assault, attack’) = Go-ng (‘attack’) = jı- (‘attack’)
Zhàn-dòu 戰鬥 (‘warring-tussle’) = Zhàn (‘fight’) = Dòu (‘fight’)
Sha--lù 殺戮 (‘kill-kill’) = Sha- (‘kill’) = Lù (‘kill’)
Subordinating structures are not as flexible as coordinating structures in their
ability to form disyllabic combinations out of monosyllabic words without affecting the semantic interpretations of the phrase. In other words, the subordinating
structure cannot be freely used without changing the original meaning of the
phrase in which it occurs. However, the coordinating structure can do this easily
by simply adding a synonym to the original monosyllable verb, noun or adjective
in any position of a sentence without changing the basic syntactic structure and
meaning of that sentence.This, as we have seen before, is what Zhao Qi did in his
Menzi Zhangju (e.g., (17)). Given this analysis, it follows that the coordinating structure has an advantage over subordinating structures in creating disyllabic phrases.
If the coordinating structure is the structure by which phrases could be created expressly for prosodic purposes, and if as we have argued before, it is more
likely that naturally occurring phrases would bear two-syllable feet, than that
coordinating structures would be created purposely for prosody, we would then
predict that there must be statistically more naturally occurring disyllabic phrases
(i.e., more subordinating phrases) than coordinating disyllabic forms in the earlier
stages, because it requires less effort to make use of naturally occurring phrases
than to create new ones.This is also borne out as seen in Table 2.7.There were 67
tokens of subordinating structures, but only 48 tokens of coordinating structures.
If the coordinating structure is used to create disyllabic phrases, and if the
creation of disyllabic forms is required only when the disyllabic foot became
more dominating, we would further expect that a reverse situation would occur
in the language, i.e., there would eventually be more disyllabic combinations
56 Prosodic word in Classical Chinese
that were formed by coordinating structures than by subordinating structures,
because when the prosodic requirement becomes stronger and stronger, making use of naturally occurring phrases would not be efficient and productive,
so the phrases created for prosody would come to dominate in late stages. This
analysis receives support from Cheng’s (1981:112; (1985:337) statistical data
given in Table 2.9. (“Total Comp” = Total compound words, “CC” stands for
Coordinating compounds and MH for Modifier Head compounds).
Table 2.9 Percentage of CC and MH compounds in Confucius, Mencius and Lunheng
Chronology
Text
Comp
Total
CC
Total
%
Total
MH
%
c. 550 BC
c. 300 BC
c. 100 AD
Confucius
Mencius
Lunheng
180
333
2,088
48
115
1,401
26.7
34.5
67.24
67
100
517
37.2
30
24.76
Table 2.9 shows that making use of naturally occurring phrases was replaced
by coordination as a way to meet the prosodic requirement. Since coordinating
structures have certain productivity advantages over subordinating structures
in creating disyllabic forms, coordinating word structures came to dominate in
the later stages.
The theory presented here explains why some compounds undergo a process of de-phrasalization (making use of naturally occurring phrases), while
others (created for prosodic requirement) do not. It also explains why there
were more subordinating compounds in earlier stages than later on, and why
compounds created for disyllabicity were mostly found at the later stage (most
examples of this type given by Cheng (1981) are from Han Feizi, c. 230 BC).
Secondly, the theory presented here also explains why SP, VO and VR structures are disfavored structures for forming compounds as seen in section 2. 2. Let
us consider the VO construction first.The reason why VO compounds were very
rare has to do with sentence prosody. As seen in Chapter 4, the sentential normal
stress (i.e., the NS seen above) in SVO languages such as English (Liberman and
Prince 1977) and Chinese (Chao 1968) generally falls on the right-most element
of a sentence (e.g., (45), and note 23). Since VO phrases in Classical Chinese
frequently appear at the ends of sentences (Feng 1995), the object of the verb in
a sentence will often be the target of the nuclear stress. As seen in section 2.5.4,
according to Liberman and Prince’s relative prominence principle (1977), a strong
node must be licensed by a weak node.Therefore, a single node alone cannot realize the stress. Since one syllable cannot serve as a branching node in a prosodic
structure, another syllable must be attached to it to form a disyllabic foot in order
to realize stress. Thus if the object is a monosyllabic word, that word must attach
to the preceding verb to become a part of a foot in order to realize the stress.
However, when the VO predicate becomes a foot and the primary stress
has been realized upon it, the VO foot must fulfill the requirement assigned
by nuclear stress. As a result, a VO structure is bound with sentential stress in a
Prosodic word in Classical Chinese
57
sentence. In other words, the nuclear stress on VO structures will always require
the VO to be a verb with an object in a sentence.28 Hence it is difficult for them
not to serve as the main predicate of the sentence, and thus it is hard for dephrasalization or lexicalization to take place. This explains why there are hardly
any VO verb compounds in Classical Chinese. Under this analysis, the way for a
VO combination to become a compound is for it to avoid acting as a predicate
of the sentence. This is achieved by changing its part of speech, i.e., acting as a
noun, such as sı--mă 司馬 ‘control-army, general’, which is precisely what has
been observed in the literature.
As for VR, since the verb-resultative complement structures are a later development in the language (starting from the Han dynasty, see Jiang Shaoyu, Cao
Guangshun (2005) and Feng 2014), it is no surprise that VR compounds are
rare (if not none) before the Han. In addition, SP compounds are even rarer,
simply because there are hardly any SP phrases in the language,29 partly due to
the fact that the subject in Classical Chinese is often dropped.
2.7. Conclusion
To elaborate the main points in this chapter: First, we have shown that twosyllable (compound) words in Classical Chinese appeared in large numbers
during the Han dynasty because of the advent of a disyllabic prosodic foot
structure during that period.
Second, because the earlier, bimoraic, monosyllabic foot could no longer
be supported by a syllable structure that had undergone simplification following the loss of consonant clusters and syllabic-final consonants, the syllabically
based Foot Formation Rule (FFR) will consequently emerge from the phonological change of Old Chinese. This is also because the newly developed
tone-bearing syllables can no longer naturally and systematically differentiate
mora-weights between syllables that contain fewer numbers of moras after syllable simplification, inevitably weakening the moraic foot system of Archaic
Chinese, giving rise to a new type of prosodic structure (i.e., the disyllabic foot
formation) in Medieval Chinese.
Furthermore, given the fact that Classical Chinese is basically a monosyllabic
language, the Word Formation Rule (58) is thus derived from the Prosodic
Hierarchy and the Foot Binarity Principles adopted in Chapter 1. The theory
presented here requires that the prosodic integrity of being a single foot and
the syntactic relation of being a phrase interact to cause PrWd, Idiomatized
PrWd and compounding in Classical Chinese: syntax determines the structural
relations between each elements in a compound, and prosody determines the
metrical size of that compound; therefore, a newly developed prosodic morphological system was born.
Under this analysis, the functional approach that the increase of compounds
around the time of the Han dynasty was due to a decrease in the number of
phonologically distinct syllables is no longer attractive, and problematic. We
argue against this functional hypothesis, because it is difficult for this analysis to
58 Prosodic word in Classical Chinese
explain the structural mechanism of the morphological development of compounding and most importantly, the functional analysis cannot account for the
facts that Polar Compound (i.e., dòng jìng 動靜 ‘active-quiescent, activity’) and
Partial Compound (pia-nyì fùcí 偏義複詞using one meaning of the two parts,
i.e., sì zhu- shì-cháo 肆諸市朝 kill it-at market-(imperial) court ‘to execute him
at the market [cf. Execution is not allowed in the imperial court]’) are highly
counter-functional but quite productive at that time.
It has been shown that using this prosodical analysis to account for the
development of classical compounding, we have explained a wide range of
phenomena, such as why there are more Modifier Head compounds than coordinate compounds at earlier stages of compounding, and why the reverse situation occurs later on, i.e., more coordination compounds than Modifier Head
compounds.
The argument made here is quite different from the traditional analysis in
many aspects. First, in the traditional analysis, the only connection between
phonological change and compounding is that phonological change resulted in
more homophones, causing the development of compounds (e.g., 2.4.1). The
present study took a new look at phonology and compounding from a prosodic
point of view. By taking prosody into account, we reached a new understanding
of the phonological change in the development of compounding.
Secondly, the importance of “Foot” has been recognized in the literature for
quite a long time (e.g., Guo 1938; Chen 1979; Shih 1986) and until recently it
has been made a connection between the Foot Formation Rule (Chen 1979;
Shih 1986) and the development of disyllabicity. Historically, linguists (for
example, Guo 1938) had believed that disyllabicity is merely a stylistic device,
and that the disyllabic foot occurred throughout the history of the Chinese language.The present study has made an attempt, following Feng (1997b, 2015), to
motivate a Foot Formation Rule based on the phonological system of Classical
Chinese. It is argued that the Foot Formation Rule was established during the
Han dynasty based on characteristics of syllable structure and the development
of the four-tone system.
Third, compound words are traditionally known as syntactic words in Chinese (e.g., Chao 1968). In this chapter, it is shown that compound words are
not merely syntactically structured, but also prosodically motivated. As a result,
the so-called compounds in Classical Chinese can naturally be divided into two
categories: a word category and a phrasal category, and both are listed in the
dictionary. The former are compounds based on lexicalization (or a categorychanging rule, cf. sı--mă 司馬 ‘charge-military’, ‘general’ as in (4d)). The latter
are Idiomatized PrWds based on their frequency of usage (cf. yı--cháng 衣裳
‘shirt-skirt’, ‘clothes’; jiă-bı-ng 甲兵 ‘armor-weapon’, ‘military’). It is also possible
that some items can be listed twice in the dictionary, once as a lexical word (cf.
tia-n-zı̆ 天子 ‘Emperor’), once as an idiomatic item (cf. tia-n-zı̆ ‘Heaven-Son’,
tia-n zhı- zı̆ yě 天之子也 ‘Son of the Heaven’).
Strictly speaking, Idiomatized PrWds are neither (free) phrases nor words, but
are idioms created by the prosodic system and fixed in usage, exhibiting special
properties: they are listed in the dictionary, used as lexical items, bear the same
Prosodic word in Classical Chinese
59
prosodic shape as a compound word, but still retain some phrasal properties yet.
Therefore, Idiomatized PrWds constitute an intermediate category between
free phrases and words in the morphological system of Classical Chinese (as
seen above) and in Modern Chinese, as shown in the following chapters.
Notes
1 The term Classical Chinese used here covers the language from the Warring States
Period (500 BC–200 BC) to the Han dynasties (206 BC–220 AD).
2 Ju-n-chén is not a compound by the semantic criterion given in (8) in contexts such as the
following:
3
4
5
6
7
8
9
(i) 君以計畜臣， 臣以計事君， 君臣之交, 計也。
Ju-n yı̆ jì xù chén， chén yı̆ jì shì ju-n， ju-n-chén zhı- jia-o, jì yě.
Monarch use trick treat official official use trick serve monarch, monarch and official
’s relation trick prt.‘The monarch uses tricks to gain officials and officials use tricks to
serve the monarch, the relations between them are nothing but tricks.’ (Hanfeizi. Shixie)
Huang (1984) also proposed criteria for Modern Chinese compounds. His criteria are
based on the Lexical Integrity Hypothesis which says, roughly, no phrasal structure rule
may apply to a lexical item, and the Phrase Structure Constraint which requires, roughly,
that no two constituents appear after the last verb. The Lexical Integrity Hypothesis
works (but not completely for Classical Chinese as we can see below), but the Phrase
Structure Constraint will not apply to Classical Chinese, simply because two constituents are allowed to appear after the verb.Therefore, although Huang’s criteria are important for Modern Mandarin Chinese, the Phrase Structure Constraint is not relevant to
Classical Chinese.
Derivative compounds are different; see below.
In traditional philology, this has sometimes been called lián lèi èr jí 連類而及 ‘bring two
words of the same kind together’, which means A is added to B, because it is the same
semantic category. In this case, usually one part of AB functions as a dummy placeholder,
which has no semantic interpretation at all.
The term Syntactic Word refers to compounds which are found according to syntactic
relations such as Subject + Predicate (SP), Modifier + Head (MH),Verb + Object (VO),
Verb + Resultative complement (VR) and Coordinate Constructions (CC).
The term “others” refers to meaning specializations, such as tia-n-xià (sky below, ‘the
Emperor’): “below the sky” → “all below the skies” → “the world of men” → “society”
→ “the Emperor”.
Of course, the best way to study compounding in Mencius zhangju may be to list all of
the monosyllable words in Mencius that have been translated into two-syllable combinations in Zhao Qi’s commentary, i.e., to provide an exhaustive listing of the “one-to-two”
notes. However, since time does not allow for such an investigation, I will analyze two
chapters of Mencius Zhangju, namely, the Liang Huiwang Shang and the Gongsun Chou
Xia. These two chapters (c. 300 BC) constitute nearly 15% percent of the entire book.
This 15% sample size is sufficient to postulate (1) a different proportion of compounds
in the Pre-Qin period and the Han dynasty and (2) the basic linguistic properties of
compounds in these two periods.
Although the no-open-syllable hypothesis for Old Chinese has been questioned by
scholars (see Norman 1988 and Baxter 1992), there are scholars who accept this hypothesis, such as, Lu Zhiwei (1947), Li Fang-kuei (1980), Ting Pang-Hsin (1979), and Yu
Naiyong (1985). Most importantly, as argued by Ting (1979) and illustrated by Yu (1985),
syllable structure was clearly more complex in Old Chinese than in Middle Chinese.
In this chapter, I adopt Ting’s hypothesis that the basic (minimal) syllable structure of
Old Chinese is CVC. Note that even though not all syllables in Old Chinese are CVC,
most scholars agree that the majority of syllables in Old Chinese had a minimal CVC
60 Prosodic word in Classical Chinese
10
11
12
13
structure. If this is so, the theory developed in this chapter can still be held without
assuming the strong form of the ‘no-open syllable’ hypothesis as we will see below.
This is why Baxter introduces the term ‘pre-initial’ for first segment of initial clusters
(*s- of *sk-) and ‘post-coda’ for final segment of syllable-final clusters (*-s of *-ks), for
Old, but not Middle, Chinese (see Baxter 1992:7).
For example, yú 愚 ‘foolish, stupid’ and yú 虞 ‘anxiety, worry’ are phonologically different in Old Chinese, but they became homophones in Middle Chinese, as did jı-ng 京
‘city’ and jı-ng 驚 ‘surprise’; see Wang 1980.
The term “coda” refers to segments immediately following the main vowel; and “postcoda” refers to final segment of syllable-final clusters.
The distinctive function of the new tone system can be seen clearly from the fact that
the number of etymological words which are distinguished by tonal differences (for
example, Level Tone of nouns cognately related with Departing Tone of verbs – the
change of category from noun to verbs [Mei 1980]), dramatically increased during the
late Han dynasty. For example (taken from Chou 1962:54):
(i) Noun (Level Tone) Verb (Entering Tone)
冠 kuan ‘cap’ 冠 kuan ‘to cap’
衣 jei ‘clothing’ 衣 jei ‘to wear (clothes)’
14 In Mencius, shi-chao can also be used to mean only shi (market) but not shi chao (market and imperial court):
(i) 撻之于市朝。
Tà zhı- yú shì-cháo.
Whip him at market-court
‘Whip him at the market.’
15 There are more examples of this type (see Gu Yanwu [1613–1682 AD], Ri­zhi-lu,
Juan.27)
(i) 擅兵而別，多他利害。
Shàn bı-ng ér bié，duo- ta- lì-hài。
Take army and leave, more other benefit-harm.
‘Take the army and leave, there will be more harm (to use).’ (Shiji. Wuwang Bi Zhuan)
(ii) 生女不生男，緩急無可使者。
she-ng nǚ bù she-ng nán，huăn-jí wú kě shı̆ zhě.
Born female not born male, unhurried-hurried no can use prt.
‘If one has only girls but no boys, there is no help for urgency.’ (Shiji. Canggong Zhuan)
(iii) 先帝嘗與太后有不快，幾至成敗。
Xia-n dì cháng yŭ tài hòu yŏu bù kuài，jı- zhìchéng bài.
Late Emperor before with Queen have unhappy, almost cause cheng-bai.
‘The late Emperor often had a fight with the Queen, it almost causes a failure.’ (Houhanshu. Douhe Liezhuan)
In Modern Chinese, there are also compounds of this type. For example:
(iv) 他要是有個好歹，孩子怎麼辦？
Ta- yàoshi yǒu gè hǎodǎi, háizi zěnme bàn?
She if have one good-bad, children how do
‘If she has a disaster, what about her children?’
16 As argued in Feng (1991, 2009), the development of the four-syllable poetic line (i.e., the
[2+2] rhythmic pattern during the Zhou Dynasty, 1000 BC) to a five-syllable poetic line
(i.e., a [2+3] rhythmic pattern after the Han dynasty, c. 100 AD) parallels to the change
of prosodic structure of the language.
17 Along the lines of Chen’s nasal attrition (1975), Wang (1993) proposed the reduction
of the post-nuclear consonant (n, ng): from a consonant into an approximant, forming a
part of the V. That is, the nasal endings of the syllable rime are all [-consonantal], and can
be viewed as part of a diphthong. As a result of her analysis, Beijing Mandarin syllables
Prosodic word in Classical Chinese
18
19
20
21
61
are all arguably CV, where the V covers both single vowels and diphthongs, which is
similar to Swahili in the sense that “Swahili syllables are coda-less – they can only end
in vowels.” (See William O’Grady, John Archibald, Mark Aronoff, and Janie Rees-Miller.
Contemporary Linguistics: An Introduction. New York: Bedford/St. Martins: p. 309)
Regarding the medial segment in Old Chinese, here I would like to claim that the
medial is part of the onset, based on recent analyses by in which the prenucleus (i.e., y, w,
medials) in Mandarin Chinese is not analyzed as part of the rhyme. Therefore, whether
the syllable contains a medial or not, the syllable weight.
One may argue that although Arabic is one of the languages that is sensitive to the prosodic weight of syllables, it does not mean that (Old) Chinese is also sensitive to prosodic
weight. However, as shown in 2.5.4, Archaic Chinese was indeed a prosodic-weightsensitive language (see also Feng 2013).
At this point, I should point out that the assumption that the CVC syllable structure
of AC is capable of forming an independent foot does not mean that the replacement
of bimoraic one-syllable feet by two-syllable feet is an all-or-none operation, i.e., it is
unlikely that one-syllable feet suddenly were all considered ill-formed and two-syllable
feet were immediately dominant. What seems natural is that the phonological basis for
the monosyllabic foot was lost step by step and monosyllabic feet became more and
more disfavored, while disyllabic feet became more and more common and dominant,
the result of an increase in certain environments for favored disyllabic elements and
a corresponding decrease in the same environments for disfavored monosyllabic ones
(Kroch 1989).This follows because the syllable structure reduction in OC and the ensuing four-tone system in MC actually took a quite long time to be finally completed
(probably by the late Han, see Xu 1996:269). Nevertheless, the unacceptability of the
monosyllabic foot can be evidenced clearly from both Archaic Chinese as seen below
and Modern Chinese as seen in the following chapters of this book.
One may argue that since diphthongs in Chinese (Middle Chinese and Modern Mandarin) can also be analyzed as consisting of two moras, a syllable that contains a diphthong
can still be a bimoraic foot even if the coda is lost. However, I will not consider diphthongs in Chinese to be able to form a standard foot using long vowels as they can in
Archaic Chinese. The reasons are as follows: first, there is no evidence of a phonological
contrast between long vowels and short vowels in Modern Chinese of all dialects, therefore there is no evidence to show that diphthongs are distinctively longer (or heavier)
than monophthongs in general. Secondly, it is well known that Mandarin syllables are of
the same length for single rhymes (monophthongs) and compound rhymes (diphthongs)
(see Duanmu 1990, and Wang 1993); therefore, if diphthongs are considered as long
vowels so that they can form a bimoraic foot, then monophthongs must also be considered as being able to form a foot, because there is no length difference between these
two types of syllables. However, it has been widely recognized in the literature (Chen
1979, Shih 1986; also see examples given in note 19), that there are clear prosodic contrasts between two-syllable and one-syllable units in poetic prosody (Chen 1979). Also,
a monosyllabic word must be grouped with another foot in the Tone Sandhi domain
defined by Foot Formation which normally contains at least two syllables (Shih, 1986).
This contrast is also observed in syntactic structures as discussed in 2.5.4 below. On
the other hand, there is no prosodic contrast between diphthongs and monophthongs
in the language. Therefore, if we consider a minimal foot as being formed by two syllables, the prosodic and syntactic properties of one-syllable and two-syllable units can be
captured. If, on the other hand, a monosyllable is considered a normal foot based on an
analysis that monophthongs consist of two moras, one cannot explain why diphthongs
do not differ from monophthongs. In addition, a significant generalization about the
prosodic properties of one-syllable and two-syllable units is lost. Therefore, no matter
how one analyzes diphthongs, prosodically speaking, diphthongs must be considered
equivalent to monophthongs in terms of prosody, and both lack the ability to form a
foot (for more arguments on this and related questions, see Feng 1995:246–252, 2000,
2015).
62 Prosodic word in Classical Chinese
22 There may be an alternative account for how to motivate the disyllabicity (or the Foot
Formation Rule [52] given below) from the phonological changes (e.g., 2.3.1 and 2.5.1)
in Old Chinese. San Duanmu has suggested to me that the incapability of Mandarin to
form an independent foot with only one syllable is due to the tonal system of the language (personal communication through e-mail). If this is so, according to the hypothesis that disyllabic feet were newly developed in Classical Chinese and the fact that the
tonal system followed the loss of final consonants, the development of disyllabicity could
also be attributable to the development of the tonal system in Classical Chinese. As mentioned above (2.3.1), the tones of Middle Chinese were developed from Old Chinese
codas and post-codas: *-s > Departing Tone; and *-ʔ > High-rising Tone. By the time
of the Han period, the tonal system was partially (if not completely) established (see note
13). Given this fact, if one syllable cannot form an independent foot in a tonal language
in general, then the development of the tonal system would be another factor to motivate the Foot Formation Rule given in (52). Nevertheless, the tonal­based account also
supports my analysis for the establishment of the Foot Formation Rule.
23 See either Duanmu’s Non-head Stress Rule (1990), or Cinque’s (1993) hypothesis that
phrasal stress is assigned universally as follows: in [XP Y] or [Y XP], stress goes to XP,
or the syntactic complement. The Sentence Prosodic Rule in (49) given below can be
derived from Duanmu’s and Cinque’s hypothesis: that is, within a VP, if the language is
SVO, the Sentence Normal Stress falls to the right of the verb, i.e., the complement of
the head of the VP.
24 Note that in the surviving SOV structure hé-zuì zhı- yŏu, zhı- yŏu ‘it-have’ can never be
separated. This indicates that zhı- in hé-zuì zhı- yŏu must be a pro­nominal clitic form
cliticized onto the verb (e.g., zhu- is a fusion form of zhı--yú ‘it at’ [Z =f ]).
25 At this point, one may argue that the nonexistent structure of *[[hé-zuì] yŏu] is not
due to whether a monosyllable can be a foot or not, but to the contrast between two
syllables hé-zuì versus one syllable yŏu. In other words, it might be argued that a foot
that consists of fewer syllables cannot compete with a foot that contains more syllables.
However, note that a disyllabic foot is able to compete with a trisyllabic foot, as seen in
(ia) and (ib) below:
(i) a. 吾何邇封之有？《左傳·昭公九年》
Wú hé ěr fe-ng zhı- yŏu？
I what near-fiefdom it have
‘What near fiefdom do I have?’ (Zuo.Zhao 9)
b. 吾子相之，老夫抱之，何幼君之有？《公羊傳·成公十五年》
Wú zı̆ xia-ng zhı-， lăo fu- bào zhı-, hé yòu ju-n zhı- yŏu？
You assist him, I carry him, what young prince it have
‘You assist him; I carry him; what kind of young prince do we have?’ (Gongyang.
Cheng 15)
In (ia), hé ěr-fe-ng zhı--yŏu is the last phrase, and in (ib) hé yŏu-ju-n zhı--yŏu forms
the last phrase. According to the SPR (Sentence Prosodic Rule), in both cases the
left node X contains three syllables he er-feng (what near­fiefdom) or he you-jun
(what young-prince), while the right node Y contains only two syllables zhi-you.
Yet, unlike (43), (ia) and (ib) are grammatical.The contrast between (43) and (ia-b)
is illustrated as follows:
VP
*VP
43. a.
X
[σ
σ]
Y
|
[σ]
X
[σ
σ σ] [σ
Y
σ]
Prosodic word in Classical Chinese
63
This strongly suggests that two-syllable units behave differently from one-syllable units.
Given the different prosodic behaviors between monosyllabic units and disyllabic units,
the argument for the one-syllable foot can no longer be held. The fact is that a standard
foot can always stand alone, but one syllable is incapable of doing so, as exemplified in
(40). It follows that a one-syllable unit, unlike a two-syllable unit, cannot form a standard
prosodic word.
26 In Classical Chinese, word order was the fundamental means for indicating grammatical
relations between the elements of a sentence. Therefore, combinations of words must be
constrained by phrase-structure rules of the language.
27 Note that there is no theoretical reason to expect that all X0-level constituents would be
semantically non-compositional, nor any reason to expect that all X’- or X‘’-level constituents would be semantically compositional (see Liberman and Sproat 1992).The semantic
criterion (8) is unsatisfactory in this connection. The Foot Formation Rule, on the other
hand, encourages the development of disyllabic lexical units given the theory presented
here. The Word Formation Rule, a formal constraint for prosodic words, is theoretically
motivated. Therefore, compounds in Chinese can be formally derived by (58) alone.
28 The iambic stress on VO compounds in Modern Chinese sometimes causes speakers to
treat them as phrases (Chao, 1968). For example:
A. wŏ hěn da-n-xı-n ta-.de jia-ng lái
I very bear-heart he/she future
‘I worry about his/her future.’
B. Nî da-n shénme xı-n a!
You bear what heart prt.
‘What on earth are you worrying about!’ It has been suggested (Chao 1968:431; Feng
1995:107) that sentence stress can ionize an iambic compound into a phrase in Modern
Chinese in certain contexts.This analysis supports the assumption that sentence stress on
VO structures causes them to be construed as phrases.
29 It is also possible, as Feng (1993) has argued, that in Classical Chinese there was a pause
between the subject and the predicate in declarative sentences. If it is so, the pause may
block the natural combination of an SP structure from being a foot; hence it is harder for
the SP structure to become a compound than for other structures, given the hypothesis
that compounds must be constrained by the prosodic integrity of being one foot.
References
Barale, Catherine. 1982. A Quantitative Analysis of the Loss of Final Consonants in Beijing Mandarin. Doctoral dissertation. Philadelphia: University of Pennsylvania.
Baxter, William H. 1992. A Handbook of Old Chinese Phonology. Berlin: Mouton de Gruyter.
Behr, Wolfgan. 2004. “Language Change in Premodern China-Notes on Its Perception and
Impact on the Idea of A ‘constant way’ ”. In Achim Mittag and Helwig-Schmidt-Glintzer
(eds.) Ideology and Historical Criticism (Special Issue of Historiography East and West). Leiden:
E.J. Brill, 13–51.
Benedict, Paul K. 1972. Sino-Tibetan: A Conspectus (Contributing editor: James A. Matisoff ).
Cambridge: Cambridge University Press.
Bodman, Nicholas C. 1980 “Proto-Chinese and Sino-Tibetan: Data Towards Establishing the
Nature of the Relationship”. In F. Van Coetsem and Linda R. Waugh (eds.) Contributions
to Historical Linguistics. Leiden: E.J. Brill, 34–199.
Chao, Yuen-Ren. 1968. A Grammar of Spoken Chinese. Berkeley: University of California
Press.
64 Prosodic word in Classical Chinese
Chen, Matthew Y. 1975. “An Areal Study of Nasalization in Chinese”. Journal of Chinese
Linguistics 3: 16–59.
Chen, Matthew Y. 1979. “Metrical Structure: Evidence From Chinese Poetry”. Linguistics
Inquiry 10: 371–420.
Cheng, Xiangqing.1981. A Study of Disyllabic Words in Pre-Qin 先秦漢語雙音詞研究. In
Cheng, Xiangqing (eds.) Studies of Pre‑Qin Chinese. 1882. Shandong Education Press.
Cheng, Xiangqing. 1985. “Lunheng Fuyinci Yanjiu 論衡複音詞研究”. In Cheng Xiangqing (ed.) Studies of Pre-Qin Chinese 先秦漢語研究. Jinan: Shandong Education Press,
262–340.
Chou, Fa-kao. 1962. A Historical Grammar of Ancient Chinese 中國古代語法. Taipei: Academia Sinica, Institute of History and Philology, monograph 39.
Cinque, Guglielmo. 1993. “A Null Theory of Phrase and Compound Stress”. Linguistic
Inquiry 24.2: 239–297.
Dobson, W. A. C. H. 1959. Late Archaic Chinese: A Grammatical Study. Toronto: University of
Toronto Press.
Dong, Tonghe. 1954. The History of Chinese Phonology 中國語音史. Taipei: Zhōnghuá Wénhuà Chu-bǎn Shíyèshè Press.
Duanmu, San. 1990. A Formal Study of Syllable, Tone, Stress and Domain in Chinese Languages.
Doctoral dissertation. MIT, Cambridge, MA.
Feng, Shengli. 1991. “Prosodic Structure and Word Order Change in Chinese”. The PENN
Review of Linguistics 15: 15–21.
Feng, Shengli. 1993. “The Copula in Classical, Chinese Declarative Sentences”. Journal of
Chinese Linguistics 21.2: 211–311.
Feng, Shengli. 1994. “Stress Shift and Object Post-Posing in Early Archaic Chinese
論上古漢語的重音轉移與賓語後置” Language Research 26: 79–93.
Feng, Shengli. 1995. Prosodic Structure and Prosodically 語言研究 Constrained Syntax in Chinese.
Ph.D. dissertation. Pennsylvania: UPENN.
Feng, Shengli. 1997a. “Prosodic Structure and Compound Words in Classical Chinese”.
In Jerome Packard (eds.) New Approaches to Chinese Word Formation. Berlin: Mouton de
Gruyter, 197–260.
Feng, Shengli. 1997b. Interactions Between Prosody, Morphology and Syntax in Chinese
《漢語的韻律詞法與句法》. Beijing: Peking University Press.
Feng, Shengli. 2000. Prosodic Syntax in Chinese 漢語韻律句法學. Shanghai: Shanghai Education Press.
Feng, Shengli. 2005. Prosodic Grammar in Chinese 漢語韻律語法研究. Beijing: Peking University Press.
Feng, Shengli. 2012. The Syntax and Prosody of Classifiers in Classical Chinese. In Xu, Dan
(ed.) Plurality and Classifiers Across Languages of China. Berlin: Mouton de Gruyter, 67–99.
Feng, Shengli. 2013. New Insights in Phonological Study on Archaic Chinese 上古音韻研
究的新視角. In Feng, Shi and Gang, Peng (eds.) 大江東去 —— 王士元教授八十
歲賀壽文集. City University of Hong Kong Press, 71–84.
Feng, Shengli. 2014. “Historical Syntax of Chinese”. In Huang C.-T. James, Li Y.-H. Audrey
and Andrew Simpson (eds.) The Handbook of Chinese Linguistics. Chichester, UK: Wiley
Blackwell, 537–598.
Feng, Shengli. 2015. “Five Comments on ‘A Chinese Phonological Enigma’ by Professor
Geoffrey Sampson”. Journal of Chinese Linguistics 43: 734–739.
Feng, Shengli. 2016. Question-and-Answers About Prosodic Grammar in Chinese
漢語韻律語法問答. Beijing: Beijing Language and Culture University Press.
Prosodic word in Classical Chinese
65
Goldsmith, John A. 1990. Autosegmental and Metrical Phonology. Oxford: Basil Blackwell.
Guo, Shaoyu.1938.“The Function of ElasticWord Length in Chinese 中國語詞之彈性作用”.
Yenching Journal of Chinese Studies 燕京學報 24: 1–34.
Haudricourt, Andre. 1954. “De L’origine des tons en vietnamien”. Journal Asiatique 242:
68–82.
Harbsmair, Christopher. 2013. Zhao Qi and the History of Colloquial Chinese. Written Chinese:
the Present and the Past. Beijing: Peking University Press.
Haudricourt, Andre. 1972. ‘Problemes de phonologie diachronique.’ Langes et civilisations a tradition
orale 1. Paris: Societe pour l’Etude des Langues Africaines.
Hayes, Bruce. 1981. A Metrical Theory of Stress Rules. Doctoral dissertation. MIT, Cambridge,
MA; IULC. 1980.
Ho Dah-an. 2016. “Such Errors Could Have Been Avoided—Review of Old Chinese:
A new reconstruction By William H. Baxter and Laurent Sagart”. Journal of Chinese Linguistics 44.1: 175–230.
Hong, Bo. 1999. Worship Even Number/The Beauty of Harmony and Disyllabification
of Chinese Vocabulary 偶數崇拜/中和之美與漢語詞彙的雙音節化. In Hong Bo Jianguo Collection: Studies of Han-Tai Languages 堅果集：漢台語錐指. Tianjin: Nankai daxue
chubanshe: 160–172.
Huang, C.-T. James. 1984. “Phrase Structure, Lexical Integrity, and Chinese Compounds”.
Journal of the Chinese Language Teachers Association 19: 53–78.
Huang, C-T. James. 2005. “Syntactic Analyticity: The Other End of the Parameters”. 2005
LSA Summer Institute Lecture Notes, MIT and Harvard, Cambridge, MA.
Jiang, Shaoyu and Guangshun, Cao. 2005. Summary of Research on Historical Chinese in PreModern Times. 近代漢語語法研究綜述. Beijing: Commercial Press.
Kager, Rene. 1992. “Alternatives to the Iambic-Trochaic Law”. Natural Language and Linguistic Theory 11: 381–432.
Kiparsky, Paul. 1979.“Metrical Structure Assignment Is Cyclic”. Linguistic Inquiry 10: 421–442.
Kroch, Anthony. 1989. “Reflexes of Grammar in Patters of Language Change”. Journal of
Language Variation and Change 1: 133–172.
Labov, William. 1987. “The Overestimation of Functionalism”. In Rene Dirven and Vilem
Fried (eds.) Functionalism in Linguistics. Amsterdam: Benjamins, 311–332.
Li, Fang-Kuei. 1980. Study of Old Chinese Phonology 上古音研究. Beijing: Commercial Press.
Liberman, Mark. 1975. The Intonation System of English. Ph.D. dissertation. MIT, Cambridge, MA.
Liberman, Mark and Prince, Alan. 1977. “On Stress and Linguistic Rhythm”. Linguistics
Inquiry 8: 249–336.
Liberman, Mark and Sproat, Richard. 1992. “The Stress and Structure of Modified Noun
Phrases in English”. In Ivan A. Sag and Anna Szabolcsi (eds.) Lexical Manners. Stanford:
Stanford University Press, 131–181.
Lu, Zhiwei. 1965 [1957/1964]. Chinese Morphology 漢語的構詞法. Beijing: Science Press.
McCarthy, John. 1979. “On Stress and Syllabification”. Linguistics Inquiry 10: 443–466.
McCarthy, John and Prince, Alan. 1991. Prosodic Minimality. Lecture presented at University
of Illinois Conference the Organization of Phonology.
McCarthy, John and Prince, Alan 1993. “Generalized Alignment”. In G. Booij and J. van
Marle (eds.) Yearbook of Morphology 1993. Dordrecht: Kluwer, 79–153.
Mei, Tsu-lin. 1970. “Tones and Prosody in Middle Chinese and the Origin of the Rising
Tone”. Harvard Journal of Asiatic Studies 30: 86–110.
Mei, Tsu-lin. 1980. “Chronological Strata in Derivation by Tone-Change” 四聲別義中
的時間層次. Chinese Philology 中國語文 6: 427–443.
66 Prosodic word in Classical Chinese
Mei.Tsu-lin. 1994. Notes on the Morphology of Ideas in Ancient China. In Willard J. Peterson, et al. (eds.) The Power of Culture, 37–46.
Nespor, Marina and Vogel, Irene. 1986. Prosodic Phonology. Dordrecht: Foris.
Norman, Jerry. 1988. Chinese. Cambridge: Cambridge University Press.
Pan,Wuyun. 2000. Chinese Historical Phonology 漢語歷史音韻學. Shanghai: Shanghai Education Press.
Pan,Wuyun. 2001. “The emphatic and weakened pronoun forms in Archaic Chinese 上古指
代詞的強調式和弱化式”. In Fan Kaitai 范开泰 and Qi Huyang 齐沪扬 (eds.) Mianxiang 21Shiji Yuyan Wenti Zai Renshi 面向二十一世紀語言問題再認識. Shanghai: Shanghai Education Press, 297–313.
Pulleyblank, Edwin G. 1962. “The Consonantal System of Old Chinese”. Asia Major 9:
58–144, 206–265.
Pulleyblank, Edwin G. 1977–1978. “The Final Consonants of Old Chinese”. Monumenta
Serica 33: 180–206.
Pulleyblank, Edwin G. 2000. “Morphology in Old Chinese”. Journal of Chinese Linguistics
28.1: 26–51.
Selkirk, Elisabeth. 1980a. “Prosodic Domains in Phonology: Sanskrit Revisited”. In Mark
Aronoff and Mary Louise Kean (eds.) Juncture Anma Libri. Saratoga, CA, 107–129.
Selkirk, Elisabeth. 1980b. “The Role of Prosodic Categories in English Word Stress”. Linguistic Inquiry 11: 563–605.
Shi, Xiangdong. 2014. Discussions on Coda, Mora and Tones of Yinsheng Syllables in Archaic
Chinese 關於上古漢語陰聲音節的韻尾、韻素和聲調問題的探討. New Explorations
on Chinese Prosodic Grammar. Shanghai: Zhongxi Press, 344–66.
Shih, Chilin. 1986. The Prosodic Domain of Tone Sandhi in Chinese. Ph.D. dissertation. University of California, San Diego.
Ting, Pang-Shin. 1979. “The Syllable Structure in Archaic Chinese 上古漢語的音節結構”.
Bulletin of the Institute of History and Philology Academia Sinica 50: 717–739.
Wang Li. 1980. The History of Chinese Language 汉语史稿. Beijing: Zhonghua Press.
Wang, Zhijie. 1993. The Geometry of Segmental Features in Beijing Mandarin. Ph.D. dissertation.
Newark: University of Delaware.
Xu, Dan. 2006. Typological Change in Chinese Syntax. Oxford; New York: Oxford University
Press.
Xu, De’an. 1981. “Early Chinese Morphology Shown in Commentary Materials
漢語早期構詞法 以爾雅、方言同郭注的對照為例”. Journal of the Xinan Nona Institute
西南師範學院學報 3: 31–39.
Xu, Tongqiang. 1996. Historical Linguistics 歷史語言學. Beijing: The Commercial Press.
Yu, Min. 1989 [1981]. Collections ofYu Min’s Works in Linguistics 俞敏語言學論文集. Ha’erbin:
Heilongjiang People’s Publisher.
Yu, Naiyong. 1985. Study of the Old Chinese Sound System 上古音系研究 . Hong Kong: The
Chinese University Press.
Zhao, Pusong. 2014. Moraic Contrast in Archaic Chinese: A Case Study of First Personal Pronouns
Wu and Wo. PhD Dissertation. Hong Kong: The Chinese University of Hong Kong.
Zhengzhang, Shangfang. 2003. Old Chinese Phonology 上古音系. Shanghai: Shanghai Education Press.
Zhengzhang, Shangfang. 2017. “Prosody in Chinese Dialects and Old Chinese”. Studies in
Prosodic Grammar 2.1: 1–2.
Prosodic word in Classical Chinese
67
Zhengzhang, Shangfang. 2017. “The Prosodic Features in Chinese Dialects and Archaic Chinese Pronunciation”. Studies in Prosodic Grammar 2: 28–31.
Zhu, Qingzhi. 1992. A Study of the Relationship Between Buddhist Scriptures and the Vocabulary
of Middle Sinitic 佛典與中古漢語詞彙研究. Taipei: Wenjin Press.
Zubizarreta, Maria Luisa. 1998. Prosody, Focus, and Word Order. Linguistic Inquiry Monograph
33. Cambridge, MA: MIT Press.
3Monosyllabicity and
disyllabicity
The Foot Formation Rule established in Classical Chinese did not shape
the modern vocabulary into all disyllabic forms completely. Monosyllabic
words still have existed all along since Classical Chinese and in fact the
status of monosyllabic morpheme has not changed even today. How can
we explain the fact that there are disyllabic compound words demanded
by disyllabic foot formation since the Han dynasty on the one hand, and,
monosyllabic words and monosyllabic morphemes are still functioning in
the language up to now (Sproat 1998) on the other hand? Indeed, the
situation has resulted in a notorious controversy in the field of Chinese
linguistics over the syllabic property of Mandarin Chinese: whether it is
a monosyllabic or a polysyllabic language. These issues are investigated in
this chapter, before discussing the prosodic word formation in Mandarin
Chinese in next chapter.
Following Feng (1995), we propose, first, that there are two separate interacting rules responsible for the ‘monosyllabic myth’ in Chinese morphology, that
is, the Morphosyllabicity Rule (MR) operating on the morphemic level (under
the Morphosyllabic Constraint (MC) discussed in the previous chapter), and
the Foot Formation Rule (FFR) that applies at the syntactic level (phrasal rule,
see Feng 2003). Under the two-rule system presented here, it is natural to see
why there are two seemingly opposite processes in Chinese morphology: monosyllabic forms are forced to become disyllabic, while disyllabic morphemes
develop towards monosyllabicity.
3.1. Background
There has actually been a ‘war’ in the field of linguistics over the syllabic property of Chinese language/s: whether Mandarin Chinese is a monosyllabic language or a polysyllabic language. The arguments in the literature over the last
30 years have resulted in two paradoxical generalizations:
1． (i) Mandarin Chinese is a monosyllabic language.
(ii) Mandarin Chinese is a poly-(di-)syllabic language.
Monosyllabicity and disyllabicity
69
The monosyllabic approach (see, Karlgren 1920, 1949; Li 1973; Chao and Yang,
1962; Chao Y.R, 1968; Chou, F.G. 1982, and others) insists that since almost
every syllable of Chinese has a meaning, Chinese should be characterized as a
monosyllabic language. Thus, Chao writes: “The so-called ‘monosyllabic myth’
is in fact one of the truest myths in Chinese mythology” (1968:139).
The polysyllabic approach (see Kennedy, 1951; DeFrancis, 1950; Jing, 1969;
Li and Thompson, 1981 and many others) observes that more than 80% of
words in a running text or an ordinary Chinese dictionary consist of two or
more syllables.1 If the majority of words are polysyllabic, there is no reason to
consider Chinese as a monosyllabic language. As Kennedy has reasoned, we
have no record of Chinese ever having even as many as 4,000 distinct syllables,
a number which, if representing only words of one syllable, would have been
quite inadequate to represent the scores of thousands of expressions that the
Chinese with their highly sophisticated cultures must have needed to express
themselves (1964:104–118, 274–322).
Although the ‘war’ seems to be over by now, the controversy remains unresolved. In contemporary linguistic literature, one can easily take the monosyllabic approach for granted,2 while others may freely take the disyllabic property
as given.3 It seems that no one would deny that Modern Chinese is a monosyllabic language in the sense that every syllable has a meaning, while disyllabicity
is also a property in its morphology. So the problem we face is: If Chinese is
a monosyllabic language, how could its disyllabic words account for 80% of
the vocabulary? On the other hand, if disyllabicity has become a characteristic
of the language, how could it still sustain a monosyllabic property? As Tang
(1988:569) has put it “We can certainly assert that the tendency towards disyllabicity of the Chinese lexicon will be developed more rapidly. Of course, we
must realize that although the number of monosyllabic words is not great, the
frequency of their occurrences is much higher than that of polysyllabic ones.
As a result, the question of whether Chinese is a monosyllabic language or a
polysyllabic language is still difficult to decide” (My translation).
In this chapter, I will propose a two-rule system of Chinese morphology in
section 2, based on the theory of prosodic morphology developed by Mc­Carthy
and Prince (1993) and Feng (1995, 2003). I shall demonstrate that the morphological system of Chinese can be constructed by two independent rules: the
Morphosyllabic Rule (MR) and the Foot Formation Rule (FFR).4
Furthermore, I will argue, in section 3, that these two rules apply at two
different levels of the grammar: MR applies at the morphemic level, and
FFR at the phrasal (or syntactic) level. Given the two levels for the application of these two rules, it follows that the input of FFR must be the output
of the MR. Under this system, the monosyllabic property is determined by
the rule of morphosyllabicity (see 2.1) and the characteristic of disyllabicity
is derived from the rule of Foot Formation (see 2.2). Section 4 investigates
the interaction between the MR and the FFR, in which the [2-to-1] and
[1-to-2] morphological processes are discussed. Given the two-rule system
70 Monosyllabicity and disyllabicity
hypothesis, a number of theoretical consequences will be summarized in
sections 5 and 6.
3.2. Two rules in Chinese morphology
3.2.1. Morphosyllabic Rule
It has been widely recognized that morphemes in Chinese are overwhelmingly
monosyllabic. However, there are neither generalizations nor linguistic representations to consider the monosyllabicity as a morphological constraint of the language.Traditionally, linguists only gave general statements that Chinese is basically
monosyllabic language (Chou, 1982), but never considered it as a rule in Chinese
(or in human languages). Some crucial reasons, I think, are these: First, we lack a
theory of how it is possible for a language to observe a monosyllabic rule. Second,
there are disyllabic monomorphemes in everyday speech, even though they are
very rare. For example (a “.” before a syllable indicates it is neutralized, thereafter):
2. húdié 蝴蝶
yua-n.yang 鴛鴦
tángláng 螳螂
bo-.li 玻璃
pú.tao 葡萄
‘butterfly’
‘mandarin duck’
‘mantis’
‘glass’
‘grape’
However, as we will see below, these disyllabic morphemes cannot disprove
the general fact that morphemes in Chinese are overwhelmingly monosyllabic. Based on reasons that will be detailed below, Feng (1995) suggested that
the property of syllabic monomorphemic formation should be considered as
a general constraint in Chinese morphology. This constraint is stated by using
DeFrancis’ term (1986:187) as “morphosyllabicity” and formulated in (3):
3. Morphosyllabic Rule (MR) (“M” stands for morpheme and “σ” for syllable):5
M
│
σ
In Chinese, a syllable must correspond to a morpheme.
Note that the M here strictly refers to indigenous root morphemes in Chinese, thus the loan words such as bı-ngláng 檳榔 ‘areca’ that may operate pursuant to different conditions and rules.6
According to MR, the M node dominates a syllable directly, and hence a syllable corresponds to a morpheme.The mapping between an “M” and a “σ” will
immediately result in a grammatical output: one syllable with one morpheme.
According to Feng (1995, 2009), the MR can also be naturally derived from
the theory of prosodic morphology developed by McCarthy and Prince (1993).
Monosyllabicity and disyllabicity
71
More specifically, it is derived from the theory of alignment. Previous studies of
the syntax-prosody interface have proposed an edge-based theory (Chen 1987;
Selkirk 1986; among others), which proposes that the domains of sentence phonology are specified by rules like “The right/left edge of some grammatical
constituent coincides with the corresponding edge of some phonological constituent” (cf. Selkirk 1986).This rule has been defined in terms of edge alignment
(ALIGN), with the following general schema by McCarthy and Prince (1993):
4. General Schema for ALIGN
In ALIGN (GCat, GEDGE, PCat, PEdge), the GEdge of any GCat must
coincide with PEdge of some PCat, where,
GCat ≡ Grammatical Category, among which are the morphological
categories
MCat ≡ Root, Stem, Morphological Word, Prefix, Suffix, etc.
PCat ≡ Prosodic Category ≡ Mora, Syllable, Foot, PrWd, PhPhrase, etc.
MEdge, PEdge = Left, Right
McCarthy and Prince’s edge alignment schema extends the Chen/Selkirk’s
theory in two ways: the grammatical and prosodic categories subject to alignment are the word-internal morphological constituents, root, suffix, etc., and
the word-internal prosodic constituents, syllable, foot, etc.; and alignment of
different edges may also be required. It has been observed, for example, that
many languages require free-standing (nonclitic) words to be of a minimal
prosodic size (typically disyllabic or bimoraic). Subminimal items are either
barred from the lexicon or made to be free-standing through various augmentation processes. As pointed out by Kenstowicz (1993), in Australian language
Yidiny, all roots conform to a CVCV(CV) template, and therefore are minimally disyllabic.
If there is a general correlation between certain prosodic categories (mora,
syllable) and certain morphological categories (root, morpheme, words) in
human languages, then it is reasonable that the MR in Chinese might simply
be the result of the general principle of Edge Alignment.The alignment principle is parameterized in terms of a correspondence between a prosodic category
syllable and a morphological category morpheme. In other words, the MR is
a subcase of a more general principle of Edge Alignment between prosody
and morphology. Given this, the Morphosyllabic Rule can be interpreted in
terms of Edge Alignment7 between syllables and morphemes (‘M’ stands for
root morphemes and ‘σ’ for syllables, ‘[’ and ‘]’ represent left and right edges,
respectively):
5. ALIGN: [M] = [σ]
This constraint relates the prosodic category syllable to the morphological
category morpheme, demanding that they begin and end together. There are
72 Monosyllabicity and disyllabicity
some interesting linguistic consequences resulted from the ALIGN in Chinese
grammar.8
One effect to be noted here is the implication that any operation which
breaks the desired relation between the morphological and prosodic constituency of a form will be a violation of the MR, since ALIGN requires sharply
defined morpheme edges. For ALIGN to be satisfied, the morpheme final
consonant or vowel must occupy the final position in the corresponding syllable, and the morpheme initial C or V must occupy initial position in that
syllable. Consequently, a “morpheme mid syllable/consonant” will de-align a
morpheme (see McCarthy and Prince 1993:38). This explains why there is no
resyllabification in Chinese we mentioned in Chapter 1 briefly and elaborated
here in (6) as follows (where “.” marks the morpheme boundary):
6. CVC.VC
→ *(CV(CVC)
Lín.a-n 林庵
→ *Lí.nán 黎楠.
Tia-n.a-n.mén 天安門 → *Tia-.na-n.mén
Resyllabification like (6) is a phonological process by which a postvocalic
consonant of a preceding syllable occupies the onset of the next syllable (part
of a more general tendency to avoid onsetless syllables). In many languages
the onsetting of postvocalic consonants takes place even if a grammatical or
word boundary intervenes. For example, in English, a word like ‘condensation’
divides grammatically as [condens+ation] but syllabifies as [con.den.sa.tion].
However, in Chinese this type of resyllabification, i.e., the second syllable takes
the final consonant of the preceding syllable as its onset, never happens, confirming that syllables in Chinese cannot be resyllabified in ordinary situations.9
Another important implication of the constraint [ALIGN: M=σ] is that the
essential or primitive constituents for morphological operations in the language are monosyllabic morphemes or words. From a prosodic viewpoint, any
combination of two or more morphemes into one syllable will interfere with
the organization of syllables. From a morphological point of view, any association of two or more syllables that represent a single morpheme will affect
the morphological structure. Because of this, the organization of the grammar
beyond individual syllables (in the prosodic system) or individual morphemes
(in morphological system) will activate relative principles and constraints, both
prosodic and morphological.
Note further that the [ALIGN: M=σ] also entails that the following structure
given in (7) is ill-formed:
7.
*M
σ
σ
In (7), the M node dominates two syllables, and hence disyllabic monomorphemes will be generated. Obviously, if MR in (3) is a general rule in Chinese
Monosyllabicity and disyllabicity
73
morphology, structures like the one given in (7) is ill-formed. In other words,
in the morphological system of Chinese, there will be no syllable that is not
a morpheme on the one hand, and no morphemes that contain two or more
syllables on the other hand. As we will see, these two predictions are borne out
as exemplified in the following sections. In what follows, I will first provide
evidence supporting the possibility of taking the MR to be a general rule (or
constraint) in Chinese morphology.
3.2.2. Morphemization of polysyllabic words
As shown in (2), seemingly counterexamples to the MR can easily be found in
Chinese. However, as Su 蘇錫育(1989) and Sproat & Shih (1993) pointed out,
for some disyllabic morphemes such as húdié 蝴蝶 ‘butterfly’, of which the two
syllables represent only one morpheme, one of the two syllables can be used as
an independent morpheme and combine with other morphemes. For example:
8. 蝴蝶
螞蟻
螺絲
蟾蜍
螳螂
húdié ‘butterfly’
măyı̆ ‘ants’
luósı- ‘snail’
chánchú ‘toad’
tángláng ‘mantis’
‘butterfly stroke’
蝶泳 dié-yŏng ‘male ants’
雄蟻 xióng-yı̆ ‘whorl’
螺紋 luó-wén ‘toad cake’
蟾酥 chán-su- 螳臂當車 táng-bì da-ng-che- ‘antis arm stop car’
(an antis stops a car with its arm)
鴛侶yuan-lǚ ‘affectionate company’
鴛鴦 yua-n.yang
‘mandarin duck’
‘glass fiber’
玻璃 bo-.li ‘glass’
玻纖 bo--xia-n 狐狸 hú.li ‘fox’
狐臭 hú-chòu ‘fox smell’
‘cobweb’
蜘蛛 zhı- zhu- ‘spider’ 蛛網 zhu--wăng 螃蟹 páng.xiè ‘crab’ 蟹青 xiè-qı-ng ‘greenish-grey’ (color)
‘camel hair’
駱駝 luò.tuo ‘camel’ 駝毛 tuó-máo Note that if dié-yŏng is a compound formed by two morphemes, dié must
be considered as an independent morpheme. If dié is analyzed as a morpheme,
then the first part of hú.dié, i.e., hú, must also be analyzed as a morpheme. That
is, given the following diagram, if B in [B C] of (9b) and in [B D] of (9c) is a
morpheme of compound words [B C] and [B D], then A in [A B] should also
be considered as a morpheme.
9. a.
b.
c.
A B
B C
B D
This treatment is parallel to a morphological analysis of cranberry, huckleberry
and boysenberry. If berry is a morpheme, the other part of the words, i.e., cran in
cranberry, huckle in huckleberry and boysen in boysenberry must also be analyzed as
morphemes although no such forms cran, huckle and boysen exist in the English
74 Monosyllabicity and disyllabicity
lexicon. That is, the morphological theory must allow some morphemes that
are not meaningful in isolation.10 Morphemes of this type acquire meaning by
virtue of their connection with other morphemes to form words. Likewise, if
dié in dié-yŏng is a morpheme, we have a good reason to consider hú in hú-dié
to be a morpheme, even though hú may not be independently listed in the
lexicon (analogous to cran, huckle and boysen). If hú is a morpheme according
the above analysis, then the correlation of a syllable with a morpheme satisfies
the MR given in (3). In other words, as long as hú and dié are two morphemes,
they meet the morphosyllabic constraint: a syllable is directly dominated by a
morpheme. If this is the case, the so-called Butterfly-cases would not be ‘true’
exceptions to the general constraint in Chinese morphology that a syllable
must coincide with a morpheme.
This above analysis entails that as long as one syllable in a disyllabic form is
morphemized, the disyllabic word is no longer contradictive to the rule of the
MR. Indeed, Su (1989) and Sproat & Shih (1993) have provided rich documentation of a strong tendency of morphemization in Chinese that word formation
frequently picks one morpheme in a polysyllabic form to form a compound.
This process is very active in Chinese, namely, to pick up one syllable from a
polysyllabic word and make it a morpheme by combining it with other morphemes/words that already exist in the language. Note that the morphemization process happens not only to those where one part of the polysyllabic form
was used in Classical Chinese as Sproad and Shih noticed (1993:194), it happens also to loan words borrowed from other languages. For example,
10. fútè
wătè
mı̆tutèdílún
fótuó
měnggŭ
< ‘volt’
< ‘watt’
< ‘mètre’
<‘terylene’
< ‘Buddha’
< ‘Mongol’
伏特
瓦特
米突
特迪倫
佛陀
蒙古
fú ‘volt’
wă ‘watt’
mı̆ ‘meter’
dí-lún ‘terylene’, dí-mián ‘polyester fiber’
fójı-ng ‘Buddhist sutra’
měngyı- ‘Mongolian doctor’
These examples demonstrate that morphemization is a productive way of
making new compounds in Chinese morphology, which provides strong support to the grammatical function of the MR in (3). Given this, it is reasonable
to conclude that the tendency towards morphemization of a polysyllabic form
is motivated by operations of the Morphosyllabic Rule and that this rule functions actively in Mandarin Chinese.
However, there are some disyllabic morphemes that do not follow the rule
given in (3). For example:
11. a. bı-ngláng
b. níngméng
c. pánggua-ng
d. hú.lu
檳榔
檸檬
膀胱
葫蘆
‘betel nut’
‘lemon’
‘balder’
‘bottle gourd’
Monosyllabicity and disyllabicity
75
Furthermore, there are syllabic epenthesis forms in the language:
12. a. *ziIm → xún.me 尋麼
b. *tʃɣɛp → zhă.me 眨麼
‘look for’
‘blink’
The two syllables in the above examples are inseparable, and none of them has
been used to form a compound with another syllable. As a result, they are exceptions to the analysis given above. However, examples in (11) can be accounted
for in terms of loan words that are exceptional to native rules and epenthesis
syllables like the ones in (12) are naturally excluded from root morphemes.
Even if there are still a few exceptions such pánggua-ng 膀胱 ‘balder’ (medical terminology) and hú.lu 葫蘆 ‘bottle gourd’ (the neutralized second syllable
functions as a register particle), see Chapter 6 below, they can be explained in
terms of MR interacting with other rules and constraints. As we will see below,
the MR given in (3) is inoperable in the domain of operations of Foot Formation Rule (FFR) in the language. We will discuss the interaction between the
MR and the FFR below. For now, we may safely conclude that the MR applies
only to root morphemes,11 or to put it another way, only root morphemes are
sensitive to the MR, hence non-root morphemes like loan words and functional
categories, are not governed by the MR. Since the disyllabic forms in (11) and
the second syllable in (12) are not root morphemes, the MR would not see
them within this system.12 This can further be seen from the fact that in the Fifth
Edition of Xiàndài Hàny Cídiǎn 現代漢語詞典 ‘Modern Chinese Dictionary’
(2005), there is a total of 41,915 words, of which only 849, or 3%, are polysyllabic morphemes, which are arguably all non-colloquial or non-indigenous in
nature in the sense that they are technique terms or loan words either borrowed
from other languages or passed down from Classical Chinese thousands of years
ago (Shen 2007). Despite the fact that the complexity of the origins of polysyllabic words, they are neither indigenous in character, nor root morphemes in
morphology in Mandarin Chinese (see Sproat and Shih 1996, Feng 2009).
Empirically, the Morphosyllabic Rule (3) can further be tested by the fact
that polysyllabic loan words were/are often morphemized into, and thus indigenized (or nativized) as a monosyllabic morpheme in today’s morphological
process (Spoart and Shih 1996). For example:
13. fótuó
fó-jı-ng
fó-diǎn
fó-fǎ
fó-jiào
fó-xué
chéng-fó
dà-fó
huó-fó
佛陀
佛經
佛典
佛法
佛教
佛學
成佛
大佛
活佛
‘Buddha (loaned into China around the first century)’
‘Buddhist sutra’
‘Buddhist Document, Sutra’
‘Buddhist doctrine/power’
‘Buddhist teaching, Buddhism’
‘Buddhist Study’
‘become a Buddha’
‘great Buddha’
‘living Buddha’
76 Monosyllabicity and disyllabicity
Even if loan words behave differently from ordinary words morphologically in
general, the above examples show that they are still affected by the MR in Chinese
when they are used as root morphemes under the process of morphemization.
3.2.3. Anti-disyllabicity
Although disyllabicity has been believed to be a strong tendency in Modern
Chinese, anti-disyllabicity can also be observed in the language. That is, disyllabic forms of some sort tend to be reduced to a monosyllabic form again.
Chou (1982) pointed out that the reduction of the nominal suffix er must be
considered as a process of anti-disyllabicity (1982:9). The reduction of er is further elaborated in Xu (1990). He argued that Chinese exhibits a tendency for
disyllabic forms to be reduced to monosyllabic forms, by observing that, hái-ér
孩兒 ‘child-son’ is a disyllabic word formed by hái 孩 ‘child’ plus a monomorphemic nominal suffix –er, which was developed from the Medieval Chinese
word er meaning ‘son, small’. However, the second syllable er in almost all nouns
of Mandarin Chinese has been reduced to only a /r/ feature fused with the
proceeding syllable, thus yielding a monosyllabic word.This type of process can
be formulated as follows:
13.
a. Monosyllabic Word
→ Disyllabic Word → Monosyllabic Word
hái
hái-ér
háir13
孩
孩兒
孩
‘child (literary/elevated)’
‘child-son’
‘child (informal)’
b. hé (Literary/elevated)
何
‘what’
hé-wù/shén.me
何物（什麼)
‘what-thing’
shá (colloquial)
啥
‘what’
Chou then concludes that the operation of two-syllable words becoming
one-syllable words, i.e., [2-to-1], is a very active morphological process in many
Mandarin dialects. The operation here is thus generalized as follows: first, the
last morpheme in a two syllable compound loses its meaning with its syllabic
status; second, the phonological feature of the reduced syllable is joined to the
first one according to general phonological constraint/s, yielding what Kratochvil (1968) called a fusion syllable (i.e., háir and shá). Given the observations
provided by Chou (1982), Kratochvil (1968), Xu (1990) and Wang (1994), that
even if functional morphemes are beyond its scope, the MR still wants every
root morpheme to be formed by itself independently. Therefore, the suffix has
a tendency to lose its syllabic status.14 If it is so, the er-reduction not only provides examples of anti-disyllabicity, but also argues for the dynamic power of
the MR.
Monosyllabicity and disyllabicity
77
Another type of anti-disyllabicity concerns with the process that only onesyllable morpheme of a disyllabic word is used as root morpheme for new
compounds. For example,
cáichǎn 財產 ‘wealth’ : pòchǎn 破產 ‘to go bankrupt’
she-ngyı-n 聲音 ‘sound’ : lùyı-n 錄音 ‘sound recording’
miànkǒng 面孔 ‘face’ : huìmiàn 會面 ‘to meet’
zérèn 責任 ‘responsibility’ : fùzé 負責 ‘to be responsible’
she-ntǐ 身體 ‘body’ : xiànshe-n 獻身 ‘to devote’
ju-nduì 軍隊 ‘army’ : ca-nju-n 參軍 ‘to join the army’
Note that the monosyllabic morphemes taken from disyllabic words are purposely used to make new disyllabic words. This is so because the circumscribed
monosyllables are bound morphemes, hence they cannot be used along in natural speech.15 This further indicates that root morphemes are overwhelmingly
monosyllabic, which is a result of the MR.
The newly coined monosyllabic word dı- 的 ‘taxi’ may serve as a good example of monosyllabic morphemization for the MR. Di comes from a sound
translation taxi, originally as díshì, a disyllabic form. The English word taxi was
first translated into Cantonese and Taiwan Min dialects and then loaned into
Mandarin dialect. By using the light verb dă 打 ‘hit/do, ‘to take a taxi’ in Chinese is expressed as dă dı-shì 打的士 ‘to take a taxi’. ‘Dă dı-shì’ is perfectly grammatical but dă dı- 打的 ‘to take a taxi’ is more preferable because it fits the
disyllabic foot formation. After the VO disyllabic expressions used frequently
over the past 30 years, a newly coined monosyllabic word dí 的 for ‘taxi’ was
born. Now everyone knows what do you mean when you say wŏ xiăng dǎ yı-gè
dı- 我想打一個的 ‘I want (to call) a taxi’. Dı- is a full-fledged monosyllabic
word newly created in the language. As a result, the morphemization of díshì as
dí is a perfect example for the operation of MR (turning dı-shì into dı-) within
the disyllabic phrasal environment ‘dă dı-’ motivated and licensed by the FFR
as seen below.
3.3. Foot Formation Rule
As seen above, the second fundamental rule in Chinese prosodic morphology
is the Foot Formation Rule (FFR for short). Following the evidence and arguments given in Chapter 2 and the sandhi domain discussed by Shih (1986),
Feng (1995) suggested further that the FFR established in Classical Chinese
also functions in Mandarin Chinese. That is:
14.
f
σ
σ
78 Monosyllabicity and disyllabicity
As indicated in previous studies of Modern Chinese (see, among others, Chen
1979, Shih 1986, and especially Feng 2000: Chapter 2 and Duanmu 2000), Chinese employs disyllabic foot structure in everyday speech. For example,
15. A: Nı̆ duo- dà le?
you how big Asp.
‘How old are you?’
B: a. *Wŭ.
five
‘Five (years old).’
b. Wŭ suì.
five years
‘Five years (old).’
c. Shí wŭ.
fifteen
‘Fifteen (years old).’
16.
a.
b.
Éméi, Tàiháng, Huà-sha-n, Tài-sha-n, Jı̌ngga-ng-sha-n do-u shì Zho-ngguó de
míng sha-n.
Emei, Taihang, Hua mountain, Tai mountain, Jinggang mountain are
all China’s famous mountains.
‘Emei, Taihang, Hua, Tai and Jinggang are all famous mountains in
China.’
Éméi,Tàiháng, *Huá, *Tài, Jǐngga-ng do-u shì Zho-ngguó de míng sha-n.
Emei, Taihang, Hua, Tai, Jinggang are all China’s famous mountains.
‘Emei, Taihang, Hua, Tai and Jinggang are all famous mountains in
China.’
The examples in (15) show that a monosyllabic word cannot be used independently within the context cited above. The examples in (16) show that monosyllabic words cannot form an independent foot co-occurring with other feet.
These examples suggest that the FFR must be considered as a prosodic constraint in Chinese grammar. Of course, exceptions to the FFR may be found in
cases like the following:
17. Fàn, wŏ yı-diănr do-u bù xiăng chı-.
Food I little all not want eat.
‘As for food, I don’t want to eat it at all.’
A monomorphemic word can form an independent foot as seen in (17)
above. However, it must be supported by a pause after it (indicated by
‘#’), that is:
Monosyllabicity and disyllabicity
79
18. Fàn # wŏ yı-diănr do-u bù xiăng chı-.
Food # I little all not want eat.
‘As for food, I don’t want to eat it at all.’
It is well known (see Chao, 1968:67) that there is an overt grammatical pause
between the topic (or the subject) and the comment in Chinese, and this is
especially true when a monosyllabic form acts independently as a topic or a
subject. Hence monosyllabic forms are structurally limited only to appear in
the topic and subject positions in Chinese. In this situation, if we take the pause
to be some kind of prosodic epenthetic device for the single syllable, this type
of foot would be structurally represented as follows (a syllable with a prosodic
epenthetic pause):
19.
f
Fàn [pause],
Food,
wŏ yīdiănr dōu bù xiăng chī.
I don’t want to eat it at all.
If this is so, the generalization that a monosyllabic word cannot form an
independent foot can also hold in situations where a monosyllabic foot occurs.
3.4. Levels for operation of MR and FFR
Given the two rules (MR and FFR) outlined above, we are facing a paradoxical
situation in Chinese morphology. By the MR, monosyllabic words are licensed
and hence legitimate in the language. If this is so, how could disyllabic words
develop anyway? By the FFR, any instances of phonological words must be
disyllabic, hence all monosyllabic words must be ruled out as a violation of the
prosodic minimality. If this is correct, why are there monosyllabic words cooccurring with disyllabic forms in the first place, and why is there a tendency of
anti-disyllabicity as seen above? Obviously these two rules are mutually exclusive. That is, if there is a rule that requires prosodic morphological units to
be formed by only one syllable, a disyllabic rule could not be allowed in the
prosodic morphology. On the other hand, if a disyllabic rule is operative in the
prosodic morphology, the monosyllabic rule would have been eliminated in
that system because these two rules are incompatible. As a result, the operation
of one rule will be at the expense of the other. However, these two rules coexist in Chinese and their output occurs side-by-side. As a result, a theory of
Chinese morphology is called for so that these two rules operate freely in the
morphological system.
In this chapter we shall propose, following Feng (1995, 2003), that the
MR and the FFR are actually operating at two different levels of grammar, that
is, the MR applies at a level where morphemes or monosyllabic words are constructed, while the FFR applies at the post-morpheme level where morphemes
80 Monosyllabicity and disyllabicity
(bound or free) are put together, as seen from the following analysis of (20a,b)
(‘M’ stands for morphemes and ‘Wd’ for words).
20.
a.
S
b.
XP
|
Wd
XP
|
Wd
|
M
|
M
σσ
copy
σσ
paper
XP/S
word/phrase level
X0XP
X0/XP
M
morpheme level
word/phrase level
M
σ
σ
㼷
᮷
ঠ
Ԧ
morpheme level
(20a) represents a structural analysis where syllables are organized into morphemes (M), morphemes are combined into words (Wd), words are grouped
into phrases (XP), and finally, phrases are structured into a sentence (or a larger
phrase). This is a possible hierarchical structure for languages like English. Note
that there is no constraint on one-to-one correlation between an individual
syllable and an individual morpheme in English. However, since the MR is
a rule in Chinese, a structure like (20a), which is possible in other languages,
must be re-structured as (20b) in Chinese. That is, one syllable must correlate
directly to a morphological unit (either a morpheme or a monosyllabic word).
Since a morpheme (bound/free) plus a morpheme (bound/free) in Chinese
must be formed in the syntax,16 (20b) is a natural result of morpheme (or syllable) combination.
Given this, it is clear that only after the structure organized by the insertion of
lexical items produced by the MR can the application of the FFR take place,
as illustrated in (21).
21.
XP
a.
MR
XP
b.
XP/Wd
XP\Wd
M M
│ │
σ σ
M M
│ │
σ σ
FFR
f
XP\Wd
M
│
σ
M
│
σ
f
XP\Wd
M
│
σ
M
│
σ
Monosyllabicity and disyllabicity
81
Since the FFR is syllable-based and since the MR demands each syllable
be a morpheme, a grouping of syllables will inevitably result in a grouping of
morphemes (bound/free). Since the FFR cannot apply without a grouping of
syllables, and since syllables (= morphemes) cannot be grouped without syntax,
the application of the FFR on syllables will inevitably involve an operation of
syntax.
Note that if the language has only the FFR without the MR, then freestanding words (non-clitic) in Chinese would all have been constructed as
disyllabic forms. This is because monosyllabic words violate prosodic minimality if morphemes are sensitive to the FFR. Sub-minimal items are either barred
from the lexicon or are brought into it through various augmentation processes.
However, the MR is a dynamic rule operating at morphemic level in Chinese,
and most crucially, it functions at a level before morphemes are (syntactically)
grouped, and is different from the FFR. As a result, the applications of one rule
cannot be stopped by that of the other. That is, the MR cannot go beyond the
morphemic level to prevent the operation of the FFR, and the FFR cannot
apply to levels lower than syllables to interfere with the MR. Since each of
them has its own domain of application, the MR freely produces morphosyllabic forms below phrasal level, while the FFR unobstructedly generates Prosodic Words (PrWd, see Chapter 4 for detailed discussion) above morpheme
level. As a result, Chinese morphology is not only determined by the FFR but
also by the MR. This scheme can be represented as in Table 3.1.
Table 3.1 Two-rule system of Chinese morphology
Levels
Rules
Process
Lexicon
Phrasal
FFR
PrWd
Disyllabic forms
Morphemic
MR
M
│
σ
Monosyllabic forms
Metrical phonology
M
│
σ
Given this analysis, we see that the two levels make it possible for the coexistence of monosyllabicity with disyllabicity in Chinese morphology, and under
this system, we may simply say that there are two rules: one is the morphemic
MR and the other is the phrasal FFR, thus the monosyllabic constraint governs
nothing but morphemic elements while the FFR governs everything that is
syntactic (thus compounds are formed syntactically).
3.5. Interaction between MR and FFR
As we have seen before, both MR and FFR have their own separate domain
of application, and because of this, it becomes possible for disyllabic forms to
coexist with monosyllabic forms. However, although neither of these two rules
82 Monosyllabicity and disyllabicity
would prevent the application of the other, each of them affects the outputs of
the other. This is because the output of the MR must appear in phrases where
the FFR dominates. As a result, the outputs of the MR must be processed and
hence satisfy the FFR at the post-morphemic level. On the other hand, the
MR is also a rule that dominates at the morphemic level, hence disyllabic morphemes will violate the requirements of the MR. The domination relationship
is thus formulated as below (where “>>” means “dominates”):
22. Syntactic level
FFR >> MR
Morphemic level MR >> FFR
This situation can best be interpreted in terms of the Optimality Theory
(OT). That is, when the FFR dominates the MR, the output of the MR must
be considered as a violation of the FFR. When the MR dominates the FFR,
the output of the FFR may or may not be considered as a violation of the MR,
depending on whether the disyllabic forms produced by the FFR are represented by one or two lexical morphemes. If the disyllabic forms are constructed
by a combination of two lexical morphemes, the MR is satisfied, and if it is
represented by only one lexical morpheme (by means of idiomatization or lexicalization, etc.), the result would be a violation of the MR (unless the second
syllable is functional or the whole thing is non-indigenous.).
Thus, the intervention of one rule on another will inevitably result in the
following situation: monosyllabic words always violate the FFR at the postmorphemic level, thus they must be combined with another syllable/morpheme
to appear at the phrasal (syntactic compound) level. On the other hand, the
MR also intervenes to prevent polysyllabic morpheme (i.e., a syllable contains
two morphemes) and polymorphemic syllable (i.e., a two-syllable unit with
only one lexical meaning) from occurring in the operating system.
Given this, it is not surprising that the dominating relationship of FFR >>
MR will give rise to enormous disyllabic forms (or combinations) as exemplified in (23a,c) and also causes some prosodically constrained syntactic consequences as shown in (23d,g).
23. a. to look for
尋*ziIm
(Medieval Chinese)xún.me (Mandarin)
b. to blink
眨*tʃɣɛp (Medieval Chinese)zhă.me (Mandarin)
c. Mama, wŏ yào shuì.*(jiào)
Mom, I want sleep
‘Mom, I want to sleep.’
d. *zhòng zhí shù
‘plant trees’
e. *Bă liăn ca-
zhòng zhí shùmù
‘plant trees/woods’
Bă liăn ca--ca
ba face wipe
‘to wipe face’
BA face wipe-wipe
‘to wipe face’
Monosyllabicity and disyllabicity
83
f. *Jìnxíng găi
Jìnxíng găi-gé
‘carry out change’
‘carry out change change’
g. *Ma-, wŏ è le, wŏ yào chıMom I hungry Asp.
Mom, I want eat
‘Mom, I am hungry, I want to eat.’
Ma-, wŏ è le, wŏ yào chı- fàn
Mom, I hungry Asp. I want eat-rice.
‘Mom, I am hungry, I want to eat.’
The examples given in (23a,b) show that some monosyllabic words in Chinese have developed into disyllabic forms by adding an epenthetic syllable in
Mandarin Chinese. Example (23c) shows that, under the pressure of the FFR,
some intransitive verbs have developed into transitive verbs (e.g., shuì 睡 ‘sleep’)
forming VO compounds (shuìjiào 睡覺 ‘have a sleep’). Examples (23d,g) show
that monosyllabic words (objects or verbs) must be ruled out prosodically
because of their incapability of forming a foot and hence realizing the Nuclear
Stress (NS) at the end of the sentence (see Chapters 4–5 for more discussions
on the NSR).
Contrary to the requirement for disyllabicity demanded by the FFR, the
dominating relationship of MR >> FFR functions to motivate and thus protect monosyllabic forms as illustrated in (24).
24. [2-to-1] Process
a. Hái+ér
‘child son’
b. Jı-n+rì
‘this day’
c. Shén.me
‘what’
Háir
‘children’
Jı-r
‘today’
Shá
‘what’
Morpholization
d. Húdié
‘butterfly’
e. Fútè
‘volt’
f. Sàie-nsı‘science’
g. Dı-shì
Dă dı-
Diéyŏng ‘butterfly swim’
‘butterfly stroke’
Fú
‘volt’
Ke-xué ‘branch (of academic) study’
‘science’
‘taxi’
‘take a taxi’
As seen before, the reduced syllables in (24a,c) may be initiated under purely
phonological conditions, but it is important to realize that the possibility of all
[2-to-1] morphological processes (24d–g) is possible only if they are motivated,
84 Monosyllabicity and disyllabicity
protected or licensed by the MR, because the results of such a process would
otherwise be impossible under the requirement of FFR. Given this, it would be
better to consider Chinese morphology as a system governed by the cooperation and intervention of the two rules.
The theory presented above implies clearly that the intervention of one rule
on the other will inevitably result in [1-to-2] and [2-to-1] morphological processes under certain conditions as seen above. The [1-to-2] process is motivated
by the FFR on the phrasal (or syntactic compound) level, in which one monosyllabic form becomes a disyllabic form by combining with another morpheme
(or word), or by adding an extra epenthetic syllable. The [2-to-1] process is
protected and in some cases demanded by MR. That is, if one of the two morphemes loses its lexical meaning or the two meanings are no longer compositional after a long period of usage, the second syllable of that disyllabic from will
be reduced, yielding a match between one syllable with one morpheme (M=σ).
These two types of processes may be schemed as follows (“M” stands for lexical
meaning, “σ” for full-tone syllable, “.σ” for neutral-tone syllable):
25.
Under this schema, it is no longer surprising that there are two opposite
tendencies ([1-to-2] and [2-to-1] processes) in Chinese morphology. Furthermore, the ‘Monosyllabic Myth’ (see DeFrancis, 1986:177–188) can be resolved,
at least partially, by the cooperation as well as the interaction and intervention
between the two fundamental rules (MR and FFR) at certain levels of the
grammar.
3.6. Redundancies in the Chinese lexicon
Given the two-rule system of Chinese morphology, a natural conclusion about
the lexicon of Chinese would be that the morphological operations of MR and
FFR will result in a certain degree of redundancy for wordhood in the language.That is, one meaning may possibly be represented by two types of forms:
Monosyllabicity and disyllabicity
85
a monosyllabic form produced by MR and a disyllabic form resulting from the
FFR. This, as shown in (26), is exactly what happens in Modern Chinese.
26. a. Xiăng 想
‘miss’
Xiăngniàn 想念
‘miss’
Jia-o 澆
‘water’
Jia-oguàn 澆灌
‘water’
Fàn 飯
‘rice’
Mı̆fàn 米飯
‘rice’
Suàn 蒜
‘garlic’
Dàsuàn 大蒜
‘garlic’
wŏ xiăng ma-.ma
‘I miss (my) mother’
wŏ.men xiăng niàn zŭguó
‘We miss (our) country’
jia-o jia-o hua‘water flowers’
jia-oguàn hua-cǎo
‘water plants’
wŏ ài chı- fàn
‘I like eat rice’
wŏ chı- mı̆-fàn
‘I eat rice’
wǒ qù măi suàn
‘I go buy garlic’
wŏ măi dà-suàn
‘I buy garlic’
b. nà-lı̆ (that-place, there) 那裡
zhè-lı̆ (this-place, here) 這裡
jı-n-rì (this day, today) 今日
míng-rì (next day, tomorrow) 明日
nàr (there) 那兒
zhèr (here) 這兒
jı-r (today) 今兒
mír (tomorrow) 明兒
Both types of words co-exist in the language. The monomorphemic words
meet the MR and disyllabic forms satisfy the FFR. However, when monosyllabic morphemes/words are used (in stressed positions) alone in natural speech,
they violate the requirement of the FFR, thus a monosyllabic form must cooccur or be grouped with others to satisfy the FFR at the phrasal level. When
a combination happens to match perfectly (i.e., a minimal foot with a minimal
phrase), the prosodic morphological system will immediately identify it as a
PrWd according to the FFR given in (25). Furthermore, under the prosodic
pressure and pragmatic usages, the two elements in a PrWd will be ‘frozen’ or
‘idiomatized’ and finally become what is called a compound word in Chinese.
3.7. Conclusion
This chapter provides a two-rule system for Chinese morphology, by observing
that in previous works (though stated in different terms sometimes) misinterpretation of these two rules has caused a great confusion regarding the syllabic
property of the Chinese language (DeFrancis, 1986:177–188). Given the analysis in this chapter, it is clear that the confusion is mainly caused by the lack of a
86 Monosyllabicity and disyllabicity
proper theory to capture the rule system and to explain how it operates in such
a way that these two rules cooperate while intervening with each other in the
outputs produced by interface of the two.
The theory presented here may lead to some desirable theoretical as well as
empirical consequences.
First, a natural and plausible conclusion for the ‘monosyllabic myth’ in Chinese is that there is only a system of rules which interact to derive the lexicon
and whether Chinese is monosyllabic or polysyllabic is merely a result of rule
interaction.
Second, the notion of PrWd as seen in Table 3.1 and (25) is crucial in resolving the traditional problems of distinguishing phrases from words in Chinese
morphology. As we will discuss in detail in the next chapter, according to the
notion of PrWd, any instance of combinations of two (monosyllabic) morphemes will fall into the category of PrWd, hence examples such as báizhı̆ 白紙
‘white paper’ and kànpò 看破 ‘see through’ are prosodic words. In the present
theory of this chapter, PrWds are outputs of the morphological system in the
sense that a compound must first be a PrWd, even if a PrWd may not necessarily be a compound. Therefore, although kànpò is a PrWd, it may be used in
other environments as a phrase such as kàndepò 看得破 ‘be able to see through’,
where a functional marker (i.e., the inflected morpheme -de-) can be inserted
in between the two morphemes. According to this analysis, it is natural to have
the following categories:
27.
Phrase:
bái.de zhı̆ 白的紙
kàn.de pò 看得破
‘papers that are white’
‘be able to see through’
PrWd:
báizhı̆ 白紙
kànpò 看破
báishŭ 白薯
*bái.de shŭ 白的薯
Găishàn 改善
*găi.de shàn 改得善
‘whitepaper; white paper’
‘see through; understand thoroughly’
‘white potato, sweet potato’
‘sweet potato’
‘change better, improve’
‘be able to change better’
Compound:
Given the prosodic category of PrWd in Chinese morphosyntax, the peculiar behavior of cases like báizhı̆ and kànpò could be accounted for systematically.
Third, although the rule system of Chinese morphology produces a certain degree of redundancy in the lexicon as seen in (26), other systems of the
language will be expected to operate on these forms so that they are not pure
redundancies.The fact that monosyllabic forms and disyllabic forms have served
different stylistic-register and prosodic purposes in the language confirms the
hypothesis we have made here. Li J. (1989), for example, observed that disyllabic
forms tend to be used on formal occasions while monosyllabic forms are generally very casual. Furthermore, as seen in (23d–g) above, strong prosodic positions
tend to attract disyllabic forms and exclude monosyllabic forms, indicating that
the monosyllabic and disyllabic forms have played different roles in the Chinese
prosodic system (see Chapters 4–5) and register system (see Chapters 6–7).
Monosyllabicity and disyllabicity
87
It is also expected that further research may provide evidence from language
acquisition as to whether or not these are different steps of acquiring these two
rules. Actually, Tang’s study (1988:43–92) has already suggested that children
exhibit a tendency to first acquire the MR with an unproductive FFR. If this
is so, it is not surprising that monosyllabic forms exhibit a high frequency of
occurrence, given that the elevated usage of the FFR (i.e., forming a foot by
using classical words or morphemes) would be a more advanced adult grammar. Obviously, if the theory given here is correct, it may also shed some light
on language acquisition regarding the diglossic grammar of children and adults.
Notes
1 According to Chou’s calculation (1982:10), there are 1,478 (13%) monosyllabic forms,
6,816 (60%) disyllabic forms and 3,072 (27%) polysyllabic forms in Readings in Sayable
Chinese (eds. by Chao,Y.R.1968).
2 See, for example, Footnote 7 of Duanmu 1993.
3 See, for example, Tang 1988:3,4,568; Li, 1989:114.
4 Therefore, in Chinese, syllabic writing is de facto morphemic writing, and thus to call it
morphosyllabic is correct, but not fundamentally different from calling it logographic or
morphemic. (William G. Boltz 1989. Reviews (11) Sino-Platonic Chapters, 14)
5 The MR was formulated in Feng’s 1995 as a monomorphemic axiom, stated as follows:
Monomorphemic Axiom (MA).
6 For example, no indigenous syllable starts with /r/ in Japanese (Tsujimura 1992:477–
522) and there are no length constraints on translated foreign names in Mandarin Chinese (cf. Chapter 5 of this book), indicating that non-indigenous forms require different
phonological and morphological rules.
7 In the ongoing discussions, we use the symbol ALIGN to indicate an alignment of
both edges ALIGNL for left-edge alignment, and ALIGNR for right-edge alignment.
Furthermore, ALIGN: [X]=[Y] represents a situation where X and Y not only begin
and end together, but also are interchangeable: If it is X, it must be Y. However, ALIGN:
[]X = []Y stands for a situation where X and Y begin and end together, but Y may not
necessarily be X (although X must be Y).
8 Downing (2006) also developed a similar rule, called “one syllable one morpheme”.
9 The situation concerns resyllabification in Chinese is limited to Sentential Final Particles
(SFP) that are resyllabified with a coda consonant of the preceding syllable, such as tian-a
> tian-na. This process may also confirm the validity of ALIGN in the sense that SFPs
are toneless and not root morphemes.
10 In other words, cran is reanalyzed as a bound (or inner-word) morpheme in cranberry.
11 Root morphemes are morphemes that can be used to form a syntactic compound. Thus,
xiè (crab) in zı̆-xiè (purple crab) is a root morpheme, because it serves as a head of the
syntactic compound, but zi in yı̆.zi (chair) is not a root morpheme, because it is a suffix
(functional morpheme), and the compound is derivative, rather than syntactic. Epenthesis syllables like the ones in (12) are naturally excluded from root morphemes.
12 The distinction between root morphemes and non-root morphemes (excluding functional epenthesis syllables) can also be seen from the fact that resyllabification in Chinese
does not exist between root morphemes demanded by the Edge-Alignment given in (3),
but sometimes happens to functional morphemes such as a in the following:
Tia-n-a Tia-n-nà
Heaven-particle Goodness!
13 Note that although there may be other source/s for the coda r sound in Mandarin, the
suffix -er is undoubtedly the origin of r-suffix given the intermediate process of hai.
88 Monosyllabicity and disyllabicity
er in neighboring dialects. Thus we can formulate the process of the change as follows:
[CV(C)Root-Noun+er]COMPOUND → CV(C)Root-Noun-erSUFFIX → CV-r (where r is part of the
syllable carrying a register-functional feature). For ‘register function’ see Chapter 6 for
detailed discussion.
14 Neutralized suffixes are also motivated in this regard.
15 The opposite may also be true, the bound morphemes in a disyllabic words are taken to
form disyllabic words.
16 See Chao 1968.
References
Chao, Yuen-Ren. 1968. A Grammar of Spoken Chinese. Berkeley: University of California
Press.
Chao, Yuen Ren and Lien Sheng, Yang. 1962. Concise Dictionary of Spoken Chinese. Cambridge, MA: Harvard University Press.
Chen, Matthew Y. 1979. “Metrical Structure: Evidence From Chinese Poetry”. Linguistics
Inquiry 10: 371–420.
Chen, MatthewY. 1987.“The Syntax of Xiamen Tone Sandhi”. PhonologyYearbook 4: 109–150.
Chou, Fa-kao. 1982. “Re-Examination of Monosyllabicity in Chinese 中國語單音節性之
再檢討. Chinese Study 中國語文研究 5: 5–12.
DeFrancis, John. 1950. Nationalism and Language Reform in China. New Jersey: Princeton
University Press.
DeFrancis, John. 1986. The Chinese Language: Fact and Fantasy. Honolulu: University of
Hawaii Press.
Downing, Laura J. 2006. Canonical Forms in Prosodic Morphology. Oxford: Oxford University
Press.
Duanmu, San. 1993. “Rime Length, Stress, and Association Domains”. Journal of East Asian
Linguistics 2.1: 144.
Duanmu, San. 2000. The Phonology of Standard Chinese. Oxford: Oxford University Press.
Feng, Shengli. 1995. Prosodic Structure and Prosodically Constrained Syntax in Chinese. Ph.D. dissertation. Philadelphia: UPENN.
Feng, Shengli. 2000. Prosodic Syntax in Chinese 漢語韻律句法學. Shanghai: Shanghai Educational Press.
Feng, Shengli. 2003. “Prosodically Constrained Postverbal PPs in Mandarin Chinese”. Linguistics 41.6: 1085–1122.
Feng, Shengli. 2009. “Monosyllabicity and Disyllabicity in Chinese Prosodic Morphology”.
Macao Journal of Linguistics 1: 4–19.
Jing,Yunjing. 1969. “Disyllabicity of Modern Mandarin”. Chinese Culture 10.4: 88‑104.
Karlgren, Bernhard. 1920. “Le proto-chinois, langue flexionelle”. Journal Asiatique 15: 205–
232. 由馮承鈞譯為原始中國語為變化語說 ，東方雜誌 1929年3月10日, 26.5: 77–88.
Karlgren, Bernhard. 1949. The Chinese Language: An Essay on Its Nature and History 漢語的
本質和歷史. New York: The Ronald Press Company.
Kennedy, George A. 1951. “The Monosyllabic Myth”. Journal of the American Oriental Society
71.3:161‑166.
Kennedy, George A.1964. Selected Works of George A. Kennedy (Edited by Li, Tien-yi). New
Haven: Far Eastern Publications.
Kenstowicz, Michael. 1993. “Evidence for Metrical Constituency”. In Kenneth Hale and
Samuel J. Keyser (eds.) The View From Building 20 – Essays in Linguistics in Honor of Sylvain
Bromberger. Cambridge, MA: MIT Press, 257–273.
Monosyllabicity and disyllabicity
89
Kratochvil, Paul. 1968. The Chinese Language Today: Features of an Emerging Standard. London:
Hutchinson.
Li, Charles N. and Thompson, Sandra A. 1981. Mandarin Chinese: a Functional Reference Grammar. Berkeley: University of California Press.
Li, Fang-Kuei. 1973. “Language and Dialects of China”. Journal of Chinese Linguistics 1.1:
1–13.
Li, Jiashu. 1989. “Monosyllabic and Disyllabic Words in Everyday Oral and Writing Languages 單音節詞和雙音節詞在口語和書面語上的表現”. In Jiashu Li (eds.) Chinese
Study and Chinese Teaching 語文研究和語文教學. Hong Kong: Commercial Press.
McCarthy, John and Prince, Alan. 1993. “Generalized Alignment”. In Geert Booij and Jaap
van Marle (eds.) Yearbook of Morphology 1993. Dordrecht: Kluwer, 79–153.
Selkirk, Elisabeth. 1986. “On Derived Domains in Sentence Phonology”. Phonology Yearbook
3: 371–405.
Shen Huaixing. 2007. “On Entries of Binome Dictionary and Related Questions”
《聯綿字典》的收詞及相關問題, Lexicographical Studies 辭書研究 4: 85–92.
Shih, Chilin. 1986. The Prosodic Domain of Tone Sandhi in Chinese. Ph.D. dissertation. University of California, San Diego.
Sproat, Richard. 1998. Review of Packard (1998): New Approaches to Chinese Word Formation. In the Yearbook of Morphology.
Sproat, Richard and Shih, Chilin L. 1993. “Why Mandarin Morphology Is Not Stratum
Ordered”. In Geert Booij and Jaap van Marle (eds.) Yearbook of Morphology. Dordrecht;
Boston; London: Kluwer Academic Publishers, 185–217.
Sproat, Richard and Shih, C. L. 1996.“A Corpus-Based Analysis of Mandarin Nominal Root
Compound”. Journal of East Asian Linguistics 5.1: 49–71.
Su, Xiyu. 1989. “Discussion on Morphemization of Syllabic Symbols in Modern Chinese
Polysyllabic Forms” 試說現代漢語複音單純詞中的音節符號向語素的轉化. Chinese
Study 漢語學習 2: 21–25.
Tang, Tingchi湯廷池. 1988. Hànyǔ Cífǎ Jùfǎ Lùnjí 漢語詞法句法論集 [Studies on Chinese
Morphology and Syntax]. (2 Volumes) Taipei: Xuesheng Shuju 學生書局.
Tsujimura, Natsuko. 1992. “Licensing Nominal Clauses: The Case of Deverbal Nominals in
Japanese”. Natural Language and Linguistic Theory 10: 477–522.
Wang, Hongjun. 1994. “Two Phonetic Word Formations in Common Use in Chinese
漢語常用的兩種語音構詞法”. Language Study 語言研究 1: 6578.
Xu, Tongqiang. 1990. “Structural Asymmetricity and the Cause of Language Change
結構的不平衡性和語言演變的原因”. Chinese Philology 中國語文 1: 1–15.
4Prosodically constrained
compound formation
Based on the prosodic theory developed in previous chapters, this chapter concerns with the mechanism of compound word formation in Mandarin Chinese, proposing that compound words in Chinese must first be a prosodic word
(PrWd, thereafter) even if a prosodic word is not by necessity a compound
word.
This chapter is organized as follows: section 1 gives an overview of different types of PrWds in Chinese; section 2 offers a detailed characterization of
prosodic words in Mandarin Chinese; section 3 focuses on the mechanisms
that generate compound words through PrWd formation; section 4 discusses
exceptions to our prosodic account in considering the contrast between [2+1]
syllable structure (i.e., 2 syllables + 1 syllable, thereafter) and [1+2] syllable
structure with respect to their morphological and syntactic properties; and section 5 is a summary.
4.1. Different types of PrWds
According to the present theory, the tremendous disyllabic forms in Mandarin
Chinese (MC for short, thereafter) are one of the results of prosodic word formation. There are at least nine types of PrWds in MC as listed below:
1. Reduplication
a.
b.
c.
d.
e.
tia-n 天 ‘day’
kàn 看 ‘look’
xún.me 尋麼 ‘look for’
ga-njìng 乾淨 ‘clean’
màn 慢 ‘slow’
tia-n-tia-n 天天 ‘every day’
kàn-kan 看看 ‘take a look at’
xún.me-xún.me 尋麼尋麼 ‘try to look for’
ga-n-ga-n jìng-jìng 乾乾淨淨 ‘to make it clean’
mànma-r 慢慢兒 ‘slowly’
A monosyllable plus a neutralized syllable like xúnme ‘look for’ is acceptable for reduplication, but a word with two full-tone syllables is barred from
undergoing reduplication. This restriction applies to not only nouns, but also
verbs, adjectives and adverbs, even if the reduplicated forms may have different
syntactic or semantics functions, for example:
Prosodically constrained compound formation
2. a. noun
b. verb
c. adjective
d. adverb
91
xı-ngqı- 星期 ‘week’
*xı-ngqı--xı-ngqı- 星期星期 ‘every
week’
jìsì 祭祀 ‘worship’
*jìsì-jìsì 祭祀祭祀 ‘try to worship’
zhěngqí 整齊 ‘neat’
*zhěngqí-zhěngqí 整齊整齊 ‘make
it clean’
gănkuài 趕快 ‘quickly’ *gănkuài-gănkuài 趕快趕快 ‘quickly’
3. Affixation
a. yǐ 蟻 (a bound morpheme) ‘aunt’ mă -yǐ 螞蟻 ‘aunt’
b. h 虎 (a bound morpheme) ‘tiger lăo-hǔ 老虎 ‘tiger
c. zhuo- 桌 (a bound morpheme) ‘table’ zhuo--zì 桌子 ‘table’
4. Epenthesis
a. to look for
b. to blink
尋**ziIm (Medieval Chinese) xún.me (Mandarin)
眨**tʃɣɛp (Medieval Chinese) zhă.me (Mandarin)
5. Exclamation
a. tia-n 天 ‘heaven’
b. ma- 媽 ‘mother’
tia-n-na 天呐 ‘My Goodness, curse’
ma--ya 媽呀 ‘Oh! When danger happens’
6. Abbreviation
a. bĕijı-ng dàxué 北京大學 ‘Beijing University’ → běidà北大
b. běijı-ng shı-fàn dàxué 北京師範大學 ‘Beijing Normal University’ →
běishı-dà 北師大
c. xia-nggǎng zho-ngwén dàxué 香港中文大學 ‘The Chinese University of
Hong Kong’ → gǎngzho-ngdà 港中大
7. Truncation
三十而立
sa-n shí ér lì
three ten and stand
‘A man stands up when he is 30 years old.’
而立之年
ér lì zhı- nián
and stand’s year
‘The year of 30’
8. Morpheme Combination
a. yán 鹽 ‘salt’
b. suàn 蒜 ‘garlic’
vs.
vs.
xián-yán 鹹鹽 ‘salty-salt, salt’
dà-suàn 大蒜 ‘big-garlic, garlic’
9. Phrasalization (a process to associate a head word with a phrase category)
a. xuě 雪 ‘snow’
b. bı-ng 冰 ‘ice’
bái xuě 白雪 ‘snow’
white snow
liáng bı-ng 涼冰 ‘ice’
cold ice
92 Prosodically constrained compound formation
10. Onomatopoeia
a. pe-ng pe-ng pe-ng 砰砰砰 (sound of knocking)
pengpeng, pengpeng *pengpeng peng pengpeng
b. ho ng lónglóng 轟隆隆 (sound of thunder)
honghong longlong *honghong longlonglong
honglong honglong *honglong honglonglong
c. dı-ngdı-ng da-ng 叮叮噹 (sound of bell)
dingding dangdang *dingding dangdangdang
dingdang dingdang *dingdingding dangdang
All the above examples show that disyllabic forms (and at most trisyllabic
forms) made by prosodic word formation are the most favorable and legitimate shapes observed in Chinese grammar such as reduplication and affixation, etc.
4.2. Prosodic property of Chinese word formation
If a foot in Chinese is formed by at least two syllables, and each syllable is
aligned with a morpheme or a word, then disyllabic feet are overwhelmingly
formed by combination of two monosyllabic words or morphemes in the
morphosyllabic system of Chinese as seen in Chapter 2. Based on the facts
and the prosodic theory developed in previous chapters, a generalization of
morphological function of prosodic words is proposed as follows.
11. Morphological function of prosodic words
Prosodic words are not necessarily compounds, but a compound must be a
PrWd.
The morphological function of prosodic word is taken as a general prosodic
constraint on morphology. In other words, compound word formation in Chinese is prosodically constrained.
There are several pieces of evidence supporting the generalization given
above. First, it is generally observed since Chao (1968) and Zhu (1982) that
there are basically five types of syntactic relations that words (or morphemes)
are combined into compounds in Chinese, namely, Subject + Predicate (SP),
Verb + Object (VO),Verb + Resultative Complement (VR), Modifier + Noun
(MN) and Coordination Construction (CC). All of these five types of compounds are formed binarily by two elements with a phrasal relation exhibited
in syntax. However, there is hardly any SVO compound in the language (see
below for interesting exceptions). Given the syntactic relations among all types
of compound formation, it is difficult to explain why SVO compounds do not
exist. According to (1), however, if a compound must be a PrWd, the lack of
SVO compounds can be explained in terms of the lack of SVO PrWds (see
below).
Prosodically constrained compound formation
93
Second, VO and VR compounds are very productive in Chinese. However,
there is hardly any compound word that is formed by a disyllabic verb plus a
monosyllabic object (i.e.,VV+O where the double V represents disyllabic verb
and the following O represents a monosyllable object, thereafter) or a resultative
complement (i.e.,VV+R) (Lu 1965:91, Duanmu 2000, Lu and Duanmu 2002):
12. tí-ga-o 提高
lift-high
‘raise’
ya--suì 壓碎
*tíbá-ga-o 提拔高
lift-high
‘raise’
*ya-jı̆ -suì 壓擠碎
zhàn-wĕn 站穩 *zhànlì-wĕn 站立穩
stand-firm
stand-firm
‘stop’
‘stop’
găi-zhèng 改正 *xiu-găizhèng 修改正
press-break
press-break
change-correct change-correct
‘break’
‘break’
‘rectify’
‘rectify’
yuè-gŭi 越軌
*cha-oyuèdé-zuì 得罪 *huòdé-zuì 獲得罪
guı̆ 超越軌
gain-guilt
gain-guilt
cross-rode
cross-rode
‘offend’
‘offend’
‘transgress’
‘transgress’
zhòng-dì 種地 *ge ngzhòng-dì 耕種地 que -dé 缺德
*que-shăo-dé 缺少德
plough-land plough-land
lack-virtue
lack-virtue
‘cultivate land’ ‘cultivate land’
‘mean’
‘mean’
Third, there are no compounds that are formed by monosyllabic verbs with a
disyllabic object or a disyllabic resultative complement. For example (see Dong
1998 and Feng 2000):
13. fàng-so-ng
放鬆 loose
‘relax’
*fàng-kua-nso-ng 放寬鬆 kuò-dà 擴大
release-loose
extend-big
‘relax’
‘enlarge’
*kuò-guăngdà 擴廣大
extend-big
‘enlarge’
suo--xiăo 縮小 *suo-- we-i xiăo
縮微小
găi-zhèng 改正 *găi-zhèngquè
改正確
shrink-small
‘reduce’
change-correct change-correct
‘rectify’
‘rectify’
shrink-small
‘reduce’
diàn-jia-n 墊肩 *diàn-jia-nbăng
墊肩膀
fill-shoulder fill-shoulder
‘shoulder-pad’ ‘shoulder-pad’
shè-fă 設法
set-method
‘try’
*shè-fa-ngfă 設方法
set-method
‘try’
zhàng-yì 仗義
*zhàng-yìqì 仗義氣
hold-justice
‘have a sense
justice’
que--dé 缺德
lack-virtue
‘mean’
hold-justice
‘have a sense of
justice’
*que--dàodé 缺道德
lack-virtue
‘mean’
94 Prosodically constrained compound formation
Fourth, expressions that are formed by more than three syllables do not look
like words judged by native Mandarin speakers (Lu and Duanmu 2002). For
example:
14. măn-chéng-fe-ng-yŭ 滿城風雨
full city wind rain
‘the talk of the town’
yı--yı--dài-shuı̆ 一衣帶水
one cloth belt river
‘close neighbors separated only by a trip
of water’
These expressions have been traditionally termed as Sìzì Gé 四字格 ‘Foursyllable Pattern’, but not compounds (as will be seen in Chapter 7). However,
if we ask: Why are they not words? What is the difference between Sizige and
compounds and where does the native speaker’s language intuition come from?
Hardly any answer can be offered either syntactically or morphologically. However, if we look at them from a prosodic point of view, the answer becomes
clear: compounds must be PrWds and four-character patterns are Compound
PrWds (i.e., PrWd+PrWd) which are not basic or simple compound words.
(see Chapter 7 for more detailed analysis about these forms).
Finally, trisyllabic compounds are formed mostly by the MN type (Lǚ1962):
15. a. diàn yǐng-yuàn 電影院
movie-theater
b. jiào yù-jú 教育局
education-department
shàonián-go-ng 少年宮
youth-club
lánqiú-duì 籃球隊
basketball-team
However, if we contract a four-syllable expression such as píxié go-ng
chǎng 皮鞋工廠 ‘leather-shoe work-place’ into a trisyllabic unit (i.e., making
a [2+2] form into a [1+2] or a [2+1] form), then only [2+1] is acceptable.
The [1+2] forms are generally not grammatical (see section 5 below for more
analysis of this type of example). For example, píxié go-ng 皮鞋工 ‘shoe-maker’
and píxiéchǎng 皮鞋廠 ‘shoe factory’ are both perfect compounds, but *xié
go-ngchǎng 鞋工廠 ‘shoe factory’ and *pí go-ngchǎng 皮工廠 ‘skin factory’ must
be ruled out.
The following classical example shows this point even more clearly: fùyìn
wénjiàn 複印文件 could be a well-formed VO phrase ‘to copy document’ or an
MN compound ‘copied documents’. However, if it is contracted into trisyllabic
units, fùyìn jiàn 复印件 (a [2+1] form) must be an MN compound (‘copied
document’) while yìn wénjiàn 印文件 (a [1+2] form) must be analyzed as a
VO phrase (‘to copy document’). The question is why only [2+1] patterns can
be lexicalized as compounds whereas [1+2] patterns are either ruled out or
considered as phrases. Given the prosodic word consideration, the reason can
be this: the [1+2] patterns cannot produce PrWds in Chinese prosodic morphology as we will see in next section (see also Duanmu 2000 for a different
Prosodically constrained compound formation
95
prosodic account). If [1+2] pattern is not a legitimate PrWd by definition (i.e.,
not produced by natural footing), then it cannot produce compounds according to (11). All the above evidence supports the claim that word formation in
Chinese is constrained by prosody. In other words, compounds are essentially
outputs of the prosodic morphological system in terms of PrWd template in
Chinese.
4.3. The mechanism of prosodically generated
compounds
How does the prosodic system generate compound words precisely? There are
two proposals being proposed in the literature, and each could generate wellformed compound words in Chinese. Let us look at what is called “PrWd-onTrees” (or P-Morphology, for short) first, and then the second one, i.e., “foot
directionality” (or F-Morphology, for short), respectively.
4.3.1. Compound generated through PrWd-on-Trees
Feng (1995) proposed a PrWd-on-Trees system dealing with the compound
formation in Chinese.This operation is technically based on a Foot Assignment
Principle, which is stated as follows:
16. Foot Assignment Principle (FAP)
In a phonetically realized syntactic tree (or phrase) structure, count the syllables from right to left until the number of syllables legitimately satisfies
the foot structural requirement.
The operational procedures of generating compound by prosody on syntactic
trees are illustrated as follows:
First, by the FAP, the foot assignment starts from the right of a tree, i.e., from
Y of the [X Y]XP (where the Y and X nodes contain a monosyllabic word or
morpheme, respectively):
17.
PrWd
σ
Y
σ]Foot
X
XP
Second, by Foot Formation Rule (i.e., a standard foot), the above tree will
produce a well-formed foot, and hence a standard disyllabic PrWd is born:
96 Prosodically constrained compound formation
18.
PrWd
[σ
Y
σ]Foot
X
XP
According to (11), foot determines PrWd, and PrWd then provides a template for compound formation. These two constraints will automatically give
rise to the impingement of prosodic structure on templatic morphology
(McCarthy and Prince 1986) – a typical case of prosodic morphological operation as shown in (12) (taken from Feng 2016):
19.
Compound
Prosodic Morphology
PrWd
[σ
Y
σ]Foot
X
Prosodic Phonology
XP
This operational mechanism provides various possibilities of building more
complex systems to produce different shapes of compounds motivated and further determined by different types of PrWds generated by FAR on the tree.
Specifically, the forms of [1+1] (standard PrWd by standard foot) and of [2+1]
(super-PrWd by super-foot) are well-formed outputs while those of [1+2] are
not legitimate. The detailed analyses of the five types of compounds (i.e., VO,
VR, SP, MN, CC) are shown in the following tree diagrams.
20.
VP
V’
SPEC
N
σ
V
σ
[
X (X=N or A = complement of the V)
σ
←
footing direction
Foot]
PrWd
Prosodically constrained compound formation
97
In the above tree diagram, the outputs of (20) are disyllabic VO and VR, all
of which are well-formed in the system and favored in the language. The same
operations also apply to tree structure (21) below for SP compounds.
21.
VP
NP
V’
N
σ
[
V/A
σ
Foot]
(V=intransitives; A=adjectives)
←
footing direction
PrWd
Here we have well-formed outputs generated by the operation illustrated on
the tree, such as:
22. dì
nián
ěr
dǎn
xìng
zhèn
qı-ng
ruăn
xiăo
jí
地震 ‘earth shake, earthquake’
年青 ‘year green, young’
耳軟 ‘ear soft, credulous’
膽小 ‘gallbladder small, timid’
性急 ‘temper impatient, short-tempered’
Note that there are no SVO or SVR compounds because if the number of
syllables satisfies the disyllabic foot already by the FAP (which says “count the
syllables from right to left until the number of syllables legitimately satisfies
the foot structural requirement”), there is no need for the footing to go beyond
the verb to get to the SPEC of VP node, i.e., the NP, forming a trisyllabic
super-foot (i.e., trisyllabic foot). In other words, as long as the disyllabic footing
is satisfied, the whole process of footing for compound is finished, thus leaving
no room for the trisyllabic foot to be formed further in the system. This means
that (23) is out:
23.
?
Y
σ
X
σ
[
σ
Foot]
PrWd
← footing direction
98 Prosodically constrained compound formation
If X is a disyllabic PrWd already, there is no further morphological footing
process required or allowed further. Thus, there are no [1+2] compounds in
Chinese no matter if they are SVO or SVR, or VOO or VRR forms. As a result,
no trisyllabic PrWds and compounds of the [1+2] structure (i.e., SVO/R,
VOO/RR) are produced by the system of prosodic morphology in Chinese.
On the other hand the system produces the [2+1] trisyllabic compounds in
the tree structure as follows:
24.
N’
N
Modifier
σ
[
σ
σ
Foot]
← footing direction
PrWd
Based on the process of counting the syllables from right to left required by
FAP, the operation will not stop if the number of syllables does not legitimately
satisfy the foot requirement. As seen in (24), the rightmost N is a monosyllabic
element and hence illegitimate for being a foot. However, it cannot form a foot
with the preceding syllable since it is a part of the PrWd existing before the
operation. Instead, it combines with the whole disyllabic modifier to form a
super-foot. That is to say, the outputs of (24) are acceptable because the footing
direction is from right to left, so the process starts from the rightmost monosyllabic head N, and it will be grouped with the disyllabic modifier, and since the
modifier is a disyllabic word already formed in the lexicon (a compound word),
the footing process must respect the Lexical Integrity and group the monosyllabic N with the disyllabic modifier, thus forming a trisyllabic Modifier +
Noun compound with a [2+1] syllable structure.
Different from (20), (21) and (24), the process of (25) produces ill-formed
outputs within [Modifier Head] (rather than SVO/R) structures, just as that of
(23) produces ill-formed SVO/R structures, as seen below:
25.
N’
Modifier
σ
N
σ
[
σ
Foot]
PrWd
← footing direction R-to-L
Prosodically constrained compound formation
99
This is so because, as seen in (25), the left-footing process will be terminated when
the footing process is minimally satisfied. In other words, the stray syllable to the left
of a foot will be taken as first syllable of another foot and restarting the counting
process again, which will not give a trisyllabic result either.Thus there are no [1+2]
compounds formed by this process.This is evidenced by examples given below:
26. *dì-zhèndòng 地震動
earth-move
‘earthquake’
*shu--sha-ngdiàn 書商店
book-store
‘bookstore’
*pí-go-ngchǎng 皮工廠
leather factory
‘leather factory’
*tóu-téng tòng 頭疼痛
head-ache
‘headache’
*xié-go-ngchǎng 鞋工廠
shoe-factory
‘shoe-factory’
4.3.2. Compound generated through foot directionality
As we have seen from the processes of P-Morphology (i.e., PrWd-on-Trees) given
above, all of the legitimate shapes of the five types of compounds (i.e., the SP,VO,
VR, CC, MH) are naturally generated whereas the illegitimate ones are automatically ruled out by the same process of ‘PrWd-on-Trees’. The systematic results
cannot be obtained from purely morphological, semantic and syntactic analyses
although all of the five types of compounds are formed binarily by two elements
with a phrasal relation exhibited in syntax. More specifically, the lack of SVO
compounds in the language rules out a purely syntactic operation in Chinese
morphology while the illegitimate reduplication of disyllabic words eliminates a
purely morphological account as well as a semantic account in Chinese morphology. All these facts motivate a more plausible theory for Chinese morphology; that
is, words in Chinese that are produced by all possible morphological processes
are constrained by prosody. Under these considerations, a more general question
is raised: are there more fundamental factors behind the prosodic morphological
process of generating compounds in Chinese? This question motivates the second
proposal for compound formation which was triggered thousands of years ago
(as seen in Chapter 2) and is constrained by prosody till nowadays, namely, “foot
directionality”. The basic mechanism for a simpler account with a deeper understanding of prosodic morphology in Chinese is stated as follows (Feng 2005):
27. Footing direction of Chinese morphology (F-Morphology, for short)
In Chinese morphology, only a right-footed operation results in a compound (2005:4–8).
The F-Morphology is considered as the key factor in Prosodic Grammar and it
is abstracted from the rightward-footing applications on different tree structures.
The insight of the ‘Footing Morphology’ captures not only all grammatical
100 Prosodically constrained compound formation
results by simply a footing direction application, seen as in (20), (21), (23) and
(24), but also the fact that opposite footing direction on the same trees will
result in either ungrammatical outputs (as (25)) or phrases (but not compounds)
as seen in section 5 below.
The two operational proposals introduced above, namely, P-Morphology and
F-Morphology, each works efficiently for Chinese prosodic morphology even
if the latter is simpler than the former. Note further that, given the two directional footing processes in prosodic phonology and its grammatical effects in
morphology and syntax, scholars have recently summarized related prosodically
determined grammatical patterns in terms of yòu xiàng gòu cí 右向構詞, ‘rightward-footing for compounds’ as contrasting with zuǒ xiàng gòu yǔ 左向構語
‘leftward-footing for phrases’1 in current prosodic morphology literatures (see
among others, Duanmu 2000, Lu and Duanmu 2002; Wang HJ 20002; Wang LJ
2009, 2014; Pei 2016, Zhuang 20123 and references cited there).
While both the P-Morphology and the F-Morphology are able to generate grammatical results in Chinese morphology, they are not the only rules in
morphology because there are complexities of words formed in syntax but controlled or influenced by factors of other components of the grammar. As a result,
there are exceptions (and seeming counterexamples) to the rules and principles
of prosodic morphology in the language, which we will discuss in next section.
4.4. Exceptions to prosodic constraint on compounds
There are a few types of exceptions to the prosodic constraints on compound
formation outlined above, and each type of exceptions consists of a sub-category
of semantic field in the ordinary language.
4.4.1. Exceptions from special terminology
In Chinese gastronomy, for example, the cuisine terminologies are obviously
violating the prosodic constraints outlined in previous sections. The cooks or
the professionals have always given all the dishes beautiful names, which are
vividly and artistically expressed, thus attracting one’s taste. For example,
28. a. 龍虎鬥
lóng hǔ dòu
Dragon Tiger bucket
‘Dragon and Tiger Bucket Dish’
b. 涮羊肉
shuàn yángròu
boil mutton
‘boiled mutton’
c. 螞蟻上樹
mǎyǐ shàng shù
ant climb tree
‘sautéed vermicelli with spicy minced pork’
Prosodically constrained compound formation
101
d. 雞蛋炒西紅柿
jı-dàn chăo xı-hóngshì
egg fry tomato
‘egg-fried tomatoes’
e. 西紅柿炒雞蛋
xı-hóngshì chăo jı-dàn
tomato fry egg
‘scrambled eggs with tomatoes’
The cuisine terminologies indicate that forming special technique terminologies is
different from forming ordinary words (compounds) in everyday speech (everyday
life without specific arts or techniques). Given this, it follows that particular fields
where arts or techniques are acquired may give rise to special terms in/for that
area (or semantic field). Gastronomy is a good example, and Chinese folk dancing
is another: the term èr’rénzhuàn 二人轉 ‘two people swing’ and shı-zi gǔn xiùqiú
獅子滾繡球 ‘lion turns the silk ball’ are special names for folklore performance
on stages performed by Chinese actors/actresses. The former has a Number + N
+ V sequence which can be analyzed as a rightward-footing [2+1] compound
with a SP structure while the latter has clearly an SVO structure with a mixture of
left- and right-ward footing. Both forms violate the prosodic constraint outlined
in section 4.3, but they are special usages created for specific purposes.
There seems to be a general rule that the morphological processes that form
words used in ordinary life by ordinary people are different from and independent
of those that are used by specialists to create words in particular fields for special
purposes (such as words created by chemists for chemistry studies, etc.).4 Of course,
there are varying degrees of being a specialist together with varying degrees of
being considered as a special discipline; thus, to what extent a degree of special
discipline can affect morphology is a new area deserving further study in the future.
4.4.2. Exceptions or evidence: the trisyllabic coordination
Aside from the exceptions created by specialists for special disciplines with
special rules, the coordinating trisyllabic compounds such as lăo zho-ng qı-ng
老中青 ‘old, middle, young; every generation’ are also seemingly exceptions to
the prosodic constraint proposed in 4.3, because, according to the FAP given in
(16), a trisyllabic string with a flat structure (i.e., [σσσ]) will be blocked by the
system, as seen in the structure (25), repeated here as (29).
29.
N’
Modifier
N
σ
σ
σ
[
Foot]
PrWd
← footing direction R-to-L
102 Prosodically constrained compound formation
As seen above, the leftward-footing process must be terminated so long as
footing is minimally satisfied. As a result, the leftmost word in a trisyllabic coordinating compound is left as an unfooted stray syllable – an incomplete result
of the footing process. It is incomplete because there is no further syllable to its
left for it to be grouped with, indicating that a trisyllabic coordinating string is
inoperable. Thus, theoretically, there would have been no [1+1+1] compounds
formed by this process. However, trisyllabic coordinating compounds do exist
in Chinese. For example,
30. a. go-ng‑nóng‑bı-ng 工農兵
‘workers, peasants and soldiers’
b. fú-lù-shòu 福祿壽
‘fortune, wealth, longevity’
c. dù-liàng-héng 度量衡
‘length, capacity and weight’
d. lù-hăi-ko-ng 陸海空
‘land, sea and air’
If coordinating compounds have a flat structure as is traditionally assumed, there
would have been no motivation for their existence as trisyllabic compounds.
By the FAP stated in (16), if a standard foot would locate its left edge at the
second syllable then no super-foot can be formed. However, given the analysis
in Feng (1995, 2000) and Zhang Ning (2006),5 the coordinating structure can
be formed binarily by the adjunction of a disyllabic coordinating compound
to a monosyllabic word, forming a new coordinating compound, as shown
below:
31.
N3
N1
N
|
lă o
old
N2
N
|
zhōng
middle
qīng
young
That is, the first two elements on the left are combined first, following foot
direction from right to left (or left to right, which makes no difference within a
[1+1] structure), and then they adjoin to the third element to their right, making a [2+1] coordinating structure. Given this structure, trisyllabic coordinating
compounds will be legitimately generated, since the left edge of the foot does
not break a PrWd. Hence, super-foot formation takes place, exactly as it does in
trisyllabic subordinating compounds given in (24). If coordinating compounds
Prosodically constrained compound formation
103
are all formed in this way, they will not be considered as exceptions for the
analysis given here.
The following examples show that trisyllabic coordinating compounds do
exhibit this structural property:
32. A-B-C
go-ng-nóng-bı-ng 工農兵
workers-peasants-soldiers
fú-lù-shòu 福祿壽
ortune, wealth, longevity
dú-liàng-héng 度量衡
length-capacity-weight
lù-hăi-ko-ng 陸海空
land-sea-air
shù-lı̆-huà 數理化
math-physics-chemistry
A-B
go-ng-nóng 工農
workers-peasants
fú-lù 福祿
fortune, wealth
dù-liàng 度量
measures
lù-hăi 陸海
land-sea
shù-lı̆ 數理
mathematical
B-C
*nóng-bı-ng 農兵
peasants-soldiers
*lù-shòu 祿壽
wealth, longevity
*liàng-héng 量衡
*capacity-weight
*hăi -ko-ng 海空
sea-air
lı̆-huà 理化
physics-chemistry
Note that there is no evidence that the last two items (i.e., BC) must first be
formed in the lexicon and then adjoin to the leftmost item, but there is strong
evidence that the first two items (i.e., AB) can be used independently in the
language as shown above. It is highly plausible to assume that the trisyllabic
coordinating structures are formed by adjunction of the first two items to the
last one. If this is so, the trisyllabic coordinating compounds are not exceptions
to our account; on the contrary, they are predicted by the theory presented here.
4.4.3. New discoveries from the so-called exceptions
Given the above analyses (25) and (28), it is expected that the [1+2] pattern of
compounds in ordinary speech are not generally acceptable unless they are special terminologies as seen in 4.4.1. However there are indeed [1+2] compounds
such as zhı̆ lăo hŭ 紙老虎 ‘paper tiger’. It is perfectly acceptable in everyday
speech and so is the existence of jı-n xiàngliàn 金項鏈 ‘gold necklace’. What
is going on here? We are facing a tough choice: either the [1+2] constraint is
wrong, or the [1+2] examples are exceptions.
As seen above, items in the lexicon are not all produced by morphological
rules in morphology: some are created purely by prosody (like phonological word ‘gonna, don’t’); some by idiomatization (e.g., mǎyǐ shàng shù 螞蟻上
樹 ‘vermicelli with spicy minced pork’); some by head-to-head movement in
the syntax (such as fàng-zài 放在 ‘put-on’); some by lexicalization (such as tian-zı̆ 天子 ‘the son of heaven, emperor’). The prosodic morphology in Chinese, however, distinguishes coordination from subordination, and thus, forms
of N+N, V+V, P+P, A+A are recognized as (compound) words whereas those
of N+V/A, A+N, V+N/R, P+N are phrases. The [2+1] forms can be generated in morphology as words (through a category-changing rule) if they obey
the prosodic constraints (i.e., the FAP) given in 4.3.1, while all trisyllabic [1+2]
104 Prosodically constrained compound formation
forms are phrases generated in syntax and they become wordhood through
some other ways, such as idiomatization, lexicalization, conflation, etc.
The above analysis receives support from the common assumption that the
combination of Noun+Noun (like ‘leather factory’) creates compound words,
while that of Adjective+Noun (‘big factory’) produces phrases in Chinese
(Duanmu 1990) as seen in previous chapters, for example:
33. 皮鞋工廠
pí-xié
go-ng-chǎng
leather-shoe
worker-mill
‘leather shoe
factory’
皮鞋工
pí-xié go-ng
leather-shoe
worker
‘leather shoe
worker’
皮廠
pí chǎng
皮工
pí go-ng
鞋工
xié go-ng
leather mill
leather worker
shoe worker shoe mill
‘leather
factory’
‘leather worker’
‘shoemaker’
皮鞋廠
píxié chǎng
leather-shoe mill
*鞋工廠
*xié go-ng-chǎng
shoe worker-mill
‘leather shoe
factory’
‘shoe-factory’
大皮鞋
dà píxié
big leather shoe
‘a big leather shoe’
鞋廠
xié chǎng
‘shoefactory’
*皮工廠
*pí go-ng-chǎng
leather
worker-mill
‘leather factory’
小工廠
xiǎo go-ng-chǎng
small worker-mill
‘a small factory’
The generalization of the footing-effect is this (Lü 1962, Lu and Duanmu
2002): N+N compounds favor the rhythmic pattern of [2+1] while the A+N
phrases prefer the [1+2] pattern.
Why then, there are N+NN compounds of the [1+2] phrasal pattern anyway? We certainly do not want to say that our theory is too beautiful to
be wrong, but we are convinced to commit ourselves to the fact that the
[N+NN] forms of [1+2] in Chinese morphology are absolutely not productive. This is so because the existence of jı-n xiàngliàn 金項鏈 ‘gold necklace’
and zhǐ lǎohǔ 紙老虎 ‘paper tiger’ does not permit jı-n ‘gold’ and zhǐ ‘paper’
to be freely used to create the [1+2] noun compound elsewhere, as shown in
the following examples:
34. a. *金工廠
*jı-n go-ngchǎng
gold factory
‘a gold factory’
b. *紙工廠
*zhǐ go-ngchǎng
paper factory
‘a paper factory’
Prosodically constrained compound formation
105
Why is jı-n xiàngliàn 金項鏈 ‘gold necklace’ acceptable but *jı-n go-ngchǎng
金工廠 ‘a gold factory’ is not? What is going on here? The unacceptable cases
are predicted while the acceptable ones have forced us to look for reasons as to
why they are different and deviated from what the theory predicts. After years
of research, we have finally found out, first, that there is a semantic contrast
between jı-n xiàngliàn 金項鏈 ‘gold necklace’ and jı-n go-ngchǎng 金工廠: the
meaning of ‘gold necklace’ can be rendered as ‘the necklace that is made of
gold’, while the semantics of “gold factory” means ‘the factory that produces
gold’. In other words, the meanings of ‘gold’ in ‘gold necklace’ and ‘gold factory’ are not the same. In addition to the fact that the surface meanings of ‘gold’
are different, there is something more going on internally: the jı-n 金 meaning
“made of gold” and the jı-n 金 meaning “producing gold” have two different
internal relationships with their head nouns. Technically speaking, ‘gold necklace’ and ‘gold factory’ have a different argument structure within the noun
compounds. In the former case, jı-n ‘gold’ is the material for necklace, whereas
in the latter case jı-n ‘gold’ is a product of the factory. The material gold and
product gold make the noun jı-n ‘gold’ syntactically different and so it is semantically. More precisely speaking, when gold is used with a meaning of “made
of ” as in jı-n xiàngliàn 金項鏈 “gold necklace,” it functions as a property classifying the head “necklace,” while when ‘gold’ is used to mean a ‘product’ as in
jı-n go-ngchǎng 金工廠 “gold factory,” it occurs in a position generated by compound formation. Compare:
35.
GOLD
Necklace
Figure 4.1 Property relation
Factory
produces
gold
Figure 4.2 Product relation
The property relationship entails that an entity contains A and B, while the
product relationship signifies that A gives birth to B, even if both relationships can
be expressed by phrasal syntax, i.e., Modifier + Noun (gold necklace) and Subject
+ (V) + Object (factory that produces gold). However, the latter expression does
not use ordinary phrasal syntax to express the statement that factory produces
gold. Rather, the phrasal expression is contracted into the compound word: factory that produces gold → gold factory. In other words, jı-n xiàngliàn 金項鏈
‘gold necklace’ observes a phrasal order but jı-n go-ngcháng 金工廠 ‘gold factory’
106 Prosodically constrained compound formation
has undergone a morphological (compound) formation, thus it does not observe
the phrasal order like jı-n xiàngliàn ‘gold necklace’ does. As a result, jı-n go-ngcháng
‘gold factory’ is a compound formed in morphology. This is why jı-n xiàngliàn
‘gold necklace’ can use the phrasal prosody of [1+2] to describe the head, and the
result is acceptable but jı-ngo-ngcháng ‘gold factory’ cannot use the [1+2] phrasal
prosody, and thus the [1+2] result is unacceptable.This can be explained properly
by prosodic word formation as seen in 4.3 which cannot produce [1+2] compound because of the footing direction. Since phrases can be generated by either
leftward or rightward footing, the word jı-n xiàngliàn ‘gold necklace’ is generated
in syntax and lexicalized in lexicon. In other words, the [1+2] forms like jı-n
xiàngliàn ‘gold necklace’ are not produced by morphology, and the [1+2] prosody
in jı-n-xiàngliàn 金項鏈 ‘gold necklace’ and zhı̆ lăohŭ 紙老虎 ‘paper tiger’ has
nothing to do with prosodic morphology, thus being not a violation of the FAP.
What is important to note here is the fact that if jı-n-go-ngchǎng 金工廠 ‘gold
factory’ is understood, ‘a factory that is made of gold’, even if the semantics are
unrealistic, the result is acceptable exactly like jı-n xiàngliàn 金项链 ‘gold necklace’ (the same is true with zhǐ go-ngchǎng 紙工廠 ‘paper factory’ if it is understood as ‘a factory that is made of paper’). Apparently, the so-called exceptions
(or counterexamples) are not true exceptions and they are indeed predicted
by the prosodic system that recognizes the phrasal semantics and compound
semantics by allowing the former with [1+2] and latter with [2+1], which
shows the grammatical function of foot directionality in the sense that a noun
in the leftmost position of [N+NN], such as jı-n ‘gold’ in jı-n xiàngliàn ‘gold
necklace’, will be forced to function as an adjective (i.e., N+NN → A+NN,
like gold → golden in English), denoting the property of the head noun and
hence escaping from being ruled out by FAP.
4.4.4. Conditioned exceptions
According to recent work in Huang and Duanmu (2013), Duanmu (2015), Feng
(2012, 2016), Huang (2015) and many others, new evidence and insightful explanations are offered to the seemingly exceptional cases and thus have deepened
the study in prosodic morphology. For example, a corpus-analysis (Duanmu
2011) shows that only 1.1% of N+N compounds in Chinese are formed by
following the [1+2] syllable pattern such as jı-n xiàngliàn 金項鏈 ‘gold necklace’,
zhǐ lǎohŭ 紙老虎 ‘paper tiger’ . . . etc. For example (taken from Duanmu 2012),
36. Token counts of 2+2, 2+1, 1+2 and 1+2 [N N] compounds
Pattern
2+2
2+1
1+2
1+1
All
Raw
1,325
508
1,479
1,020
4,332
Errors
46%
88%
46%
43%
Corrected
711
63
798
580
2,152
Supplemental
0
18
40
2,169
2,227
Final
711
81
838
2,749
4,379
%
16.2%
1.8%
19.1%
62.8%
100%
Prosodically constrained compound formation
107
60%
50%
40%
30%
20%
10%
0%
2+2
2+1
1+2
1+1
Figure 4.3 Length patterns of [N N] (token frequency)
It shows that even if the N+NN compounds do exist in the language, their
currents are not robust enough (1.1%) to serve as triggering experience for
next generation to come up with a categorical rule in the language. Thus,
the N+NN compounds may be automatically reanalyzed by speakers as
Modifier+NN phrases rather than the N+NN compounds, as argued in the
previous studies.
Secondly, among all the exceptions observed in the literature, there are [1+2]
compounds where the first element is a prefix-like one. Following Lü and
Rao (1981), the first element in the examples below (i.e., fù bùzhăng 副部長
‘vice-minister’, zhŭn qìyue- 準契約 ‘quasi-contract’) can be analyzed as prefixes,
rather than root morphemes.
37. fù bùzhăng
secondary-minister
‘vice-minister’
zhŭn qìyuelike-contract
‘quasi-contract’
This type of prefixes includes at least nán 男 ‘male’, nǚ 女 ‘female’, zhŭn 准
‘quasi’, zhŭ 主‘chief ’, fù 副 ‘vice’, xiăo 小 ‘small’, făn 反 ‘anti-’, lèi 類 ‘quasi’,
guì貴 ‘expensive’, hăo 好 ‘nice’, qián 前 ‘before’, gè 個 ‘each’, mĕi 每 ‘every’, etc.
(see among others, Chao, 1968; Lü and Rao, 1981; Liu 1991; Tang, 1988; see
also, Pan, 1993:129 and references cited there). These elements are sometimes
called quasi-prefixes (Liu, 1991) or semantic prefixes (Tang, 1988). It has also
been observed that the first elements in [l +2]MN compounds are very limited
in source and number (Lu and Duanmu, 1991), and that they exhibit a cluster
of properties in the grammar:
38. (i)Listablility (despite the fact that there are controversies about which
ones should be on the list);
108 Prosodically constrained compound formation
(ii) High frequency of occurrence (or versatility, see Chao, 1968);
(iii) Positional predictability (i.e., dìngwèicí 定位詞 “positionalized morphemes”, see Zhu, 1982).
Although there are different analyses about the prefixal nature of the morphemes mentioned above, the present theory goes with the affixation analysis
and provides a prosodic account for why the X in [X [Y Z]]MN forms should be
analyzed as prefixes. If the X in [X [Y Z]]MN is analyzed as a prefix, it will not be
considered as a part of compound produced by the compound formation rule
on the one hand, and the footing process given by the FAP will ignore it when
footing takes place within compound formation on the other hand. As a result,
the prefix element will be recognized as a case of extrametricality.
Thirdly, we also acknowledge that some of the [X [Y Z]]MN compounds do
not seem to be formed by prefixation. For example,
39. shu-da-i.zi 書呆子
book-blockhead
‘book worm’
táng hú.lu 糖葫蘆
sugar-calabash
‘sugar-coated haws on a stick’
In shu-da-i.zi ‘book blockhead, bookworm’, shu- 書 ‘book’ cannot be versatilely used to add to other phrases or compounds, even if the semantics seem
appropriate. For example, we have shu-diàn 書店 ‘book store; bookstore’. In
other words, shu- 書 ‘book’ violates the requirement of versatility of prefixation.
However, this type of trisyllabic compound exhibits another property, that is,
the disyllabic head does not form a full foot, because the second syllable of the
disyllabic head is neutralized (i.e., vowel reduction without tone): da-i.zi 呆子
‘blockhead’. As is known, some disyllabic forms with a neutral-toned syllable
have a length of only about one-syllable (Lin 1990). If the prosodic weight of
Chinese is primarily based on ‘length’ (Duanmu 1990; Lu and Duanmu, 1991;
and Lin, 1990), this type of disyllabic head can be considered as exception of
the system because neutralized disyllabic heads can no longer be analyzed as
a full foot. Thus, this type of [1+2] output does not violate the FAP as far as
‘length’ is concerned. The prosodic function of neutralization can be further
evidenced by examples given below:
40. 他可以看清楚黑板上的字。
ta- kĕyı̆ kàn qı-ngchu he-ibǎn shàng de zì.
he can see clear blackboard’s character
‘He can see clearly the characters on the blackboard.’
他可以把黑板上的字看清楚。
ta- kĕyı̆ bă he-ibăn shàng de zì kàn qı-ngchu.
he can see clear blackboard’s character
‘He can see the characters on the blackboard clearly.’
你要把問題想明白。
nı̆ yào bă wèntíxiǎng míngbai.
Prosodically constrained compound formation
109
you should BA question think bright
‘You should think about the question clearly.’
你要想明白問題。
nǐ yào xiǎng míngbai wèntí.
you should think bright question
‘You should think about the question clearly.’
你要把道理想透徹。
nǐ yào bǎ dào lǐxiǎng tòuchè.
you should BA reason think thorough
‘You should think the reasons thoroughly.’
你要想透徹*道理。
nǐ yào xiǎng tòuchè*dàolǐ.
you should think-thorough reason
‘You should think the reasons thoroughly.’
請你把道理說透徹。
qǐng nǐ bǎ dàolǐ shuo- tòuchè.
please you BA reason talk thorough
‘Please explain the reasons thoroughly.’
請你說透徹*道理。
qǐng nǐ shuo- tòuchè *dàolǐ.
please you talk thorough reason
‘Please explain the reasons thoroughly.’
In recent developments of Prosodic Grammar in Chinese (Feng 2017, Wang
C-S. 2017, Wang C. 2017), NS is assigned through the verb to its (resultative or
object) complement by following the so-called Government-based NSR, and
hence there is only one complement (an R, or an O, or an XP, but not both) is
allowed, giving rise to the following grammatical contrasts:
41. Ba+Object+V-RR
把道理想透徹
Bǎ dàolǐ xiǎng tòuchè
Ba reason think-thorough
‘Think the reasons thoroughly’
把道理說透徹
Bǎ dàolǐ shuo- tòuchè
Ba reason talk-thorough
‘explain the reasons thoroughly’
V-RR+Object
*想透徹道理
*xiǎng tòuchè dào lǐ
think-thorough reason
‘think the reasons thoroughly’
*說透徹道理
*shuo- tòuchè dàolǐ
talk-thorough reason
‘explain the reasons thoroughly’
However, when the resultative complement is monosyllabic or a monosyllable plus a neutralized syllable, the sentence is acceptable.The contrast between
two syllables and one (and a neutralized syllable) syllable/s indicates that a full
syllable and a full syllable plus a neutralized syllable are the same in terms of
the prosodic weight. It has been suggested that in situation like the above,
110 Prosodically constrained compound formation
neutralized syllables are prosodic variables, meaning that they can be analyzed as
extrametrical elements in certain contexts. Given the fact and the consideration
in previous studies, the [N + NN] (full-tone N + full-tone N with a neutralized N) will not be a counterexample to the present theory. On the contrary,
the characteristics of one syllable plus a neutralized syllable constitute a unique
property that supports the prosodic account of morphology in Chinese.
So far we have analyzed the most recognized exceptions and counterexamples
to the prosodic account for Chinese morphology, including (but not limited
to) special terminologies like lóng hǔ dòu 龍虎鬥 ‘Dragon Fighting against a
Tiger (a dish with a snake being cooked with a cat)’, trisyllabic coordination
like go-ngnóng bı-ng 工農兵 ‘workers, peasants and soldiers’, category-changing
cases like zhǐlǎohǔ 紙老虎 ‘paper tiger’, and prefixations fù jı-nglǐ 副經理 ‘vice
manager’ and neutralizations shu-da-izi 書呆子 ‘book worm’, etc. All of them
indicate that compounds in Chinese are prosodically governed, which gives rise
to the complexities of wordhood in Chinese, not only in its history as we say
in Chapter 2 and Chapter 3, but also in Mandarin Chinese as we have seen in
this chapter. The crucial point that needs to be addressed is that in facing the
complex wordhood situation in Chinese, which lacks what is normally observed
morphology in Indo-European languages, it is better not to use a few exceptions
to deny a well-tested pattern that exists in the language, but try to bear in mind
that different lexical items can obey different parameters, such as ‘very good’
[Modifier+Adj] vs. ‘good enough’ [Adj+Modifier] in English (Kayne 2005), and
to explore why, how and where the exceptional cases have occurred and what
the rules are governing them, thus figuring out their interacting with other rules
in Chinese morphology. By doing so, we may finally reach a full-fledged picture
of Chinese morphology, and contribute new ideas to general linguistics as well.
4.5. Conclusion
This chapter has discussed what a PrWd is and how it is realized in the Chinese prosodic morphological system. Furthermore, we have seen how compounds are derived through a specific prosodic morpho-syntactic mechanism.
It is argued that the notion of prosodic word determines the formation of
compounds in Chinese. Based on the new perspective, we are able to look at
Chinese morphology from a completely different point of view: compounds
are constrained by PrWds and PrWds are determined by foot formation. Under
this theory, we have reached the following conclusion: the system of Chinese
morphology would not be complete without taking prosody into account.
Based on previous studies such as Duanmu’s arguments about the length
of words in Chinese (Duanmu 1997, 2000, Huang and Duanmu 2013), L-J.
Wang’s observation about the morphological function of prosody (2014) and
many others, the present study provides new perspectives with many important
implications for Chinese morphology.
First, it is well known that Lü pointed out in 1979 that words in Chinese
cannot be too long and too complex. However, there is no solution to the question of how long is “not too long” and how complex is “not too complex”.
Prosodically constrained compound formation
111
Apparently, there was no theory about the length of words before. Given the
argument in this chapter, the maximality of word length can easily be derived
according to the definition of PrWds: if maximal PrWds have a length of three
syllables, then the maximal length of compound words must not be longer than
three syllables. As a consequence, the size or the length of a (compound) word
in Chinese must be limited to three syllables. Forms that are longer than three
syllables, for example, four-syllable expressions (Sìzìgé) are PrWd-compounds.
Expressions longer than four syllables will be analyzed as combinations of
disyllabic PrWds (or compounds) with trisyllabic PrWds (compounds), and
therefore prosodic phrases in prosodic phonology and complex compounds in
morphology. It should be noted that loan words such jia--lì-fú-ní-yǎ 加利福尼亞
‘California’ are outside of the indigenous system of rules.
Another important implication of this study is that the notion of PrWds
may provide new insights for traditional debate about the distinction between
words and phrases in the sense that if PrWds are formed by two morphemes or
words (with each morpheme or word corresponding to one syllable, as seen in
Chapter 2), the requirement of PrWds will be satisfied, regardless of whether
they are words or phrases. As a result, phrases such as niànshu- 念書 ‘read books’
look like words sometimes because they are PrWds – a linguistic unit defined
in prosodic morphology; words such as da-nxı-n 擔心 ‘carry heart, worry’ act like
phrases sometimes because they are phrases generated on syntactic tree structures and then become PrWds (by prosodic morphology) and compounds (by
morphology). Both of the words and phrases are motivated by the Foot Formation Rule and the PrWd formation. However, the process of forming PrWds is
by no means to destroy all the syntactic properties they originally have. Hence
their morphosyntactic behavior between word and phrase is expected because
most PrWds are idiomatized – a category between (free) phrases and words.
And most importantly this characteristic gives what has always been studied for
centuries about the elastic property of Chinese wordhood since Huang Kan
(1932), Guo (1938), Lü (1962), Lü (1963), Lu and Duanmu (1991/2002) and
about the syntactic property of being a word or phrase since Chao (1968), Lu
Zhiwei (1965), Pan WG (2004)6 and J. Packard (1997), etc. If the theory and
the procedural mechanisms given in this book are correct, the notion of PrWd
which is independently motivated by the prosodic grammatical system may
provide a key as well as a fundamental notion for Chinese morphology, as we
will see in next few chapters.
Notes
1 It should be noted that the contrast between rightward footing and leftward footing is
effective only in the domain of morphology; in syntax, however, both the rightward footing and the leftward footing are permissible.
2 See Wang HJ (2000).
3 See also Pei (2011, 2016), Zhuang (2011) and Zhuang and Liu (2012).
4 How the differences are distinguished among different special areas in terms of semantic
fields and what kind of rules are involved in what types of professions are all interesting
questions for morphological studies motivated by the theory developed here.
112 Prosodically constrained compound formation
5 See Zhang Ning (2006).
6 潘文國、葉步青、韓洋 (2004)《漢語的構詞法研究》 上海: 華東師範大學出版社。
References
Chao, Yuen-Ren. 1968. A Grammar of Spoken Chinese. Berkeley: University of California
Press.
Dong, Xiufang. 1998. “Prosodic Constraints on [[Verb+Resultative]+Object]]
述補帶賓語句式中的韻律制約”. Linguistic Study 語言研究 1: 55–62.
Duanmu, San. 1990. A Formal Study of Syllable, Tone, Stress and Domain in Chinese Languages.
Doctoral dissertation. MIT, Cambridge, MA.
Duanmu, San. 1997. “Phonologically Motivated Word Order Movement: Evidence from
Chinese Compounds”. Studies in Linguistic Sciences 27.1: 49–73.
Duanmu, San. 2000. The Phonology of Standard Chinese. Oxford: Oxford University Press.
Duanmu, San. 2011 [2016]. “Wordhood in Chinese”. In Wolfgang Behr, Gu Yueguo, Zev
Handel, C.-T. James Huang, and Rint Sybesma (eds.) Encyclopedia of Chinese Language and
Linguistics. Leiden: Brill.
Duanmu, San. 2012. “Word-Length Preferences in Chinese”. Journal of East Asian Linguistics
21.1: 89–114.
Duanmu, San. 2015. A Study of Elastic Word Length in Monomorphemic Nouns in Chinese:
‘Monosyllabic-only’ Nouns in the Lexicon and in Actual Use 漢語單語素名詞的詞長
彈性研究：詞典中和實際使用中的‘只單不雙’名詞. In Shengli, Feng (ed.) New Explorations in Chinese Prosodic Syntax 漢語韻律語法新探. Shanghai: Shanghai Zhongxi Book
Company, 4–21.
Feng, Shengli. 1995. Prosodic Structure and Prosodically Constrained Syntax in Chinese. Ph.D. dissertation. Philadelphia: UPENN.
Feng, Shengli. 2000. Prosodic Syntax in Chinese 漢語韻律句法學. Shanghai: Shanghai Educational Press.
Feng, Shengli. 2005. Prosodic Grammar in Chinese 漢語韻律語法研究. Beijing: Peking University Press.
Feng, Shengli. 2012.“Prosodic Variation and Grammatical Function of Neutral Tone in Mandarin Chinese 北京話的輕聲及其韻律變數的語法功能”. Linguistic Sciences 語言科學
6: 586–595.
Feng, Shengli. 2015. “Five Comments on ‘A Chinese Phonological Enigma’ by Professor
Geoffrey Sampson”. Journal of Chinese Linguistics 43: 734–739.
Feng, Shengli. 2016. Question-and-Answers About Prosodic Grammar in Chinese 漢語韻律
語法問答. Beijing: Beijing Language and Culture University Press.
Feng, Shengli. 2017. A Cartographical Account of Prosodic Syntax in Chinese. In Fuzhen
Si (ed.), Studies in Cartographic Syntax. Beijing: China Social Sciences Press. Forthcoming.
Guo, Shaoyu. 1938.“The Function of ElasticWord Length in Chinese 中國語詞之彈性作用”.
Yenching Journal of Chinese Studies 燕京學報 24: 1–34.
Huang, Kan. 1936 (2013). Notes on Paleography. In Huang Zhuo (ed.) Ancient Phonology and
Exegesis in Chinese 文字聲韻训诂笔记. Wuhan: Wuhan University Press.
Huang, Lijun and Duanmu, San. 2013. A quantitative study of elastic word length in Modern
Chinese 現代漢語詞長彈性的量化研究. Linguistic Sciences 語言科學12.1: 8–16.
Huang, Mei. 2015. Monosyllabic Word Used in Disyllabic Template in Chinese 漢語嵌偶單音詞.
Beijing: Beijing Language and Culture University Press.
Prosodically constrained compound formation
113
Kayne, Richard S. 2005. Some notes on comparative syntax, with special reference to English
and French. In Gugliemo Cinque and Richard S. Kayne (eds.) The Oxford Handbook of
Comparative Syntax. Oxford & New York: Oxford University Press.
Lin, Tao. 1990. Collected Words of Phonological Inquiry 語音探索集稿. Beijing: Beijing Language and Culture University Press.
Liu, Yongquan. 1991. “Difficulties in Chinese Information Processing and Ways to Their
Solution”. In Victor H. Mair and Yongquan Liu (eds.) Characters and Computers. Amsterdam: IOS Press.
Lu, Bingfu and Duanmu, San. 1991.“A Case Study of the Rhythm Between Rhythm and Syntax in Chinese”. Paper presented on the Third American Conference on Chinese Linguistics.
Lu, Bingfu and Duanmu, San. 2002.“Rhythm and Syntax in Chinese: A Case Study”. Journal
of the Chinese Language Teachers Association 37.2: 123–136.
Lu, Zhiwei. 1965 [1957/1964]. Chinese Morphology 漢語的構詞法. Beijing: Science Press.
Lü, Shuxiang. 1962. On ‘free’ and ‘bound’. Chinese Philology 中國語文 1: 1–6.
Lü, Shuxiang. 1963. “An Preliminary Research on Mono-Disyllabic Combinations in Modern Chinese 現代漢語雙音節問題初探”. Chinese Philology 中國語文1: 11–23.
Lü, Shuxiang. 1979. Problems in the Analysis of Chinese Grammar 漢語語法分析問題. Beijing:
Commercial Press.
Lü, Shuxiang and Changrong, Rao. 1981. On Non-predicate Adjectives 試論非謂形容詞.
Chinese Philology 中國語文 2: 81-85.
McCarthy, John and Prince, Alan. 1986. Prosodic Morphology. MS. University of Massachusetts
and Brandeis University. Unpublished manuscript.
Packard, Jerome Lee. (eds.) 1998. New Approches to Chinese Word Formation – Morphology, Phonology and the Lexicon in Modern and Ancient Chinese. Berlin: Mouton de Gruyter.
Pan, Wenguo, Yip, Po-Ching [Ye, Buqing] and Han, Yang. 1993. Studies of Chinese Word-Formation: 1898–1990 漢語的構詞法研究. Taipei: Xuesheng Shuju 學生書局.
Pan, Wenguo, Ye, Buqing and Han, Yang. 2004. Studies of Chinese Word-Formation
漢語的構詞法研究. Shanghai: East China Normal University Press.
Pei,Yulai. 2011. “The Study on Verb-Included Compound From the Perspective of Prosodic
Morphology”. Journal of Xi’an International Studies University 9: 41–52.
Pei,Yulai. 2016. “The Study on Syntax of Compound Word OVH”. Chinese Language Learning 2: 48–55.
Tang, Tingchi. 1988. Studies on Chinese Morphology and Syntax (2 Volumes) 漢語詞法句法
論集. Taipei: Xuesheng Shuju 學生書局.
Wang, Changsong. 2017. “A Study on Noncanonical wh Particles From the Syntax-Prosody
Interface – a Case Study of shenme in ‘V shenme (V)/(NP)’ construction”. Studies in
Prosodic Grammar 2.1: 73–100.
Wang, Chi. 2017. “Prosodic Constraints on the Predicate of Causative Constructions in
Mandarin Chinese”. Studies in Prosodic Grammar 2.1: 101–135.
Wang, Hongjun. 2000. “The Prosodic Word and Prosodic Phrase of Chinese”. 漢語的韻律
詞和韻律短語. Chinese Philology 中國語文 6: 525–536.
Wang, Lijuan. 2009. On the Morphological Function of Disyllablicity in Mandarin Chinese–A Case
Study of Nouns and Verbs 從名詞、動詞看現代漢語普通話雙音節的形態功能. Doctoral Dissertation. Beijing: Beijing Language and Culture University.
Wang, Lijuan. 2014. Morphological Prosody in Chinese 漢語的韻律形態. Beijing: Beijing Language and Culture University Press.
Wang, Lijuan. 2014. “A Prosodic Syntactic Study on[N de V]in Mandarin Chinese 漢語兩類[N的V]結構的韻律句法考察”. Chinese Teaching in the World
世界漢語教學 1: 70–77.
114 Prosodically constrained compound formation
Zhang, Ning. 2006. “On the Configuration Issue of Coordination”. Language and Linguistics
7.1: 175–223.
Zhu, Dexi. 1982. Teaching Materials on Syntax 語法講義. Beijing: Commercial Press.
Zhuang, Huibin and Liu, Z. Q. 2011. “Morphology of Chinese Synthetic Compounds and
Their Prosodic Constraint 漢語合成複合詞的構詞機制與韻律制約”. Chinese Teaching
in the World 4: 497–506.
Zhuang, Huibin and Liu, Z. Q. 2012. “On ‘de’: A Perspective in Chinese Prosodic Grammar
‘的’的韻律語法研究”. Chinese Language Learning 漢語學習 6: 34–42.
5Minimal and maximal
word effects
Minimal and maximal word effects are another type of important morphological phenomenon in Chinese morphology. They are of unique significance
in characterizing certain prosodic morphological behaviors in not only word
formation but also phrasal construction.
In this chapter I will explore the minimal word as well as maximal word
effects in Mandarin Chinese and demonstrate that the minimal and maximal
word phenomena in Chinese may provide different types of evidence supporting McCarthy and Prince’s minimality theorem, and most importantly, as I will
argue in this chapter, that the minimal word constraint can also be extended
to much broader applications such as compound formation and its interaction
with syntactic constructions in languages like Chinese: the so-called minimal
word effect is essentially morphosyntactic rather than merely morphological.
5.1. What is Minimal Word?
The notion of Minimal Word (MinWd, for short) is a fundamental concept in
the prosodic morphological system developed by McCarthy and Prince (M and
P) (1990). It is a prosodically circumscribed domain which may be selected as
the locus of morphological transformation in lieu of the whole domain (M &
P 1990, 1993, 1998). Theoretically, the notion of MinWd is derived from the
interaction of both Prosodic Hierarchy and Foot Binarity, as seen in Chapter 1
and restated in the following (taken from M and P 1998:284) for convenience:
1. Prosodic Hierarchy
Prosodic Word
Foot
Syllable
Mora
PrWd
|
Ft
|
σ
|
μ
116 Minimal and maximal word effects
2. Foot Binarity
Feet are binary under syllabic or moraic analysis.1
Since Prosodic Hierarchy impinges on every prosodic word so that each contains at least one foot, while Foot Binarity demands that every foot be bimoraic
or disyllabic, a prosodic word therefore must contain at least two moras or syllables according to the transitivity of the Prosodic Hierarchy. The Minimal Word
is therefore a single PrWd in the system. Note that the result of this reasoning
may end up with a simple statement: “A minimal word is just a foot.”Why, then,
do we need the notion of ‘minimal word’ if it is indeed a foot? As we will argue
in the section below (see also Chapter 2 of this book and Feng 2015 Minimal
Word in Chinese), the answer is this: there are different types of foot in Chinese
(i.e., degenerated foot, standard (disyllabic) foot, trisyllabic foot, bounded foot,
unbounded foot and so on), but only the disyllabic foot has morphological as
well as syntactic effects in the language. Thus, a general notion of foot is not
sufficient enough to capture the prosodically determined grammatical effects.
Furthermore, why is the disyllabic foot so special in the first place? As we have
seen above, without the Foot/PrWd alignment, there is no proper explanation
for why the size of a word (a morphological category) should coincide with a
foot (a prosodic category). In this book, we will follow M and P (1990, 1998)
and assume that the P >> M (prosody dominates morphology) is the fundamental hypothesis in prosodic morphology and that the Minimal Word is of
singular importance in characterizing a wide range of prosodic morphological
phenomena in languages known in previous studies, but also prosodic morphosyntatic phenomena, as seen in this book, in Mandarin Chinese.
Regarding the minimality condition of prosody on morphological output in
Chinese, we must be aware that there are complexities in Chinese morphology
(see the last chapter of this book for more discussion on this topic), which may
not be accounted for by one rule, namely, the MinWd constraint.2 However,
what I am going to demonstrate is this: there are phenomena that can only be
explained in terms of a minimal word analysis, which plays a central role in
Chinese prosodic morphology.
This chapter is organized as follows: section 2 consists of a study on the verbobject structure and shows that among all VO forms in the language, only the
ones that meet the minimal word requirement exhibit wordhood properties
while longer forms are all on a par with phrases. Section 3 shows that a process
of category changing from a [Aux V] verbal expression to an [Aux-V]adjective
compound is conditioned strictly on whether or not the [Aux V] is a minimal
word. Section 4 demonstrates that there is a clear distinction between MinWd
and non-MinWd [A+N] forms which are different syntactically as well and
that the Minimal Word functions as a condition for Morphosyntactic Operation. Section 5 is a summary of this chapter.
Minimal and maximal word effects
117
5.2. Minimal Word as a condition for VO compound
As seen in previous chapters, compounds are commonly formed out of VO
structures in both Classical and Modern Chinese. Although this type of compounding has already been recognized in Modern Chinese linguistics by various
morphologists, it is not clear until now how VO compounds are distinguished
from VO phrases in terms of their size differences, and this is done by the Templatic Constraint for VO compounding as formulated below:
3. Templatic Constraint on VO compounds
VO-Compd = PrWd
“The VO compound must be a prosodic word.”
Examples supporting this formulation are listed as follows:
4. 負責
fù-zé
carry-responsibility
‘be responsible for, be in
charge of ’
他負責保衛工作。
ta- fù-zé băowèi go-ngzuò.
he carry-responsibility security affairs
‘He is in charge of security affairs.’
關心
gua-n-xı-n
concern heart
‘concern, care about’
他不關心別人。
ta- bù gua-n-xı-n bié rén.
he not concern-heart others
‘He does not care about others.’
擔心
da-n-xı-n
carry heart
‘worry’
他擔心情況會有變化。
ta- da-n-xı-n qíngkuàng huì yŏu biànhuà.
he carry-heart situation will have change
‘He worries that the situation will change.’
When a VO form takes an object, it must be considered as a compound,
because the internal verb+object structure is syntactically opaque as far as the
phrase-structure condition is concerned (Huang 1982). Given this, all of the
above VO forms are doubtlessly compounds. Of course, not all disyllabic VO
compounds can take an ‘outer’ object. However, what is important to note here
is the categorical behavior of the trisyllabic (or polysyllabic) forms: No polysyllabic VO forms can take an ‘outer’ object. For instance,
5. a. *他負責任保衛工作。
*ta- fù-zérèn băowèi go-ngzuò.
he carry-responsibility security affair
‘He is in charge of the security affairs.’
b. *我開玩笑他。
*wŏ ka-i-wánxiào ta-.
118 Minimal and maximal word effects
I make-joke he
‘I make fun of him.’
Examples (4) and (5) provide a clear contrast between disyllabic VO compounds
and trisyllabic VO phrases, thus confirming the general observation that only
disyllabic VO forms can take an object, while all trisyllabic forms cannot.
6. [VO]adverb
並肩戰鬥
bìng-jia-n zhàndòu
[side-by-side]-shoulder fight
‘Fight side by side’
*並肩膀戰鬥
*bìng-jia-nbăng zhàndòu
[side-by-side]-shoulder fight
‘Fight side by side’
A related question arises as to why trisyllabic forms cannot take an object in
the way that disyllabic forms do. The answer lies in the function of the Nuclear
Stress Rule in Chinese.
5.3. Minimal Word as a condition for category changing
The minimal word effect can also be observed in [auxiliary+verb] compounds in Chinese: kěyí 可疑 ‘susceptible’, kě’ài 可愛 ‘lovable’, kěpà 可怕
‘afraid’, etc., are [Aux+V] compounds while kěyǐ huáiyí 可以懷疑 ‘can be
suspected’, kěyǐ hàipà 可以害怕 ‘can be afraid of ’ etc., are phrases. The distinction can be made clearly in terms of prosody and the condition is formulated as follows:
7. Minimal Word condition on categorical change
[kě+V] Adjective &frasl;
[kě+V]PrWd
“A [kě+V] form undergoes a process of category changing, if it is a prosodic word.”
Examples suporting this argument are given in (8):
8. a. 他非常可疑。
ta- fe-icháng kě-yí.
he extremely suspect-table
‘He is extremely suspect-able (suspicious).’
a’. *他非常可懷疑。
*ta- fe-icháng kě-huáiyí.
he extremely suspect-able
‘He is extremely suspect-able.’
Minimal and maximal word effects
119
b. 他非常可靠。
ta- fe-icháng kě-kào.
he extremely trust-able
‘He is extremely reliable.’
b’. *他非常可依靠。
*ta- fe-icháng kě yı-kào.
he extremely trust-able
‘He is extremely reliable.’
c.
他非常可信。
ta- fe-icháng kě-xìn.
he extremely trust-able
‘He is extremely trust-able.’ (reliable, trustworthy)
c’. *他非常可相信。
*ta- fe-icháng kě xia-ng-xìn.
he extremely trust-able
‘He is extremely trust-able.’
d. *這個東西非常可加工。
*Zhège do-ngxi fe-icháng kě jia-go-ng.
this thing extremely process-able
‘This thing is extremely process-able.’
The above examples show that only disyllabic [kě+V] forms can be used as
adjectives whereas all longer forms are either not found in the language or
ill-formed by this type of process, even if the verbs used in these two forms
(monosyllabic and disyllabic) are synonyms (i.e., yí = huáiyí ‘doubt’, kào=
yı-kào ‘rely’, xìn=xìnrèn ‘trust’). This shows clearly that all longer forms are
incapable of undergoing a category changing from [Aux V] verbal expressions to [Aux-V] adjectives. In other words, only disyllabic [Aux-V] forms
are allowed to form adjectives while the trisyllabic ones are prohibited
from doing so. The same effect can also be observed on another auxiliary
verb néng 能 ‘be able to’, for example,
9. a. 喬姆斯基很能產。
Qiáomǔsı-jı- hěn néng-chăn.
Chomsky very can-produce
‘Chomsky (is) very productive.’
b. 喬姆斯基很*能夠產。
c. Qiáomǔsjı- hěn *nénggòu-chăn.
Chomsky very can -produce
‘Chomsky (is) very productive.’
120 Minimal and maximal word effects
10. a. 我們學校的學生都很能幹。
wŏmen xuéxiào de xuésheng do-u hěn néng-gàn.
our school De student all very can-do
‘The students are all very capable in our school.’
b. *我們學校的學生都很能够幹。
*wŏmen xuéxiào de xuésheng do-u hěn nénggòu gàn.
our school De student all very can- do
‘The students are all very capable in our school.’
c. 學生 能夠 幹 什麼 工作?
Xuésheng nénggòu gàn shénme go-ngzuò?
student can do what job
‘What kind of job can students do?’
This clear-cut distinction once again evidences that trisyllabic forms are differentiated from their disyllabic counterparts in the language and this behavior
cannot be properly explained by purely morphological operation without taking prosody into account.
5.4. Minimal Word as a condition for morphosyntactic
operation
In Mandarin Chinese, there are many A(djective)+N(oun) compounds such as dàmı̆
大米 ‘big+rice, rice’, dàhàn 大漢 ‘big+man, burly fellow’, xiăobiànr 小辮兒 ‘little+plait,
pigtail’, etc.Traditionally, most of the [A+N] forms such as dà lăohŭ 大老虎 ‘big tiger’,
xiăo yŭsăn 小雨傘,‘little umbrella’, etc., have always been considered as phrases (Zhu
1980). However, Shih (1986), Dai (1992), Duanmu (1998, 2007), Sproat and Shih
(1991, 1996a), Chen (2000) and many other scholars have argued that the bare [A+N]
forms exemplified above should all be considered as compounds, rather than phrases.
The strongest evidence supporting this analysis is this: the adjectives in all [A+N]
forms cannot take a modifier like hěn 很 ‘very’ directly, for example:
11. * 很大樹
*hěn dà shù (very big tree)
‘a very big tree’
很大的樹
hěn dà de shù (very big De tree)
‘a very big tree’
*很大車
* hěn dà che- (very big vehicle)
‘a very big vehicle’
很大的車
hěn dà de che- (very big De vehicle)
‘a very big vehicle’
*很小雨傘
* hěn xiăo yŭsăn (very small
umbrella)
‘very small umbrella’
很小的雨傘
hěn xiăo de yŭsăn (very small De
umbrella)
‘very small umbrella’
Minimal and maximal word effects
121
Such syntactic behavior, therefore, forces one to conclude that the [A+N] forms
are not phrases for otherwise there is no reason why the adjectives cannot be
modified by hěn 很 ‘very’, as compared with English.
12. a little umbrella
big shoes
a blackboard
a smallpox
a blackbird
a bigfoot
a very little umbrella
very big shoes
*a very blackboard
*a very smallpox
*a rather blackbird
*a extremely bigfoot
In English, the adjectives of an [A+N] form cannot be modified if the [A+N]
is a compound. In Chinese, however, the adjectives in all [A+N] forms are not
allowed to be modified. Thus, it is reasonable to consider them as an N0 category (Sproat and Shih, 1991:571).
However, the [A+N] scenario is much more complicated than previously
observed. First, there is a clear distinction between different prosodic entities
among all [A+N] forms which have different syntactic behaviors.That is, disyllabic [A+N] forms (if not all) behave differently from the longer ones systematically. To see this, let us begin with Sproat and Shih’s (1991) generalization
about the ordering of “SIZE” and “COLOR” adjectives in noun phrases.3 First,
to observe:
13. Size
little
xiăo
小
big
dà
大
Color
black
he-i
黑
white
bái
白
Noun
umbrella
yŭsăn
雨傘
plate
pánzi
盤子
*black little umbrella
*he-i xiăo yŭsăn
黑小雨傘
*white big plate
*bái dà pánzi
白大盤子
Based on mounting evidence in different languages, Sproat and Shih (1991)
generalized an Adjectival Ordering Restriction (AOR), which says essentially
that: the AOR – [SIZE > COLOR] – holds if the adjectives involved are
hierarchical direct modifiers. This is argued to be a universal constraint for the
ordering of multiple adjectival modifiers, which is therefore valid not only
in English but also Chinese (and many other languages, see Sproat and Shih
1991).
Under this constraint, a clear-cut distinction emerges between disyllabic
[A+N] forms and trisyllabic [A+N] forms in Chinese. For example (taken from
Feng, 2000):
122 Minimal and maximal word effects
14.
Trisyllablic AN
*COLOR > SIZE
* 白大盤子
* bái dà pánzi
‘white big plate’
Disyllablic AN
COLOR > SIZE
紅小兵
hóng xiăo-bı-ng
‘red small-soldier, the red guard’
*紅小雨傘
*hóng xiăo yŭsăn
‘red small umbrella’
黑大漢
he-i dà-hàn
‘black big-man, a black bully’
*紅小計算機
*hóng xiăo jìsuànjı‘red small computer’
黑小辮兒
he-i xiăo-biànr
‘black little-plait, a black pigtail’
*黑大熊貓
*he-i dà xióngma-o
‘black big panda’
黑大雁
he-i dà-yàn
‘black big-goose, wild goose’
*白大蘿蔔
* bái dà luóbo
‘white big radish’
白大米
bái dà-mı̆
‘white big-rice, white rice’
*紫小蕃茄
*zı̆ xiăo fánqié
‘violet small tomato’
綠小蔥
lǜxiăo-co-ng
‘green small-scallion, green scallion’
A preliminary generalization can be drawn from the above facts as follows:
only disyllabic [A+N] forms can violate the universal constraint of [SIZE >
COLOR] whereas three (or more)-syllable [AA+N] (where ‘AA’ represents a
disyllabic adjective) forms must all obey this constraint (cf. red smallpox, black
bigfoot in English). Based on the generalization given above, the Minimal Word
condition for [A+N] can be formulated as follows:
15. Minimal Word constraint on [A+N] forms in Chinese
AN-Compd = PrWd
An [A+N] (lexical) compound is a PrWd.
The evidence here provides additional facts for the argument that the Minimal
Word in Chinese is the most harmonic prosodic form which is extremely active
in a variety of constructions in the language.
It is worthwhile to note that the prosodic effect between different types of
[A+N] gives support to, while at the same time receives support from, the analysis of disyllabic [A+N] forms as X0 category in syntax. In other words, there
are two different types of zero-level [A+N] forms in Chinese: one is lexicalized
in the lexicon and the other is formed in syntax, according to Feng’s (2001)
analysis that argues against the analysis of de-less modifier-head structures as
Minimal and maximal word effects
123
e.g., dàpánzi 大盤子 ‘big plate’ on a par with [A+N] compounds as dà-guàr
大褂兒‘unlined long gown’. The argument is based on the fact as seen above,
namely, when a modifier such as bái 白 ‘white’ is added to the noun dà-guàr
大褂兒 ‘unlined long gown’, different ordering patterns can be derived which
show a contrast of grammaticality:
16. a. dà bái pánzi 大白盤子
big white plate
‘a big white plate’
a’. *bái dà pánzi 白大盤子
white big plate
‘a big white plate’
b. bái dà-guàr 白大褂兒
white big-gown
‘a white unlined long gown’
b’. dà-bái guàr 大白褂兒
big-white gown
‘a white unlined long gown’
Given that the ordering restrictions for modifiers apply in syntax and that a
modifier relating to color must be nearer to the head noun as a modifier relating
to size, i.e., Adjectival Ordering Restriction (AOR), Feng concludes that dà-guàr
大褂兒 ‘unlined long gown’ is a compound. This is so because its internal structure is opaque to the AOR and hence (16b) is grammatical. Dà pánzi 大盤子 ‘big
plate’ in (16), however, obeys the AOR and therefore must be distinguished from
compounds like dà-guàr 大褂兒 ‘unlined long gown’ and the ungrammaticality
is due to a violation of the Lexical Integrity Hypothesis. As pointed out by Paul
(2005), Feng’s (2001) analysis captures the contrast between [A+N] compounds,
on the one hand, and de-less [A+N] modifier-head structures, on the other hand.
However, contrary to [A+N] phrasal analysis, Feng does not take this contrast as evidence in favor of the pure phrasal anaylsis of the de-less modifierhead structure crucially because of the fact that all [modifier+A+N] forms are
ungrammatical like *very blackboard in English.
However, there is one type of example cited by Paul (2005) challenging
Feng’s analysis, which is shown as follows:
17. a. 我覺得黃襯衫比紅的好看。
wŏ juéde [NP huáng chènsha-n] bı̆ [NP hóng -de Ø] hăokàn.
1SG think yellow shirt compared:to red - De pretty
‘I think that yellow shirts are prettier than red ones.’
b. 阿美 不 喜歡 黃玫瑰, 紅的 還可以。
aměi bù xı̆hua-n huáng méiguı-, hóng-de hái kěyı̆.
Amei NEG like yellow rose red - De still acceptable
‘Amei doesn’t like yellow roses, red ones are still ok.’
124 Minimal and maximal word effects
c. 不買大螃蟹， 買小的。
bù măi dà pángxiè, măi xiăo-de.
NEG buy big crab buy small- De
‘Don’t buy a big crab, buy a small one.’
d. 不買木頭桌子，買鐵的。
bù măi mùtou zhuo-zi, măi tiě de Ø.
not buy wood table buy iron De Ø
‘Don’t buy a wood table; buy an iron one.’
e. *阿美不想吃紅花，黃的還可以。
*Aměi bù xiăng chı- [N° hóng-hua-], [NP huáng-de] hái kěyı̆.
Amei NEG want eat red-flower=safflower yellow- De still acceptable
‘Amei doesn’t want to take safflower, but yellow ones are still ok.’
f. 張三不喜歡木頭桌子，鐵的還可以。
Zha-ngsa-n bù xı̆huan mùtou zhuo-zi, tiě de hái kěyı̆.
Zhangsan NEG like use wood table iron-De still ok
‘Zhangsan doesn’t like using wooden tables, but iron ones are still ok.’
According to the examples given above, Paul argues that Feng’s analysis cannot
be correct because the head noun of the [A+N] sequence in the above examples is visible to anaphoric rules operating on the phrase level. Accordingly, she
concludes that Feng’s (2001) analysis of the de-less modifier-head structure as a
(syntactic) compound wrongly predicts that the head noun in such a structure
is inaccessible.
Paul’s analysis and her solution are based on the anaphoric rule that assumingly operates on the phrase level exhaustively. However, there are some
problems with her account. Duanmu (2007) points out that this analysis is
problematic. Consider again the example Paul has given:
18. 不買木頭桌子，買鐵的。
Bù măi mùtou zhuo-zi, măi tiě de Ø.
not buy wood table buy iron De Ø
‘Don’t buy a wood table; buy an iron one.’
Paul argues that the empty noun Ø in the second clause refers to ‘table’ in the
first clause. If mùtou zhuo-zi 木頭桌子 ‘wood table’ is a compound, its internal
structure should be invisible (protected by the Lexical Integrity Hypothesis),
and the empty noun can only refer to ‘wood table’ as a whole instead of ‘table’
alone. However, it is not clear whether structures like (18) are a good test for
compounds. Consider a parallel example in English, shown in (19).
19. I don’t like meat pies; I like the one without meat.
“By all standards,” Duanmu argues (2007:117–118), “meat pie is a compound,
however, the anaphor one in (19) can still refer to ‘pie’, instead of ‘meat pie’
Minimal and maximal word effects
125
as a whole. This shows that structures like (18) and (19) are not good tests for
compounds, and there is no compelling evidence for treating [Modifier Noun]
as phrases in Chinese.”
There are more examples showing that referential devices involved here are
not necessarily of phrasal nature. Note the following examples:
20. a. *木頭桌子和鐵的，我都喜歡。
*mùtou zhuo-zi i hé tiě de Ø, wǒ do-u xîhuan.
wood table and iron De Ø I all like
‘I like both wood tables and iron ones.’
b. *我買了幾張木頭桌子和鐵的。
*wǒ mǎi le jîzha-ng mùtou zhuo-zi i hé tiě de Ø.
I buy Asp serveral Cl wood table and iron De Ø
‘I bought several wood tables and iron ones.’
In the [NP & NP] phrasal environment, the head noun of the [A+N] sequence
is not visible to anaphoric rules operating on the phrase level, indicating that
[A+N] forms are not phrasal in nature, thus arguing against Paul’s analysis. The
following examples Paul has mentioned in her study also disapprove a phrasal
analysis of the [A+N] sequence (21a).
21. a. ??一件小的大褂兒
??yı--jiàn xiăo de dà- guàr
1-Cl small De big-gown
‘a smaller big-gown’
b. 我買了一件大褂兒， *他買了一件小的。
wŏ măi-le yı--jiàn dà- guàr, *ta- măi-le yı--jiàn xiăo-de.
1SG buy-Asp-Cl big-gown 3SG buy-Asp- Cl small-De
‘I bought an unlined long gown, and he bought a small one.’
All Mandarin informants the author consulted have confirmed the judgment
for (21a) while the same informants rejected the second clause of (21b), which
clearly shows that the elements within the compound dàguàr 大褂儿 ‘unlined
long gown’ are NOT visible to the anaphoric rules operating on the phrase
level.
Two questions arise then. On the one hand: why is the head noun zhuo-zi in
mùtou zhuo-zi 木頭桌子 ‘wood table’ visible to anaphoric rules but guàr in dà
guàr 大褂兒 ‘unlined long gown’ is not, if both are [A+N] compounds? What
is the difference between these two [A+N] forms? On the other hand, why
are both head nouns are NOT visible to anaphoric rules when occurring
in [NP & NP] environments? What are the common features that they share
syntactically?
In this chapter, I will follow Feng’s (2001) analysis to provide a plausible
account for the questions given above. That is, instead of treating all the [A+N]
126 Minimal and maximal word effects
forms as compounds (Duanmu 2007) or as compounds for some and as phrases
for others (Paul 2005), I will insist on the treatment of making a distinction
between lexical compounds and syntactic compounds first, and then I will further propose two different derivations in order to account for the differences
observed in (16), repeated here for convenience:
22. a. 白大褂兒
bái dà-guàr
white big-gown
‘a white unlined long gown’
a’. 大白褂兒
dà bái guàr
big-white gown
‘a big-white gown’
b. 大白盤子
dà bái pánzi
big white plate
‘a big white plate’
b’. *白大盤子
*bái dà pánzi
white big plate
‘a big-white plate’
The argument presented here is this: dà-guàr 大褂儿 ‘unlined long gown’ is
a lexical compound formed (or lexicalized through idiomatization processes
as discussed in Chapter 2) in the lexicon and its internal structure cannot be
affected by phrase-level rules accordingly. Dà pánzi 大盤子 ‘big plate’, by contrast, is formed in the syntax by head-to-head adjunction, a similar operation
proposed by Shi (2002) as well with the following structure (see Zhuang 2015
for detailed analysis).
23. a. Syntactic word
N0
A0
|
བྷ
N0
|
㺓
b. Syntactic phrase
NP
AP
very black
N0
|
bird
Minimal and maximal word effects
127
A phrase created through syntactic operation can be expanded, while a compound created through syntax cannot be separated. This is because, according
to Feng (2001b) and Shi (2002), the adjunction of the adjectival head to the
noun head in each step results in an n0, i.e., bái pánzi 白盤子 ‘white plate’ and
dà bái pánzi 大白盤子 ‘big white plate’ are both analyzed as compounds: [n0 bái
[n0 pánzi]], [n0 dà [n0 bái pánzi]]. This entails that the difference between syntactic phrases and syntactic words lies crucially in the fact that in the syntactic
word adjunction, only lexemes may occupy both positions (the head and the
adjoined) and the result of the adjunction is still a lexeme, i.e., a syntactic atom
(X0) that occupies a single node in the syntactic tree. Whether there are structures like (23b) in Chinese is another question beyond the discussion here. For
the argument given here, it is important that, in my proposal of the syntactic
operation for compound formation that ordering restrictions can apply in the
case of two adjectives, where the adjective related to size precedes the color
term. This will give rise to the distinction between lexical compounds and
syntactic compounds since the former violate the AOR while the latter do not,
shown as follows:
24. Lexical word
你喜歡大衣，*我喜歡小的。
nı̆ xı̆ -hua-n dà-yı-, *wŏ xı̆ -hua-n xiăo-de.
1 SG like overcoat, *1 SG like small-De.
‘You like overcoat, I like a small one.’
Syntactic word
你喜歡大鞋， 我喜歡小的。
nı̆ xı̆ -hua-n dà-xié, wŏ xı̆ -hua-n xiăo-de.
2SG like big-shoe, 1SG like small-De
‘You like big shoes, and I like a small one.’
AOR dàyı- 大衣 ‘overcoat’
miăn dàyı- 棉大衣 ‘cotton-padded overcoat’
dàxié 大鞋 ‘a big shoe’
*miăn dàxié 棉大鞋 ‘a big cotton-padded shoe’
Given this analysis, there will be two types of [A+N] compounds: one is
syntactic and hence visible to anaphoric rules in certain syntactic environments, while the other is lexical and therefore opaque to all syntactic
operations (including anaphoric rules). This prediction is supported by the
following examples that strongly argue for the analysis of two types of
compounds.
25. a. John likes both meat-piesi and the onei without meat.
b. *Meat 0i and vegetable dumplings are both tasty.
c. 棉鞋和布鞋都買了。
128 Minimal and maximal word effects
mián-xié hé bù-xié do-u măi le.
cotton shoes and cloth shoes all buy Asp.
‘(I) bought both cotton shoes and cloth shoes.’
d. *棉和布鞋都買了。
*mián hé bù-xié do-u măi le
cotton and cloth shoes all buy Asp.
‘(I) bought both cotton shoes and cloth shoes.’
e. *布和棉鞋都買了。
*bù hé mián-xié do-u măi le.
cloth and cotton shoes all buy Asp.
‘(. . .) bought both cotton shoes and cloth shoes.’
f. 鐵鋸和鋼鋸都有。
tiě-jù hé gang-jù do-u yŏu.
iron saw and hacksaw all have
‘(. . .) have both iron saw and hacksaw.’
g. *鐵和鋼鋸都有。
*tiě hé gang-jù do-u yŏu.
iron and hacksaw
‘(. . .) have both iron saw and hacksaw.’
h. 車鑰匙和門鑰匙都在這兒。
che- yàoshi hé mén yàoshi do-u zài zhè-er.
car key and door key all at here
‘Both the car key and the door key are here.’
i. *車和門鑰匙都在這兒。
*che- hé mén yàoshi do-u zài zhè-er.
car and door key all at here
‘Both the car key and the door key are here.’
Only in contrastive environments like (17) the anaphoric function in syntactic compound is visible but the modifier adjective cannot be separated (or
moved) from its head noun even in syntactic compound because syntactic
operations are not allowed to apply to any element inside an X0 category.
Unlike syntactic compounds, the lexical compounds behave differently since
they strictly follow the Lexical Integrity Hypothesis (LIH):
26. 我買了一件大褂兒，*他買了一件小的。
wŏ măi-le yı--jiàn dà-guàr, *ta- măi-le
yı--jiàn xiăo-de.
1SG buy-Asp-Cl big-gown 3SG buy-Asp-Cl small-De
‘I bought an unlined long gown, (he bought a small one).’
The present theory can naturally accommodate both types of compounds with
judgments on anaphoric behaviors by native speakers under the X0 hypothesis
and the LIH for the syntactic part of the [A+N] compounding. When it comes
Minimal and maximal word effects
129
to the [SIZE+COLOR+N] ordering restrictions, the current theory also predicts nicely that only lexical compounds are exempt from the AOR restriction and all syntactic compounds are required to obey this rule, since they are
generated via syntactic X0-adjunction which is on a par with Huang, Li and
Li’s (2009) observation that Chinese syntax operates on bare lexeme, i.e., the
syntactic atom (X0) that occupies a single node in the syntactic tree.
5.3. Minimal Word effects in verb reduplication
In Chinese there is so-called verb reduplication, turning the action into a more
intimate casual activity. For example,
27. 看看 kàn-kan ‘look-look, take a look casually’
嘗嘗 cháng-chang ‘taste-taste, have a bite casually’
The verb reduplication forms can all be altered with a yı- 一 ‘one’ word in
between, i.e.,
28. 看一看 look one look ‘take a look (informally)’
嘗一嘗 taste one taste ‘have a bite (informally)’
It is commonly assumed that the V-V reduplications are phrases abbreviated
from [V [one V]] structure, i.e.,
29. a. kàn yı- kan 看一看
Look one look ‘take a look’
b. kàn le yı- kan 看了一看
Look Asp one look ‘took a look’
a. cháng yı- chang 嘗一嘗
b. cháng le yı- chang 嘗了一嘗
kàn-kan 看看
look-look ‘take a look’
kàn le kan 看一看
look Asp. look ‘took a look’
cháng-chang 嘗嘗
cháng le chang 嘗了嘗
The [V [one V]] constructions can actually be analyzed as originating from
[take a V/N] structure, pretty much like “take a look/bit” or “give a try” in English.What is the difference in Chinese from that in English? It is held that a verbcopying process is involved in Chinese but not in English, pretty much in the
sense of Baker (1988) and Hale and Kayser (1993), within the following structures:
30. a.
b.
c.
d.
take
give
DO
kàni
looki
a
a
yıone
a
look
try
kàn = [VP DO [a V/N]DP]
kàni = [VP [Vi]DO [a V/Ni]DP]
looki
In Chinese, instead of using a phonologically realized light verb like ‘have’,
‘make’, ‘do’, etc., a covert light verb DO is used in the same position as ‘take’ in
130 Minimal and maximal word effects
‘take a look’. Since the DO is phonologically empty thus it must be supported by
a phonologically realized element in order to ensure its position, a verb-copying
process takes place, giving rise to a structure of ‘look a look’ (=kàn yı- kàn). The
question why the DO position must be phonologically realized, why the DO
position needs to copy the following verb are clear at this moment. However,
what is important to note here is the fact that the copied verb carrries the stress
while the original verb is weakened almost to a neutralized syllable, for example,
31. 吃吃 CHI .chi
嘗嘗 CHÁNG.chang
看看 KÀN.kan
*chı- CHI
*cháng CHÁNG
*kàn KÀN
And this is also true in the [V one V] pattern:
32. a. CHI . yı-. chi 吃一吃
CHÁNG. yı-. chang 嘗一嘗
KÀN. yı-. kan 看一看
*chı-. yı- CHI
*cháng. yı- CHÁNG
*kàn. yı- KÀN
b. CHI . le. chi 吃了吃
CHÁNG. le. chang 嘗了嘗
KÀN. le. kan 看了看
*chı-. le CHI
*cháng. le CHÁNG
*kàn. le KÀN
c. CHI . le. yı-. chi 吃了一吃
CHÁNG. le. yı-. chang 嘗了一嘗
KÀN. le. yı-.kan 看了一看
*chı-. le. yı- CHI
*cháng. le. yı- CHÁNG
*kàn. le. yı- KÀN
The left strong pattern goes perfectly with its register property, i.e., the V.v (a
limerick prosody, see Chapter 6), exclusively occurs in casual speech. Furthermore, in natural speech, the particle yı-一 ‘a’ is weakened and finally dropped/
deleted, so that we have a shorter form that looks like a reduplicated verb as
seen in (32b). In other words, the so called verb reduplication is indeed a result
of prosodic abbreviation.
Given the above analysis we are ready to see the Minimal Word effect in the
verb-copying process. It has been observed that the yı--deletion can only be
operated within two monosyllabic verbs but not within two disyllabic ones, as
exemplified below.
33. cháng-yı--chang 嘗一嘗
taste-one-taste
‘have a bite’
tán-yı--tan 談一談
talk-one-talk
‘have a chat’
*pı̆ncháng-yı--pı̆ncháng 品嘗一品嘗
*savor-one-savor
‘try to savor it’
* jia-otán-yı--jia-otán 交談一交談
*converse-one-converse
‘have a chat’
Minimal and maximal word effects
131
However, it is not true that disyllabic verbs are barred from being reduplicated at all, because the following reduplication of VV forms are perfectly
acceptable:
34. kăolü kăolü
think-think
fe-nxi fe-nxi
analyze-analyze
考慮考慮
‘think (it) over’
分析分析
‘try to anlyze it’
Disyllabic verbs do have the abbreviated/reduplicated forms. If it is so, why
cannot disyllabic verbs have the full form with a structure of [DO/GIVE/TAKE
+an action]? There is no proper explanation for why disyllabic verbs cannot
co-occur with the yı- as monosyllabic verbs do when reduplication takes place.
At first glance, the answer is quite simple, since the VO structure is right-strong
and disyllabic verbs ([σσ]V) cannot easily be weakened to form a ([σσ]S [σσ]W)
rhythmic pattern, this is why they cannot undergo the verb reduplication. However, as seen above, disyllabic verbs (i.e., Vσσ) do have a [[V σsσw]s - [V σsσw]w]
pattern, so it is not the case that the prosodic weight of disyllabic verbs cannot
be reduced at a weakened position. Regarding all the possibilities, the best and
simplest answer to the above question is to invoke the minimality condition, that
is, verb reduplication4 in Chinese must have a minimally two- and maximally
three-syllable templates; this is why yı- is not acceptable between disyllabic verbs,
because if we assume with Duanmu (2000) that each full-tone syllable contains
two moras while a neutralized syllable has one mora, then the legitimate [[V
σsσw]s -[V σsσw]w] pattern, fe-n.xi-fe-n.xi 分析分析 ‘analyze it a little’ for example,
the two [σw] (i.e., the two .xis in fe-n.xı-) both contain only one mora (i.e., a single
μ), making the string [[V σsσw]s -[V σsσw]w] exactly six moras, which equals three
syllables long. If, however, the particle yı- appears between the disyllabic verbs,
i.e., [σσ]V-yı- -[σσ]V] (fe-n.xı- yı- fe-n.xı-), the total number of moras will go beyond
six (does not matter if yı- is a monomoraic or bimoraic syllable),5 it will always be
a violation of the maximality condition defined by super-foot formation.This is
the reason why disyllabic verbs can be reduplicated but cannot take the yı- form.
It is not because of syntax or morphology, but rather because of the minimal and
maximal constraints on reduplications (morphosyntax) in Chinese morphology.
If the current analysis is correct, then non-neutralized disyllabic verbs will
behave differently from the neutralized ones and this is borne out by the following examples:
35. *ka-ichú-ka-ichú 開除開除 [σσ-σσ] ‘fired’
*bàgo-ng-bàgo-ng 罷工罷工 [σσ-σσ] ‘strike’
*liàn ’ài-liàn ’ài 戀愛戀愛 [σσ-σσ] ‘be in love’
Disyllabic verb reduplication is not acceptable if the second syllable of the
verb is not neutralized, because, as the theory predicts, the four syllables in
132 Minimal and maximal word effects
disyllabic verb reduplication will be a severe violation of the trisyllabic maximality constraints on PrWds which prohibits any length of more than three
syllables in the relevant morphosyntactic operations.
5.8. Summary
In this chapter, I have discussed several different syntactic and morphological
constructions, namely, the [Verb+Object] forms, the [Auxiliary+Verb] forms
and the [Adjective+Noun] forms. It is seen that all these different syntactic
forms share a common property, that is, the trisyllabic ones are syntactically
different from the disyllabic ones systematically. To be more specific, only the
disyllabic ones exhibit lexical properties, hence being treated as (lexical) compounds. Thus, for [V+O] forms, only the disyllabic ones can take an object; for
[Aux+V] forms, only the disyllabic ones can be modified by a degree modifier
like fe-icháng 非常 ‘extremely’; for [A+N] forms, only the disyllabic ones can
violate the [SIZE+COLOR+N] phrasal constraint. Contrary to the case of
disyllabicity, all trisyllabic forms in these three constructions deviate syntactically from the properties of being a (true) lexical item: the separable property
and the inability to take a (post-verbal) object for all trisyllabic [V+O] forms;
the inability to undergo the process of category changing for all trisyllabic
[Aux+V] forms; and the obligation to strictly follow the phrasal constraint
[SIZE+COLOR+N] for all trisyllabic [A+N] forms.
It is possible, of course, for one to suggest a different analysis for the trisyllabic
forms. However, no matter how one analyzes them, the distinction between the
disyllabic and the trisyllabic structures will still be there. The syntactic contrast
cannot be explained away by any analysis that overlooks the disyllabic and the
trisyllabic distinctions. The facts brought to light in this chapter are quite striking: the distinctions among all three types of forms are not only syntactic, but
also prosodic. In fact, their different syntactic behaviors have not been discovered and fully understood until the prosodic distinction is brought to light. In
this sense, it is reasonable to say that the study of prosody has revealed some
important prosodic-syntactic phenomena that would otherwise be a mystery
in the language.
As we have seen, the prosodic distinction in all three different types of structures is centered upon the basic domain of disyllabicity. Why is this so? The
Minimal Word Theorem provides the best explanation and it is simply the legitimacy of being a PrWd in the language. The Chinese language is, therefore, a
language that is sensitive to the prosodic domain. Actually, it is the domain that
permits only disyllabic forms to have the priority to be or become compounds,
and it is also the domain that prevents trisyllabic forms from entering into the
realm of lexical compounding (trisyllabic [N+N] compounds are permitted
but quite limited as seen in Chapter 4). As a result, a boundary in Chinese
morphology can be set between what is a compound (PrWd applies in morphology) or permitted to be a compound (PrWd applies in syntax), and what
is forbidden from being a lexical compound, or at most a syntactic compound.
Minimal and maximal word effects
133
It is well known that the core area of prosodic morphology in previous
investigations has concerned with mainly reduplication and infixation. This
study, however, extends the notion of PrWd into the area of compounding
and its interaction with syntax. For all three types of forms discussed above, we
have seen that the minimality constraint indeed dominates the word formation of compounding in Chinese. That is, a compound (or more specifically a
lexical compound) must first be a PrWd, even if a PrWd is not, by necessity, a
compound. This conclusion has several implications in prosodic morphology.
First, the Templatic Constraint not only determines morphological operations
like reduplication and infixation, but also dominates the word formation of
compounding in languages like Chinese. This raises a question for future study
as to why and how the Templatic Constraint could also control compounding. Furthermore, it is clear, by now, that the Minimal Word requirement plays
a role in Chinese. Yet, when it does, it not only dominates word formation,
but also affects syntactic structures in a way that influences the formation of
well-formed sentences. Surprisingly, it seems that the MinWd may also circumscribe certain syntactic phrases, so that the native speaker’s intuition about
what is considered to be a “word” is unquestionably affected by the prosodic
notion of MinWd. The final question then is how and to what extent prosody
affects the grammar of the language, a question we will explore in the next
chapter.
Notes
1 See reference to the Relative Prominence Principle proposed by Liberman and Prince
1977.
2 For example, there are many trisyllabic compounds as seen in Chapter 4 as well as quadrisyllabic compounds discussed in Chapter 6; they are obviously beyond the size of a PrWd.
In fact, the Parse-all-Syllable requires that every form be fully footed, which demands
multiple feet in longer words.Yet, the All-Foot-Left requirement will never be completely
satisfied in words which have more than one foot. Now, under minimal violation of AllFoot-Left, a multi-foot form is allowed but it must have its feet as close to the beginning
of the word as possible (see M and P, 1998:298). Given this and the language-specific
constraint that every syllable is a morpheme in Chinese, polysyllabic compounds are
allowed and probably predicted: the trisyllabic (σσ)σ) compounds will be more optimal
than (σ(σσ)) because only in the former is the initial foot closer to the beginning of the
compound and therefore the better candidate in competition with the (σ(σσ) forms.
3 “>” means ‘X precedes Y’ or more specifically, ‘X is further away from the head than
Y’ in Sproat and Shih’s (1991) study. The pronominal adjectival modifiers examined by
Sproat and Shih (1991) actually include a successive hierarchical schema: QUALITY >
SIZE > SHAPE > COLOR > PROVENANCE. Here, since only the order of [SIZE >
COLOR] is directly relevant to the present analysis, other relations are omitted.
4 The reduplication, as seen above, is a reanalyzed abbreviation from the phrase [0TAKE a
V/N].
5 The following examples show that there is no tone sandhi effect on the yı- in verb
reduplication:
Verb reduplication:
吃一吃 chı- yi chı
玩一玩 wán yi wán
134 Minimal and maximal word effects
想一想 xiăng yi xiăng
看一看 kàn yi kàn
Tone sandhi : 一吃 yì chı
一玩 yì wán
一想 yì xiăng
一看 yí kàn
This indicates that the original level tone of yı- is neutralized as a weak form. As a result,
the toneless weak form would be counted as containing one mora [μ].
References
Baker, Mark. 1988. Incorporation: A Theory of Grammatical Function Changing. Chicago: The
University of Chicago Press.
Chen, Matthew Y. 2000. Tone Sandhi: Patterns Across Chinese Dialects. Cambridge, UK: Cambridge University Press.
Dai, Xiang-ling. 1992. Chinese Morphology and Its Interface With the Syntax. Doctoral dissertation. Athens, OH: Ohio State University.
Duanmu, San. 1998. “Wordhood in Chinese”. In Jerome Packard (ed.) New Approaches to
Chinese Word Formation – Morphology, Phonology and the Lexicon in Modern and Ancient Chinese. Berlin: Mouton de Gruyter, 135–196.
Duanmu, San. 2000. The Phonology of Standard Chinese. Oxford: Oxford University Press.
Duanmu, San. 2007. The Phonology of Standard Chinese. Oxford : Oxford University Press.
Feng, Shengli. 2000. Prosodic Syntax in Chinese 漢語韻律句法學. Shanghai: Shanghai Education Press.
Feng, Shengli. 2001a. “The Origin of Disyllabicity in Chinese 漢語雙音步的歷史來源”.
Contemporary Research in Modern Chinese 現代中國語研究 1: 123–138. Also included in:
Youwei Shi (eds.) 2001. From Semantic Information to Typological Study. Beijing: Beijing
Language and Cultural University Press, 22–47.
Feng, Shengli. 2001b. “The Multidimensional Properties of Word in Chinese
從韻律看漢語詞語分流之大界”. Contemporary Linguistics 中國語文 3: 161–174.
Feng, Shengli. 2011. “A Prosodic Explanation for Chinese Poetic Evolution”. Tsing Hua
Journal of Chinese Studies, 2: 223–257. June 2011.
Hale, Ken and Keyser, Samuel Jay. 1993. “On Argument Structure and the Lexical Expression
of Syntactic Relations”. In Kenneth Hele and Samuel Jay Keyser (eds.) The View From
Building 20: Essays in Linguistics in Honor of Sylvain Bromberger. Cambridge: MIT Press,
53–109.
Huang, C.-T. James. 1982. Logical Relations in Chinese and the Theory of Grammar. Ph.D. dissertation. Department of Linguistics and Philosophy, MIT, Cambridge, MA.
Huang, Cheng-Teh James, Yen-hui Audrey Li, and Yafei Li. 2009. The Syntax of Chinese.
Cambridge: Cambridge University Press.
McCarthy, John and Prince, Alan. 1990. “Prosodic Morphology and Templatic Morphology”. In Mushira Eid and John McCarthy (eds.) Perspective on Arabic Linguistics: Papers From
the Second Symposium, Amsterdam: J. Benjamins, 1–54.
McCarthy, John and Prince, Alan. 1993. “Generalized Alignment”. In Geert Booij and Jaap
van Marle (eds.) Yearbook of Morphology 1993. Dordrecht: Kluwer, 79–153.
McCarthy, John and Prince, Alan. 1998. “Prosodic Morphology”. In Andrew Spencer and
Arnold M. Zwicky (eds.) The Handbook of Morphology. Oxford: Blackwell, 283–305.
Paul, Waltraud. 2005. “Adjectival modification in Mandarin Chinese and related issues”. Linguistics 43.4: 757–793.
Minimal and maximal word effects
135
Shi, Dingxu. 2002. “The Syntactic Category of Compounds and Phrases
複合詞與短語的句法地位”. In Zhang, Bojiang (ed.) Studies and Explorations in Grammar
語法研究和探索(Volume 11): 35–51.
Shih, Chilin. 1986. The Prosodic Domain of Tone Sandhi in Chinese. Ph.D. dissertation. University of California, San Diego.
Sproat, Richard and Shih, Chilin. 1991. “The Cross-Linguistic Distribution of Adjective
Ordering Restrictions”. In Georgopoulos Carol and Roberta Ishihara (eds.) Interdisciplinary Approaches to Language. Dordrecht: Kluwer Academic, 565–953.
Zhu, Dexi. 1980. Studies on Modern Chinese Grammar 現代漢語語法研究. Beijing: Commercial Press.
Zhuang, Huibin. 2015. Syntactic Words in Chinese 漢語的句法詞. Beijing: Beijing Language
and Culture University Press.
6Prosodic register grammar
Based on the theory of prosodic grammar developed in previous chapters, this
chapter will demonstrate how the prosodic system interacts with register grammar (RG for short) giving rise to what is now called prosodic register grammar
in Chinese (PRG for short). PRG concerns with three major features in formal
written Chinese which in turn consist of (1) monosyllabic words used in disyllabic templates (i.e., Qiàn-ŏu Cí 嵌偶詞), (2) disyllabic words used in disyllabic
couplets (i.e., Hé-ŏu Cí 合偶詞) and (3) formal structures formed in syntax
(Zhèngshì Tî Jù-xíng 正式體句型). Since the formal structures are mainly concerned with register syntax (not prosody), this chapter will only concentrate
on the prosodically related register morphological features of the Qiàn-ŏu Cí
嵌偶詞 and Hé-ŏu Cí 合偶詞.
The monosyllabic words in Chinese have a dual function in the register
system: the ordinary monosyllabic words such as shǒu 手 ‘hand’, jiǎo 腳 ‘foot’,
zhu- 豬 ‘pig’, gǒu 狗 ‘dog’, tia-n 天 ‘sky’ and dì 地 ‘earth’ are informally registered
while the classical ones such as xiào 校 ‘school’ and biàn 遍 ‘all over’ are mostly
used for elevated or formal expressions.1
There is an additional distinction between monosyllabic and disyllabic words
which corresponds to casual and formal registers as first recognized by early
Republican philologist Huang Kan 黃侃 (1982), who says:
1. In Chinese, the sound (refers to morpheme, by Feng) is monosyllabic while the tone
(refers to prosodic unit, by Feng) is disyllabic, and thus monosyllabic words often
become disyllabic ones, giving rise to the fact of one meaning with two words: one
is a monosyllabic while the other is disyllabic, such as tia-n 天 ‘sky’ and huángtia-n
皇天 ‘sky’ or hàotia-n 昊天 ‘sky’.The disyllabic words are redundant in plain speech
(zhìyán 質言), but useful in literary language. Because of the properties given above,
Chinese poetry and parallel prose are created accordingly. Other types of versification
of the language are also formed by these two features.2
In addition to the difference between ordinary speech (zhìyán 質言) and
written language (wényán 文言), there is also a difference between poetic and
prosaic expressions as well. In this chapter we will concentrate on the first difference and show that, although the formal written features must be used under
Prosodic register grammar
137
strict principles, namely, the Register Coherence Principle, the Principle of
Auditory Intelligibility (PAI) and the Principle of Shaping by Prosody (PSP),
they must also be combined or mixed with some colloquial features in order
to make the language natural and keep the register system more balanced and
operative. As a result, an amalgamation principle which modulates literary diction with colloquial expression is proposed here. Furthermore, based on these
principles, a method of calculation has been developed for the quantitative
analysis of the degree of formality of a specimen of Chinese writing.
6.1. What is register grammar?
In order to see how the register morphological categories Qiàn-ŏu Cí 嵌偶詞
and Hé-ŏu Cí 合偶詞 are formed and functioning in Mandarin Chinese, it is
necessary to see first what register grammar is in a linguistic sense.
The term Yŭtı̆ grammar was coined first in Chinese by Feng (2010b, 2011).
Later it was translated into English by Feng and Vihan (2016) as register grammar. What does Yŭtı̆ exactly mean then? Feng (2010b) defines it as follows:
2. Definition of Yŭtı̆
Yŭtı̆, literally meaning speech system, is a primitive system inherently acquired
by communicators for adjusting distances or relationships among themselves when they are in face-to-face communications.3
How does Yŭtı̆ have anything to do with grammar? There are rich literatures
which have investigated the grammatical properties of Yŭtı̆s both theoretically
and empirically in Chinese linguistics. A full-fledged overview of the Yŭtı̆ theory with its empirical results is impossible here, but a brief summary can be
given in terms of the following two crucial principles:
Principle I: Form-Function Correspondence Principle (FFCP):
Every linguistic form has a function inherently interpreted by communicators (interlocutors) to adjust communicative distance in face-to-face
communications.
Principle II: The Realization of Yŭtı̆ through Grammar (RYG)
In human languages, there are essentially three types of Yŭtı̆s, i.e., formal,
informal and elevated, and each of them is realized (or identified) by a
specific form of the grammar.
Some important notions stated in the above definitions need to be clarified
first before we proceed. First, the term ‘grammar’ used here refers to Yŭtı̆ grammar (YG), paraphrased into English as register grammars. Second, it is assumed
that there is not only one system of grammar in a specific language synchronically (i.e., Chinese, for example); rather, two or more grammars are involved in
a natural language (see also Kroch 2000 and references cited there). Third, the
138 Prosodic register grammar
notions of formal, informal and elevated in the system of RG are constructed
relatively; that is, formal, informal and elevated Yŭtı̆ types exist in a relative
relation: there is no informal form if there is no formal entity to compare
with. Similarly, an elevated form could not exist by itself alone without being
compared with formal and informal ones. To put it differently, formal styles are
obtained in comparison with informal ones, and elevated style is formed by a
contrast with formal and informal ones. In such a system it is crucial that the
FFCP, together with the RYG will give rise to the fact that any piece of utterance should have a function definable by RG (register grammar). As a result,
each register (formal, informal or elevated) will correspond to, or is identified
by, a specific form of the grammars, which implies that there are different categories of RG-forms in human languages. These categories not only contrast
with each other, but also cooperate with one another to fulfill the purpose of
defining communicative distances in face-to-face situations.
The following example demonstrates how the Yŭtı̆ system works and what
exactly register grammar is about.
3. Informal *[VP PP] and Formal [VP PP] in Modern Chinese
a. *畫家要飯在紐約大街上。
*huà-jia- yào fàn zài niŭ yue- dà jie- shàng.
painter beg food at New York big street above
‘A painter begs for food at the street of New York.’
b. 畫家乞食于紐約街頭。
huà-jia- qı̆ shí yú niŭ-yue- jie- tóu.
painter beg food at New York street top
‘A painter begs for food at the street of New York.’
4. Informal *[V & V] and Formal [VV & VV] in Modern Chinese
a. *買和看了一本書
*măi hé kàn le yı- běn shubuy and read Aspect one Classifier book
‘buy and read a book’
b. 購買和閱讀了一部古典小說
gòu măi hé yuè dú le yı- bù gǔdiǎn xiáoshuobuy-buy and read-read Asp one Cl classical novel
‘buy and read a classical novel’
As seen in (3), the structure of [VP PP] is both grammatical and ungrammatical
in different register situations: in everyday speech, it is ungrammatical while in
formal speech (with a formal preposition yú ‘at’), it is grammatical. Likewise,
the coordinating structure of [V and V] in Modern Chinese is not acceptable if
it is formed by informal (monosyllabic) verbs but becomes perfectly grammatical if it is structured by formal (disyllabic) verbs. The facts show that registers
Prosodic register grammar
139
are not only realized through syntax but also identified by syntax: registers are
encoded within grammars. That is to say, the grammaticality of certain structures is systematically determined by different types of registers. As a result, the
structures of [V PP] and [V & V] are ungrammatical in an informal register
(i.e., the syntactic operations in the language do not allow such a structure
to be used in everyday speech) but perfectly acceptable in the formal register
just like the RG works in other languages (see example 5 below). Note that
there would not have been a proper explanation for the ungrammaticality of
the phenomena in (4) above and (5) below if the notion of register grammar
had not been developed in the past several years (readers can refer to the article
“Written Language versus Spoken Language” in Brill Encyclopedia of Chinese
Linguistics and “Modern Chinese: Written Chinese” in Routledge Encyclopedia
of Chinese Language and Linguistics). Besides, the effects of register grammar are
not only observed in Chinese, but are also manifested in other languages like
English. For example,
5. Dative construction in English
a. *Donate me a picture.
b. Give me a picture.
c. *Construct us a house.
d. Build us a house.
a’. Donate a picture to me.
b’. Give a picture to me.
c’. Construct a house for us.
d’. Build a house for us.
Although the register grammar is newly developed in Chinese, it has also
been found effective in other languages as well. The dative construction in
English is a typical example of this kind. It has long been a mystery as to why
some verbs cannot be used in the dative construction. The puzzle can now
be properly explained within the theory of RG as to why only colloquial (or
Anglo-Saxon) verbs such as ‘give’ and ‘build’ have the dative structure while
the Latinate verbs like ‘donate’ and ‘construct’ do not. It is so because different
registers use different grammars and Anglo-Saxon verbs are colloquial while
the Latinate ones are formal. Then, when the Latinate ones are used more frequently in everyday speech (become more colloquial and opaque to its Latinate
origin), the dative construction starts to be acceptable ((such as ‘assign’, see Yang
2016:191-198) because it is then reanalyzed (re-categorized) into the informal
grammar (i.e., operable with dative construction).
Within the RG introduced above, we can be sure about the early attempts of
stylistic variation in the studies of sociolinguistics. For example, Enkvist (1973,
see also Crystal, David and Davy Derek 1969) has proposed a stylogrammar
and claimed that “some types of stylogrammatical rules are categorical within
each style”. However, there are no clear and precise categorical rules being
formulated under the stylogrammar that would generate all grammatical sentences and rule out all ungrammatical ones. A typical example given by Enkvist
(1973:46) under his approach comes from the religious text that “always uses
‘thou lowest’，and never ‘you love’ ”. Of course, the phrase of ‘thou lowest’
can be considered as a form with elevated style for ‘you love’, but it is best
140 Prosodic register grammar
analyzed as a stylo-word, not the stylogrammar (i.e., a distance-word 語距詞 –
words used for identifying/adjusting register distance, but not a distance grammar 語距語法 – structures used to identify/adjust register distance).Thus these
examples can hardly be considered as a grammatical device of the register (Yuti)
system. This may be the reason why the term stylogrammar, together with the
theory on stylogrammatical rule disappeared in the 1980s.
Recently, Kiesling (2009) has proposed another similar theory in terms of
stance, which is directly related to the register grammar in Chinese, as stated as
follows:
6. “I define stance as a person’s expression of their relationship to their talk
(their epistemic stance – e.g., how certain they are about their assertions),
and a person’s expression of their relationship to their interlocutors (their
interpersonal stance – e.g., friendly or dominating).” (Kiesling 2009:172)
The notion of ‘stance’ as defined above has something in common with
(and is also different from) the theory of Yŭtı̆ given in (1). In other words, the
notion of stance and the theory of distance in Yŭtı̆ are different even if they
are somehow similar. Compare the definition of Yŭtı̆ again (repeated here for
convenience):
Yŭtı̆ is a primitive system of face-to-face communication inherently
acquired by communicators for adjusting distances among themselves.
The similarities are obvious: both systems characterize the use of language
to control/manage the relationships among interlocutors. However, the relationships defined in both theories are quite different. By the Stance Theory,
the relationships among interlocutors consist of two parts, one is an epistemic
stance – i.e., how certain they are about their assertions, and the other is an
interpersonal stance – i.e., friendly or dominating. For the Yŭtı̆ Theory, the
relationships among communicators consist of a substantial distance in reality
which can be realized as various types of conceptual distances (near-intimate,
far-distant/formal; high = respectful/dominating, low = modest/honorific,
etc.) in different natural environments and social circumstances among humans
(presumably among animals as well), illustrated as in Figure 6.1.
Figure 6.1 reflects the hypothetical system for tripartite division of the threeway stylistic registers (or triglossia). As seen below, several notions within this
system need to be emphasized: colloquial and standard form a pair of “contrasting concepts”, meaning that, to employ the label “formal expression” in a
theory is to realize a certain “degree of formality”, which is weighed against
an opposed notion of a “degree of informality”, or, “degree of colloquialism”.
There are no absolute formal or informal styles in this system. Thus to determine “formal” and “informal” is always a matter of “degree”. And for the same
reason, “elevated” and “non-elevated” also constitute a contrastive pair in which
the members illuminate one another and thus facilitate the differential classification of “highly elevated” and “very elevated”.
Prosodic register grammar
141
Elevated (ancient expressions)
casual
(slang
colloquial
vernacular
standard
literary)
(Modern Chinese)
Figure 6.1 The register composition in the binary stylistic system
Given the brief comparison between the theory of Stance and the theory
of Yŭtı̆ (or register grammar) above, the differences are quite apparent: Yŭtı̆
concerns with communication among all kinds of communicators, animals and
humans (although the animals’ communicative capacity and their Yŭtı̆ system
have not been fully studied and understood yet) while the theory of Stance
only concerns with communications of humans in certain situations. There is
another difference between these two theories, i.e., the epistemic stance. In the
theory of Stance, epistemic stance is part of the theory itself, while in the theory
of Yŭtı̆, it is derived from distance among the interlocutors. In other words,
distance is a more primitive notion and more like an atomic element in the
communication systems while Stance is merely a manifestation of the system of
distance – Stance is not only included but in fact derivable from system of Yŭtı̆.
In this chapter, we will focus only on human language communications.
The studies about the function of human languages have proposed the notions
of style, register, genre and even stylogrammar in past decades. However, what
I would like to point out is the fact that there seems to be no theory-orientated
investigation on the grammaticality of registers found in previous studies
(including the stylogrammar developed by Enkvist 1973 and the Stance theory
by Kiesling 2009, which are most similar to Yŭtı̆ theory). On the contrary, as
current studies show (Feng 2003a&b, 2010b, Wang 2009, 2012 and references
cited there), the grammatical functions of the three dimensions of stylistic registers (i.e., formal, informal and elevated) are manifested through or identified/
characterized by grammatical devices in human languages. To put it differently,
the human register capacity (see Labov 2014 for innate capacity of human
register competence – using language to manage relationships with others),
though encoded into their cognitive system of various kinds (i.e., body language, the musical capacity of all life forms, etc.) is indeed encoded into the
formal system of the grammar of the language they acquired, used and developed. In short, register grammar is a system of diglossia, not a stylistic choice
as traditionally assumed. In what follows we will first introduce the history
of newly established formal register grammar in Modern Chinese after the
May Fourth Movement and then discuss the register functions of Qiàn-ŏu Cí
嵌偶詞 and Hé-ŏu Cí 合偶詞.
142 Prosodic register grammar
6.2 The Chinese register grammar
6.2.1. History of register grammar in modern China
Register grammar is recognized first in the study of written Chinese. The formal Chinese writing system was newly developed after the May Fourth Movement in 1919. Before then, Chinese intellectuals (including virtually everyone
who was literate) wrote in Classical (literary) Chinese.4 The Literary Revolution, which was initiated several years before or after the May Fourth Movement, promotes writing in the vernacular with the slogan “write what you say.”
However, 100 years later, it has ended up with a new separation of writing from
speaking again today. This was not expected by the initiators of the Literary
Revolution, but can be accurately predicted by the theory of register grammar
(see also Huang Kan 1936). Of course, today’s separation of writing from speaking is not simply a reversion to the past. What is important to note here is the
fact that all languages necessitate two relative functions: formal and informal. If
“write what you say” results in a purely informal style and if Classical Chinese
served as the formal style of the past, then something had to be developed or
created to fill up the gap when Classical Chinese was abolished by the Literary Revolution because there was a need for a formal style in the RG system
within social communications.
In this chapter, we demonstrate how Modern Formal Chinese is newly
developed and a prosodic register grammar (PRG for short) has emerged. As
seen above, modern written Chinese (白話文) is a result of the May Fourth
Movement (1919) and before then, even if there were proposals for writing
in the vernacular such as Huang Zunxian (黃遵憲), who has maintained that
“my hand writes [what] my mouth [says] (我手寫我口)” in the late Qing
dynasty (1840–1911), the shift from writing in literary Chinese to writing in
the vernacular did not actually occur until the Literary Revolution (文學革命)
launched by Hu Shi (胡適) and Chen Duxiu (陳獨秀) in 1919. Ironically, the
new formal writing system in Modern Chinese still uses classical forms. Why
must the creation of a formal style require the use of Classical Chinese? It has
long been believed that the replacement of literary Chinese with the vernacular
in writing was because, according to Hu Shi, literary Chinese became a dead
language thousands of years ago. Obviously this is not totally true.What is striking is the fact that there are still ‘remnants’ of literary Chinese (strictly speaking,
there are reanalyzed formal expressions from literary Chinese) within Modern
Chinese vernacular writing. For example:
7. “Of course, this does not mean that the local people have been entirely integrated
into the Han nationality in all of the areas where the Han have penetrated. In fact,
this is not so because even now there are many minorities which have mixed with
the Han nationality within Han regions.”5
– Fei Xiaotong, The Modernization and Minority
Development in China (p. 291)
Prosodic register grammar
143
In the above paragraph there are about 44 morphemes (free, as well as bound)
and 11 of them are taken from Classical Chinese, e.g., ‘所擴及 extend to’,
‘吸收為 to take as’, ‘並不如此 is certainly not like this’, ‘即在目前 even now’,
‘聚居 to live together’ and ‘在內 within’. Actually, classical expressions like these
are not merely remnants, but are required to make the written text sound formal. Zhang (2002) has proposed that people “should incorporate some literary
expressions into their own writing 要把來自文言的東西融會到自己的筆下”
(Zhang Zhongxing 2002:134).
What is the reason for incorporating literary forms into the vernacular and
why are classical elements necessary in Modern Chinese writing? The traditional
philologist Huang Kan (黃侃) (1983: 122) answered this question precisely long
ago: “The distinction between written language and spoken language . . . cannot be merely coincidental” (“文與言判 . . . 非苟而已”《黃侃日記》). The
theory behind the “Non-Coincidental-Distinction Approach” calls for the register grammar. That is, modern written Chinese makes use of Classical Chinese to
distance itself from the vernacular. Following Huang’s insight, Feng (2005) argued
that the incorporation of literary forms into vernacular is essentially a result of
formal style requirements. As seen above, a formal style is obtained by creating a
sense of distance from everyday speech.That is to say, when linguistic expressions
are used, the more distant the expressions are kept from everyday speech,6 the
more formal sense they can create for their listeners. The demand for vernacular
Chinese to function formally after the May Fourth Movement, as Feng argued,
has motivated speakers and writers to use some classical forms to satisfy their
urgent need for stylistic-register effects. This is the reason why the separation of
writing (formal) from speaking (informal) has been resurrected in modern times.
6.2.2. The reconstruction of the register system
As seen above, it has been a traditional dilemma of separating colloquial expressions from literary diction in forming modern vernacular writing over the past
100 years in Chinese philology. It did not become clear until recently that, after
the destruction of literary Chinese 文言文, modern written Chinese has been
gradually formed by combining the vernacular vocabulary and grammar of literary expression and that the inseparability nature of the two registers has made
it possible for the birth of a formal style of writing.The following diagram gives
a brief history of how the formal and informal diglossia are distributed.
In the above diagram, what is important is the notion that written Chinese
should be defined in terms of the formal style of writing in Modern Chinese.
The notions of “informal/formal” and “spoken/written” are not completely
corresponding, i.e., “formal” does not only imply written, nor does “spoken”
always refer to ‘nformal, and vice versa. Formal Chinese is also an utterable language and is not reserved only for writing. The definition implies that formal
Chinese can be both spoken and written.
Classical Chinese and Modern Chinese should be clearly distinguished here:
Classical Chinese refers to the language with many linguistic features of the
144 Prosodic register grammar
Classical
Chinese
Ancient
Modern
Elevatad
Written
Chinese
Newly Developed
Formal Systems
Formal
Chinese
Writings in Vernacular
Spoken
Chinese
Everyday Speech
Informal
Chinese
Slang
Figure 6.2 Diagram of formal and informal Chinese
Han and pre-Han periods (i.e., up to the third century AD) which remained
prevalent up until the May Fourth Movement, while Modern Chinese is
defined in terms of its auditory comprehensibility to the ordinary people of
today. Thus, speech that cannot be understood by means of its sound alone by
an ordinary high school graduate will not be considered as modern (for a more
detailed discussion of this criterion, see Feng 2003a/b).
The diagram also shows that formal expressions in Modern Chinese developed from two major sources: Classical Chinese and completely new expressions that evolved within the formal system itself after 1911.7 Note that even
though some formal expressions have been taken from Classical Chinese, they
are no longer considered as purely Classical Chinese because they have been
accepted (or reanalyzed) into and processed by the modern system of formal
Chinese and understood as such by native speakers, as we will see below. With
all these new concepts in mind, we proceed with our discussion about the reformation of the formal system in Modern Chinese.
6.2.3 The reformation of a formal system
Based on the above observation and the previous studies, we now view Modern Formal Chinese as a new language, a new form within the diglossia system
Prosodic register grammar
145
developed after the May Fourth Movement. It is new in the sense that formal
Chinese serves as a diglossic (or register) grammar within Modern Chinese.
What is this new register grammar composed of? Feng (2005) suggests that
Modern Formal Chinese consists of several components, as illustrated in the
following diagram:
According to the diagram in Figure 6.3, written Chinese (書面語) can be
characterized as a formal language composed of a syntactic system based on
Modern Chinese (i.e., Mandarin Chinese), a set of prosodic constraints on its
morphosyntactic operations and a lexicon consisting of words or expressions
from ordinary speech, dialects, foreign languages and Classical Chinese. Of
course, this does not imply that the words or expressions from these sources
can all be assimilated directly into modern writing without any processing
and manipulation. On the contrary, all of the elements taken from spoken,
dialect, foreign and classical sources may need to be further processed and
manipulated according to the prosodic stylistic principles as represented in
Figure 6.3, namely, the Selection Principle, the Processing Principle and the
Amalgamation Principle, which are responsible for organizing different elements in Modern Formal Chinese. All these will be elaborated in next section
below.
Modern Formal Chinese
Amalgamation
Ė
Processing
Ė
Selection
Syntax
Classical
Words
Oral
Words
Prosody
Dialect
Words
……
Figure 6.3 The formation of Modern Formal Chinese
Foreign
Words
146 Prosodic register grammar
6.2.4 The principles of integration
6.2.4.1 The Selection Principle
The Selection Principle is very simple: a speaker or writer has the freedom
to select whatever element they want from the lexicon (or a dictionary) for
the purpose of expressing different styles of the language, ranging from colloquial to formal. If styles of expression can be characterized in terms of their
distance from everyday speech (or degrees of familiarity), then different sources
of expressions can be classified according to their different degrees of formality
in written Chinese. This is illustrated as follows（‘>’ stands for “is more formal
than”:
8. The Distance Gradation Principle (DGP)
classical > foreign > dialect > spoken expressions
The more familiar an expression, the closer it is to everyday speech and vice
versa. Given the DGP in (8), we are able to rank expressions into degrees of formality according to their stylistic properties, as well as their source. For example:
9. a. fǎngfú 仿佛 > sìhu- 似乎 > hǎoxiàng 好像 ‘seems, like’
b. ruò 若 > rúguŏ 如果 > yàoshì 要是 ‘if ’
These three words in (9a,b) share the same meaning. The only difference
between them is that fǎngfú and ruò are more formal than sìhu- and rúguŏ,8 which
are more formal than hǎoxiàng and yàoshì.9 Although there may not always be
triplets of degrees of formality like the example of (9), each individual expression can theoretically be ranked with a proper degree of formality according to
the DGP. In fact, in the book Expressions of Written Chinese, we (Feng 2006) have
listed 400 Hé-ŏu Cí 合偶詞, 350 Qiàn-ŏu Cí 嵌偶詞, 30 formal patterns and
90 elevated patterns and all of them can be used to achieve a formal writing,
and all these formal and elevated expressions in Modern Chinese are virtually
ranked with one of the three degrees of formality based on the principles and
methods illustrated above. As a result, one can select an expression (Qiàn-ŏu
Cí 嵌偶詞, Hé-ŏu Cí 合偶詞 or a phrase gŭjùxíng 古句型) from this reference
book in order to facilitate formal writing and speaking.
6.2.4.2 The Processing Principle
Word or phrase selection from sources or reference books is a preliminary step
towards expressing a formal style in Chinese (and presumably in other languages too). However, this does not mean that one can freely select whatever
they want. Also, it is not the case that every formal element, whether a word,
a phrase or a structure, can be used directly without some necessary grammatical and registeral modification. Thus, principles for selecting/using formal
Prosodic register grammar
147
expressions are needed and two such important processing principles are proposed in Feng 2005: the Principle of Auditory Intelligibility (PAI)10 and the
Principle of Shaping by Prosody (PSP).The PAI requires any formal expression
to be auditorily intelligible, while the PSP requires that auditorily intelligible
formal expressions must be shaped by a proper prosodic structure, since prosody
is a system that is also inherently correlated with the different types of register.
For example:
10. zhı-dào 知道 vs. zhı- 知 ‘to know’
a. 他知道今天沒有課。
ta- zhı-dào jı-ntia-n méi yǒu kè.
he know today not have class
‘He knows that there is no class today.’
b. *他知今天沒有課。
*ta- zhı- jı-ntia-n méi yǒu kè.
he know today not have class
‘He knows that there is no class today.’
c. 他不知今天沒有課。
ta- bù zhı- jı-ntia-n méi yǒu kè.
he not know today not have class
‘He does not know that there is no class today.’
11. jiěshì 解釋 vs. jiě 解 and shì 釋; and xiǎn 鮮
a. 這道題無解。
zhè-dào tí wú jiě.
this-CL problem no solution
‘This (mathematics) problem has no solution.’
b. 這道題無釋。
zhè-dào tí wú shì.
This-CL problem no solution
‘This (mathematics) problem has no solution.’
c. 多還可以，鮮則不行。
duo- hái kěyî, xiǎn zé bù xíng.
more still okay, less however not okay
‘More is fine, less wouldn’t do.’
In (10), zhı-, the monosyllabic counterpart of the common word zhı-dào is auditorily intelligible to native speakers of high school level or above, but (10b)
is not grammatical. In (11a) jiě, the monosyllabic counterpart of jiěshì, is also
intelligible. However, (11b) and (11c) demonstrate a different scenario: neither shì nor xiǎn are auditorily intelligible by a native speaker at or above
high school level, although they are free-standing, common words in Classical
148 Prosodic register grammar
Chinese. According to the Auditory Intelligibility Principle, both of them must
be excluded from Modern Chinese. In other words, a boundary between Classical and Modern Chinese can conveniently be drawn according to the Auditory Intelligibility Principle: anything that is not auditorily intelligible is not
Modern Chinese.
6.2.4.3 The Amalgamation Principle I: intelligibility
After the selection and processing are completed, we are then faced with the
next question: how can we transform formal expressions into spoken expressions? This is what has been called the ‘amalgamating problem’.
12. “When adopting a few classical words into modern writing, what can we do so that
our adoption is considered as amalgamation rather than adulteration? The situation
is extremely complex and difficult to pin down. I would have to say that we may
use the ear as a measure. Anything that sounds odd to the ear or obviously not like
everyday speech (including discussion of academic issues) will be considered adulteration, not amalgamation, and vice versa.”11 – Zhang Zhongxing, Gateway to
Literary Chinese (2002)
As noted above, it is not uncommon that one “should incorporate some literary expressions into their own writing” (Zhang, 2002). However, syncretization
does not necessarily imply an adulteration. Thus, we must clearly distinguish
amalgamation (融會) from adulteration (攙雜). The new style of formal Chinese has been formed by a process of amalgamating classical expressions into
the modern vernacular. It is not adulteration, even though there may be isolated
examples of it.Thus, the principles of intelligibility and effability must be combined together in order to derive the resulting amalgam.
13. 吝嗇的人，我們說他小氣；妒忌的人，我們也說他小氣。小氣，
自然不夠偉大；即使不是十足的小人，至少該說是具體而微的小人。
但是，如果小氣的人就算是小人之一種，則小人滿天下，而足稱為
君子者，實在太少了。（取自王力《小氣》，《龍蟲並雕齋瑣語》
p.145)
“For those stingy men, we say ‘they are petty’ while for those jealous men,
we also say ‘they are petty.’ Pettiness is of course not very great, even if
they are not entirely petty men. We should at least say they are trivial men.
However, if being a person who is petty is considered as one type of petty
man, then petty men are everywhere, and those who can be considered as
gentleman, they are truly few.”
– Wang Li, On Pettiness
The shaded portions are literary expressions from Classical Chinese and here
they are amalgamated with spoken expressions. This passage can be read aloud
and is completely intelligible and can be easily understood. Even though there
Prosodic register grammar
149
are many classical expressions in it, the way it is composed, that is, artfully combining Classical and spoken Chinese, make it a masterpiece of Modern Formal
Chinese.
6.2.4.4 The Amalgamation Principle II: proportionality
It is clear by now that modern formal writing must amalgamate literary dictions
with colloquial expressions. However, what proportion should be the correct
amount of literary expressions used in formal writings? In other words, to what
extent and how can one modulate them? Years of research (Feng 2003 a/b/c,
2005, 2006, 2008) have suggested a tentative conclusion, indicating a ratio of
2:3 between formal and informal expressions in written Chinese. Examining
the above example:
14. 吝嗇的人，我們說他小氣；妒忌的人，我們也說他小氣。小氣，
5
1 2 3 4
自然不夠偉大；即使不是十足的小人，至少該說是具體而微的小人。但
678
是如果小氣的人就算是小人之一種，則小人滿天下，而足稱為君子者，
910 11
實在太少了。（取自王力《小氣》，《龍蟲並雕齋瑣語》p.145）
12
There are a total of 12 phrases in this passage, among which five belong to
literary dictions, giving rise to a ratio of 5:7. In other words, we have about
42% formal expressions and 58% informal ones here. The formality ratio of
2:3 seems to be something like a golden mean that characterizes the ratio of
literary and colloquial dictions. Based on measurements of the stylistic ratio,
we are able to characterize the process of formal writing in terms of a wave
model:12
15. The Stylistic Wave (of formal Chinese)
Formal
1
2
8
6
3
9
4 5
7
11
10
12
Informal
This Stylistic Wave Model can also be viewed as the alternation between
literary dictions and colloquial expressions with regard to a ratio of 2:3 in
formal Chinese. As a result of the theory and the methodology presented here,
we are able to offer a solution to the problem of the incalculability of styles in
150 Prosodic register grammar
human languages. Of course, there are new questions which have arisen and
new areas of research which have been opened up and all of them are left for
future studies.
All of the principles outlined above, namely, selection, processing (PAI and
PSP) and amalgamation (intelligibility and proportionality), are necessary and
prosodic morphology plays an important role as register-forming capacity,
which shows that a prosodic register grammar is called for.
6.3. Prosodic register grammar in Chinese
6.3.1. Prosodically constructed elevated expressions (Qiàn-ŏu Cí)
If classical forms must be used to write in a formal elevated style in Modern
Chinese, then we must ask: in what ways can these classical forms be used?
Though there may be a variety of ways to make use of them, it should be
pointed out that they cannot be used directly in modern formal writing without a prosodic justification, as shown by Feng (2003 a/b) in his study of modern formal writing. For example:
16. Elevated style
我國政府出於同樣考慮,認為不宜前往。
wŏ guó zhèngfŭ chu- yú tóngyàng kăo lǜ, rènwéi bù yí qián wăng.
our country government have same consideration, decide no should
forward go
‘Our government will take the same consideration and decide not to go.’
Casual style
咱想的跟您一樣,就甭去了。
zán xiăng de ge-n nín yı- yàng, jiùbéng qù le.
we think with you same,
then no-need go Asp
‘I agree, let’s not go’.
Within the theory of prosodic morphology in previous chapters and the register grammar discussed in this chapter, we now understand that the bizarre templative behavior of words (i.e., words used in disyllabic template) is in essence
a natural consequence of the interface processes between different components
of the grammar.
It has also become increasingly clear, based on the study of prosodic grammar
over the past 20 years, that something being intelligible does not necessarily
mean that it is utterable. The example given in (10b) shows that the monosyllabic verb zhı- ‘to know’ cannot stand alone and must be used within what
is characterized as a ‘disyllabic template’ in order to be effable, thus forming
what is called Qiàn-ŏu Cí 嵌偶詞. There is a wealth of evidence that words
and phrasal patterns are ineffable without a proper prosodic shape (Feng 2003
Prosodic register grammar
151
a/b/c, 2005, Feng, Wang and Huang 2008), and the Qiàn-ŏu Cí are born with a
classical taste which gives rise to an elevated style.
While the grammars of human language are systemized between written (formal) and spoken (informal) registers, the stylistic differences between formal/
informal and elevated/plain are also grammaticalized systematically. For example:13
17. Spoken
a. 一樣
yı-yàng
‘the same’
b. 不一樣
bù yı-yàng
‘not (the) same’
c. A跟B一樣
A ge-n B yı-yàng
‘A and B are the same.’
d. 一樣的東西
yı-yàng-de do-ngxi
same POSS thing
‘the same thing’
Written
a’. 同
tóng
‘the same’
b’. 不同
bù tóng
‘not (the) same’
c’. A和B*同
A hé B *tóng
‘A and B are the same.’
d’. *同的東西
*tóng-de do-ngxi
same POSS thing
‘the same thing’
Yíyàng and tóng are semantically identical, meaning “the same”. However, the
grammars within which they are used are different, since the spoken form yíyàng
is a free form while the formal expression tóng can only occur in a disyllabic template, thus being a Qiàn-ŏu Cí. More examples of this kind are given in (18) below:
18. a. *四環以外准鳴笛
*sì huán yı̆wài zhŭn míngdí
Fourth Ring outside permit honking
‘Honking is permitted outside the Fourth Ring (road).’
b. 四環以內不准鳴笛
sì huán yı̆nèi bù zhŭn míngdí
Fourth Ring inside not permit honking
‘Honking is not permitted within the Fourth Ring (road).’
c. 待上級准假後 才能離京。
dài shàngjí zhŭn jià hòu cái néng lí Jı-ng.
wait superiors approve vacation after only can leave Beijing.
‘Only after his superiors approve his vacation can he leave Beijing.’
d. Spoken 來參觀旅遊的人(不)應該去
lái ca-ngua-n lǚyóu de rén (bù) yı-ngga-i qù
Come visit [particle] people (not) should go
‘People who come to visit should (not) go [there].’
152 Prosodic register grammar
e. Written 觀光遊客不宜*往/*宜前往/不宜前往
gua-ngua-ng yóukè bù yí *wăng /*yí qiánwăng/bù yí qiánwăng
visiting tourist not should go /should proceed /not should proceed.
‘Visiting tourists should (not) proceed (further).’
The monosyllabic words tóng ‘same’, zhŭn ‘permit’ and yí ‘should’ are taken
from Classical Chinese and are thus elevated in formal or elegant speech.
What is important to note here is the fact that without being shaped
by prosody, the examples given in (17c’,d’) and (18a,e) are all ungrammatical. Thus, prosody does affect register as far as the above examples are
concerned.
6.3.2. Prosodically constructed formal expressions (Hé-ŏu Cí)
Not only are the monosyllabic classical elevated forms prosodically conditioned
in modern formal writings, as seen above, but also the newly developed formal
expressions in the modern written language (Hé-ŏu Cí 合偶詞) are strictly
constrained by prosody. For example,
19. a. 我們一定要對這個問題進行研究。
wŏmen yı- dìng yào duì zhège wèntí jìnxíng yánjiu-。
we definitely must towards this issue carry-out study
‘We must study this question (we must carry out the study on this
question).’
b. 瞧！您的閘不靈了,得修一下了。
qiáo! nín de zhá bù líng le, děi xiu- yı- xià le.
look you POSS brake not effective Asp must repair a little Asp
‘Look, your (bicycle) brake has some problems, you have to fix it.’
c. 瞧！您的閘不靈了,*得進行修了。
qiáo! nín de zhá bù líng le, *děi jìn xíng xiu- le.
Look you POSS brake not effective Asp must carry-out repair Asp
‘Look, your (bicycle) brake has some problems, you have to fix it.’
d. 閘皮失靈,必須進行修理！
zhá pí shı- líng, bì xu- jìnxíng xiu-lı̆!
brake rubber lose efficiency, must carry-out repair
‘If the brake is not efficient, one must repair it!’
As seen above, native speakers do not use the formal expression like jìnxíng
in ordinary speech, but only on formal occasions, e.g., regulating transportation, the expression can be paraphrased formally as shown here. The intriguing
phenomena here are the fact that disyllabic light verbs like jìnxíng demand a
disyllabic complement:
Prosodic register grammar
153
20. a. 閘皮失靈,必須進行修*(理)！
zhápí shı-líng, bì xu- jìn xíng xiu- *(lı̆)!
brake rubber lose efficiency, must carry-out repair
‘If the brake is not efficient, one must repair it!’
b. 閘皮失靈了,一定要修了。
zhápí shı- líng le, yı- dìng yào xiu- le.
brake rubber lose efficiency ASP, must need repair ASP
‘If the brake is not efficient, one must repair it.’
In fact, in formal Chinese many disyllabic words are coined in modern times
and used exclusively in formal speech or writings. For example,
21. Written
a. 無法 + 學*(習)
wú fă xué*(xí)
no way study(-learn)
‘no way to study’
b. 禁止 + 說*(話)
jìnzhı̆ shuo-*(huà)
prohibit speak(-word)
‘talking prohibited’
c. 毫無 + 權*(力)
háo wú quán*(lì)
little not-have rights(-power)
‘no right at all’
d. 共同 + 談*(笑)
gòngtóng tán*(xiào)
together talk(-laugh)
‘talk together’
Spoken
a. 沒辦法學
méi bànfă xué
no way study
‘no way to study’
b. 不許說
bù xŭ shuonot allow speak
‘talking prohibited’
c. 一點兒權都沒有
yìdiănr quán do-u méiyǒu
little rights even not have
‘no right at all’
d. 一塊兒談
yí-kuàir tán
together talk
‘talk together’
The examples below illustrate again that disyllabic VV tends to require a
disyllabic VV or NN complement, and a disyllabic adverb expression tends to
require a disyllabic VV or AA head.
22. VV selects VV
加以+批*(判)
jia-yı̆ + pı-*(pàn)
give + criticism
VV selects NN
堅持+*(真)理
jia-nchí + *(zhe-n)lı̆
insist on + truth
(持理)
(chí-lı̆)
(insist on truth)
予以+批*(評)
yŭyı̆ + pı-*(píng)
give + criticism
面臨+危*(險)
miànlín + we-i*(xiăn)
face + danger
(臨危)
(lín-we-i)
(face danger)
從事+教*(學)
cóngshì + jiào*(xué)
導致+*(疾)病
dăozhì + *(jí)bìng
(致病)
(zhì-bìng)
154 Prosodic register grammar
engage in + teaching
disyllabic Adv selects VV
cause + disease
disyllabic Adv selects AA
公然+*(逃)跑
go-ngrán + *(táo)păo
openly + escape
極其 + *(容)易
jíqí + *(róng)yì
extremely + easy
共同+*(協)商
gòngtóng + *(xié)sha-ng
jointly + discuss
更加+*(美)好
gèngjia-+ *(mĕi)hăo
even more + beautiful
(cause disease)
並肩+戰*(鬥)
極為+*(不)滿
bìngjia-n + zhàn*(dòu)
jíwéi +*(bù)măn
shoulder to shoulder + fight extremely + unsatisfied
There are more than 400 Hé-ŏu Cí collected in Expressions of Written
Chinese《漢語書面用語初編》which are used within disyllabic couplets, and
all of them are formal expressions like jìnxíng進行 ‘carry out’ developed in
Modern Chinese systematically.
6.3.3. Prosodic register grammar
Given the examples of Qiàn-ŏu Cí and Hé-ŏu Cí and the fact that formal expressions are prosodically constrained, we are now able to conclude that the grammar of formal Chinese is essentially a prosodic grammar, which can be defined
as follows.
23. Prosodic grammar
If the computational system of a language operates under prosodic conditions, then the grammar of the language is characterized as a prosodic
grammar.
Regarding the characteristics of the prosodic grammar of written Chinese,
several register principles have been proposed in previous studies (Feng 2003
a/b/c), as shown in the following:
24. a. *[σ] → [σσ]foot
Disyllabic Foot Structure
A monosyllable cannot form a foot and thus cannot stand alone.
b. [σσ]PrWd → [σσ]PrWd +[σσ]PrWd
Register Coherence Principle
A prosodic word (PrWd), commonly formed by a foot, selects another
prosodic word in formal Chinese.
c. Principle of Prosodic Register
The more formal an expression is, the more prosodic words are used to
compose it, and vice versa.
Prosodic register grammar
155
(24a) gives rise to the constraint that classical monosyllabic words must occur
in a disyllabic template in Modern Chinese, while (24b) results in a “disyllabic word used in a disyllabic couplet”. (24c) affects the proportion of formal
features used in modern writing (see section 6.3.3 above). Given these principles, we would expect there to be an unavoidable interweaving of expressions
between literary dictions and colloquial patterns in Modern Formal Chinese.
This turns out to be the case as we will see below.
6.4. The linguistic properties of Qiàn-ŏu Cí
6.4.1. Bound-free and semi-free
The intriguing phenomena of monosyllabic and disyllabic words occurring in
prosodic templates and prosodic couplets did not receive a theoretical consideration until the appearance of Prosodic Morphology in 1986 (McCarthy and
Prince).
Before Prosodic Morphology was introduced into Chinese, Lü (1962, 1963)
presented his earliest observations of the relevant phenomena, and he analyzed
them in terms of the ‘semi-free morpheme’ with a [2+2] syllable-pattern tendency. At that time, there is no prosodic constraint developed in Lü’s system.
Lü’s pioneering work has profoundly influenced the field of Chinese morphology. Of course, it does not mean that his analysis goes without problems. In
fact, there are obvious limitations and problems with his framework under the
theory of prosodic morphology.
First, Lü only observed a partial phenomenon regarding to the syllabic properties of the words in Chinese. In his article “On Free and Bound”, Lü listed
five circumstances where a ‘bound’ monosyllabic morpheme can be used freely
(thus he named such kind of forms ‘semi-free morphemes’):
25. 1) in different dialects, such as xié 鞋 vs. xiézi 鞋子 ‘shoe’;
2) in written Chinese, as hua- 花 vs. hua-r 花兒 ‘flower’;
3) in everyday speech, such as bí 鼻 ‘nose’ in iatrology;
4) in idioms or fixed expressions, such as bù zhı- bù jué 不知不覺 (not
know not feel) ‘unconscious’; and
5) in special structures, such as chu--le míng 出了名 (out Asp name) ‘become
famous’.
Apparently, Lü’s ‘bound/free morpheme’ proposal is insufficient to explain
the whole picture. For example, xiào 校 ‘school’ is conceived as a semi-free
morpheme in expressions like wŏ xiào 我校 ‘my school’, ga-i xiao 該校 ‘this
school’, dào xiào 到校 ‘arrive at school’, lí xiào 離校 ‘depart from school’ in Lü’s
system. However, according to Lü’s five categories, it is difficult to explain why
the same expressions become ungrammatical when the words used with xiào
are changed to polysyllabic ones, such as in *wŏmen xiào 我們校 ‘our school’,
*zhège xiào 這個校 ‘this school’, *dàodá xiào 到達校 ‘arrive at school’, *líka-i
156 Prosodic register grammar
xiào 離開校 ‘depart from school’. Given the same syntactic and semantic environments where xiào occurs, the characterization of ‘bound/free morpheme’
cannot explain why xiào is valid for use in the former group but not in the
latter cases.
The [2+2] syllable-constraint is proposed in Lü’s (1963) paper, “A Preliminary Investigation on Mono-/Disyllabic Issues in Chinese”, where five types of
[2+2] (two syllables + two syllables) phenomena are listed:
26. 1) light verbs like jìnxíng 進行 ‘carry out’ and jia-yı̆ 加以 ‘infect’ that must
take disyllabic gerunds (i.e., 名動詞 in Zhu’s [1982] term) as their
object (i.e., [vv +VV]Gerund);
2) disyllabic verbs like guănlı̆ 管理 ‘manage’ that must take disyllabic nouns
as their object (i.e., [VV OO]);
3) objects must be disyllabic before a disyllabic verb, such as ga-ngtiě 鋼鐵
‘steel’ in ga-ngtiě she-ngchăn 鋼鐵生產 ‘steel production’ (i.e., [OO VV]);
4) disyllabic adjectives must modify disyllabic nouns, such as wěidà 伟大
‘great’ in wěidà rénwù 偉大人物 ‘great character’ ([AA NN)); and
5) a disyllabic word must coordinate with another disyllabic word such as
gòumăi yŭ yuèdú 購買與閱讀 ‘buy and read’ ([XX & XX]).
Lü’s observation is very important, but he leaves many questions unanswered,
such as why the disyllabic constraint is not on the size of the words themselves,
but on the environments in which the words occur, and on the accompanying
word with which a disyllabic word combines.
Following Lü’s ‘bound-free’ observation and Feng’s disyllabic discussions,
Dong (2004) defined the monosyllabic forms more strictly as “semi-free morphemes in disyllabic environment”. As a result, the qualified ‘semi-free’ forms in
her system are different types of syntactically conditioned monosyllabic words,
such as:
27. 1)
2)
3)
4)
5)
Adverb Verb
Auxiliary Verb
Negator
Determinative Noun
Noun Localizer
Although Dong’s study has been successful in bringing much attention to the
computation process on Chinese wordhood and lexicon, there is no prosodic
consideration for why Chinese words are sensitive to syllabic length. In addition to the lack of prosody there is no register grammar used to distinguish the
disyllabic ones from the monosyllabic forms in different register situations. Also,
many well-tested monosyllabic Qiàn-ŏu Cí are excluded from Dong’s study. As
pointed out by Wang (2008:307), a great number of Qiàn-ŏu Cí are left out
because Dong (2004) did not recognize VO forms as phrases, and thus nouns
occurring in the object position (e.g., chua-n yı- 穿衣 ‘put-on clothing’, dào xiào
Prosodic register grammar
157
到校 ‘arrive at school’) are excluded from her list. As a result, the inventory of
all the semi-free monosyllabic words in Modern Chinese is incomplete.
6.4.2 The notion of ‘syntactically free but prosodically bound’
The questions raised by Lü’s (1962) and Dong’s (2004) work demand a development of a new theory.The issue is first tackled in a study of teaching Chinese
as a second language. Feng (2003b:53–63) characterizes the first-type phenomenon, i.e., the monosyllabic words used with another monosyllabic word (i.e.,
Qiàn-ŏu Cí) in terms of ‘syntactically-free but prosodically-bound’ in the newly
developed Prosodic Word (PrWd) framework (see Pan et al., 2004 for comments on this approach). How could a syntactically free form be prosodically
bound? It is somehow a new category in general linguistics that we have never
encountered before. Let’s look at the following examples first:
28. Pronoun + xiào
a. 我校
wǒ xiào
I school
my school
b. *我的校
*wŏ DE xiào
I POSS school
my school
c. *我們校
*wǒmén xiào
we school
our school
d. *我們的校
*wǒmén DE xiào
we POSS school
our school
29. Determiner + xiào
a. ga-i xiào 該校 ‘this school’
b. běn xiào 本校 ‘our school’
c. guì xiào 貴校 ‘your school’
30. xiào + localizer
a. xiào nèi 校內 ‘inside school’
b. xiào wài 校外 ‘outside the school’
c. xiào páng 校旁 ‘beside the school’
d. xiào qián 校前 ‘in front of the school’
158 Prosodic register grammar
The above examples such as wǒ xiào 我校 ‘my school’, ga-i xiào 該校 ‘this school’
and xiào nèi 校内 ‘inside the school’ show that xiào must be analyzed as a syntactically free morpheme otherwise it cannot be used directly with a functional
element such as a pronoun, a determiner, a preposition and a localizer. In other
words, a lexical form combining with a functional element must be considered
as a phrase, not a compound. This can further be seen in the fact that xiao
‘school’ can occur with a preposition, for example:
31. Preposition + xiào
a 距校約十裡之遙，有一個小村落。
jù xiào yue- shí lı̆ zhı- yáo，yŏu yı- gè xiăo cu-nluò.
‘There is a small village ten miles away from this school.’
b. 她的男友是在校讀書時認識的。
ta- de nányŏu shì zài xiào dúshu- shí rènshi de.
‘She met her girlfriend when they were in school.’
c. 她們為校作出了重大的貢獻。
ta-men wèi xiào zuòchu- le zhòngdàde gòngxiàn.
‘They have made great contributions for the school.’
d. 教育部以校為單位進行籃球比賽。
jiàoyùbù yı̆ xiào wéi da-nwèi jìnxíng lánqiú bı̆sài.
‘The education of ministry has divided these schools into groups.’
e. 比賽時按校分組。
bı̆sài shí àn xiào fe-n zŭ.
‘The competitions are conducted among different groups of schools.’
Although prepositions in Chinese can be analyzed as co-verbs, there is no P+N
compound in Chinese and thus, and all the [Preposition+xiào] forms must be
analyzed as preposition phrases, according to which, xiào 校 ‘school’ must be a
free form independent of the proposition. However, the syntactically free form
is not utterable in terms of prosody. In other words, xiào is not free because it
cannot stand alone by itself on the one hand, and on the other hand, it cannot
be attached (or occur with) any longer form other than a monosyllabic one (if
permitted by syntax and semantics), thus giving rise to a unique category with
general syntactic but specific prosodic properties in Modern Formal Chinese
(see Feng 2003a for a detailed analysis). In other words, xiào, and all others
of this kind, must occur within a disyllabic template in order to be effable in
Modern Chinese. Obviously, the disyllabic template is what we have seen in
Chapters 3 and 4 of this book, namely, a standard foot which constitutes the
Minimal Word constraint.
Thus, *woˇmén xiào 我們校 ‘our school’, *wo˘ de xiào 我的校 ‘my school’ and
*woˇmén de xiào 我們的校 ‘our school’ in (28) are all unacceptable because they
violate the prosodic boundedness constraint – a monosyllabic word from Classical Chinese must be used within a disyllabic template – a standard prosodic
Prosodic register grammar
159
category (foot) developed since Han Dynasty.14 Note that in the present analysis, the notion of ‘semi-free’ given by Lü (1962) is not correct because logically
it means that those forms are simultaneously bound and free or sometimes are
bound and sometimes free (or even some situations are bound and some situations are free) as Lü (1962) (and his followers) has assumed. The seemingly
bound-effect (i.e., not effable by itself alone 不能單說 in Lü’s terms) is actually
a result of violations of the prosodic constraint of disyllabic templates on classical auditory intelligible forms (as examples given in [11]). Given this analysis,
it follows that the semi-free effect comes from two factors: one is syntactically
free and the other is prosodically bound, giving rise to the semi-, not fully free
effect. Note that, even if a form is syntactically free, it does not mean that it is
not subject to restrictions in syntax at all. For example, monosyllabic nouns in
Chinese can undergo a reduplication process to mean ‘every N’, such as rén 人
‘person’ → rén-rén 人人 ‘every person, people’ and jia- ‘family’ → jia--jia- 家家
‘every family’, etc. However, xiào 校 ‘school’ cannot be reduplicated as *xiàoxiào 校校 to mean ‘every school’. Of course there must be reasons for why it
cannot do so even if we don’t know yet at this moment,15 and it has nothing to
do with their syntactic characteristic of being a free form, simply because even
the non-controversial free form shı˘ 使 ‘to make, to cause’ cannot take an aspect
maker as most verbs do.
6.5 Conclusion
In this chapter, we have seen that the written Chinese as a formal device in the
system of register grammar is in fact a new language (or diglossia, so to speak)
independent of spoken Chinese and having matured only recently.We have discussed, based on the system of prosodic grammar, some of the features of formal
written Chinese that consist not only of monosyllabic words used in disyllabic
templates (Qiàn-ŏu Cí) but also of disyllabic words used in disyllabic couplets
(Hé-ŏu Cí) in the syntax. Three principles, namely, the Register Coherence,
the Principle of Auditory Intelligibility (PAI) and the Principle of Shaping by
Prosody (PSP) are proposed and they operate with some colloquial features in
order to make the language more natural. In addition, the Amalgamation Principle is also proposed to modulate literary diction with colloquial expression.
Finally, a method of calculation is developed for the quantitative analysis of the
degree of formality of a specimen of Chinese writing.
As for Qiàn-ŏu Cí, it is argued that they constitute a unique category in Chinese grammar that is syntactically free but prosodically bound, which captures
what traditionally has been characterized as being used independently but not
uttered freely (單用不單說). Given the syntactic and prosodic properties of
Qiàn-ŏu Cí，it seems natural that the traditional characterization of semi-free
morphemes are insufficient to explain their prosodic as well as syntactic behaviors (i.e., independently used syntactically 單用, but ineffable prosodically 不
單說) and fails to capture the result of interaction between syntax and prosody.
For the Hé-ŏu Cí, we have seen that the [2+2] syllable-constraint is indeed
a reflection of a PrWd-compound effect in prosodic morphology, which
160 Prosodic register grammar
generates many disyllabic words, some of which also require disyllabic words
to go with them, according to the register grammar. This theory provides a
powerful explanation for the concerned phenomena in the literature (see also
Wang 2014, Feng 2016a, Jia 2015 and especially H. Wang 2008:310–11 for
detailed comments on this work). Of course, there are still questions of whether
the templative [2+/...2] requirement is a specific effect or an across-the-board
constraint in prosodic grammar and register grammar. Furthermore, to what
extent the prosodic-templative phenomena can be considered as a grammar
and whether it is operated before or after Spell-Out, all of these questions are
worthwhile to be explored in future studies.
To sum up, all these results given in this chapter unquestionably have deep
theoretical and empirical implications. Theoretically, we now have a new perspective for how and why a new system of a language is born and how the
diglossic (or even triglossic) requirements are forced upon and satisfied by
language communicative functions, giving rise to innovative changes of the
language, as well as the birth of a new study, the prosodic register grammar.
Empirically, this study reveals that a statistical measurement for degree of formality could also be extended to a wide range of related fields, such as composition testing, readability scaling, style gradation, textbook compilation, second
language learning, literacy acquisition, etc. All of these are important areas
remaining for future research.
Notes
1 Elevated forms can be evolved into formal ones when they are frequently used in modern Chinese.
2 The original Chinese reads: “中國語言音單調複，故往往變單字為雙字。而每一
名有單名雙名二者。如天稱皇天、昊天是也。雙名在質言為贅疣，而于文言則
需用之。由於音單調複，單名多變為雙名，因之製成詩歌、駢文等文體。他如
有節奏之語言，也皆由此形成也。” 《文字聲韻訓詁筆記》99頁。
3 The original Chinese reads: 語體是實現人們在直接交際中最元始最基本屬性的、
用語言來表達或確定彼此之間關係和距離的一種語言機制。
4 Here the term ‘Classical Chinese’ refers to expressions that are commonly found in premodern Chinese writings but are not used in the everyday speech of today.
5 The original Chinese reads: “當然，這並不是說，凡是漢族所擴及的地區，
原來在這些地方居住的民族都被吸收為漢族了。事實並不如此，
即 在 目 前 ， 漢 族 聚 居 的 地 區 裏 還 是 有 少 數 民 族 雜 居 在 內 。”
——費孝通《中國的現代化與少數民族的發展》，《費孝通文集》卷7，p. 291.
6 The distance can actually be measured in terms of degrees of familiarity, which in turn,
can be calculated by frequency of occurrences in everyday speech.
7 For example, there are systematically developed formal expressions like jìnxíng 進行
‘carry out’:
我們一定要對這個問題進行研究。
wǒmen yídìng yào duì zhègè wèntí jìnxíng yánjiu-.
we definitely need towards this issue carry-out study
‘We must study this question (we must carry out study on this question).’However, in
oral speech, native speakers do not use them at all. For example:
瞧！您的閘不靈了，得修一下了。 / *得進行修了。
Prosodic register grammar
161
Qiáo! Nín de zhá bù líng le, děi xiu- yíxià le. / *děi jìnxíng xiu- le.
Look you POSS brake not effective Asp. must repair a little Asp. must carry-out repair Asp.
‘Look, your (bicycle) brake has some problems, and you have to fix it.’
Only on formal occasions, e.g., regulating transportation, can the same thing can be
paraphrased formally as:
閘皮失靈，必須進行修理！
zhá pí shı- líng, bìxu- jìnxíng xiu-lî.
Brake rubber lose efficiency, must carry-out repair.
‘If the brake is not efficient, one must repair it.’
8 One could argue that fǎngfú 仿佛 is more “literary” than sìhu- 似乎, as opposed to more
“formal”.The question of clearly distinguishing literary from formal is a topic for future
research.
9 How to decide whether one expression is more formal than another is a problem that
requires more study. Here, I make use of both a native speaker’s intuition and the frequency of occurrences inside relevant syntactic environments within our database.
10 This principle was developed on the basis of Zhang’s (2002) ear-measurement strategy
as explained in section 6.2. below.
11 The original Chinese reads: “採用少數（古代）詞語，要怎麼樣才是融會而不是攙
雜。情況千變萬化，很難具體說明。勉強說，可以用耳朵作個尺子，量一量，
凡是聽起來生硬，明顯覺得不像日常說話（包括談論學術問題）的，是攙雜而
未融會，反之是已經融會而不是攙雜。” (張中行《文言津逮》)
12 Here I only intend to give a schematic diagram. The accurate and precise result can be
easily formed by appropriate techniques.
13 The examples are taken from Feng 2003b.
14 Further research is needed in this area in order to characterize the grammatical functions
of the disyllabic template in terms of a prosody-syntactic device for elevated registers.
15 A plausible answer can be suggested as follows: the grammar of noun reduplication (meaning “every noun”) is an operation only for informal speech, not for elevated register.
References
Crystal, David and Derek, Davy. 1969. Investigating English Style. London and Harlow: Longmans Green and Co. Ltd.
Dong, Xiufang. 2004. Chinese Lexicon and Morphology 漢語的詞庫與詞法. Beijing: Peking
University Press.
Enkvist, Nils Erik. 1973. Linguistic Stylistics. The Hague: Mouton.
Fei, Xiaotong. 1999. The Modernization and Minority Development in China
中國的現代化與少數民族的發展. In Collected Essays of Fei Xiaotong 費孝通文集 (Volume 7), 288–301.
Feng, Shengli. 2003. “Prosodically Constrained Postverbal PPs in Mandarin Chinese”. Linguistics 41.6: 1085–1122.
Feng, Shengli. 2003a. “The Prosodically Restrained Written Language and the Audio-lingua Approach 韻律制約的書面語與聽說為主的教學法”. Chinese Teaching in the World
世界漢語教學 1: 87–97.
Feng, Shengli. 2003b. “An Independent Grammar for Written Chinese in Second Language
Teaching” 書面語語法及教學的相對獨立性. Language Teaching and Linguistic Studies
語言教學與研究 2: 53–63.
Feng, Shengli. 2005. Prosodic Grammar in Chinese 漢語韻律語法研究. Beijing: Peking University Press.
Feng, Shengli. 2006. Expressions of Written Chinese 漢語書面用語初編. Beijing: Beijing Language and Culture University Press.
162 Prosodic register grammar
Feng, Shengli. 2010a. A Theoretical Exploration of Prosodic Syntax. Essays on Linguistics
語言學論叢 39: 204–244.
Feng, Shengli. 2010b. “On Register Mechanism and Its Grammatical Functions
論語體的機制及其語法功能” . Chinese Philology 中國語文 5: 400–412.
Feng, Shengli. 2011. “The Literary Function of Stylistic-register Grammar
語體語法及其文學功能”. Contemporary Rhetoric 當代修辭學 4: 1–13.
Feng Shengli. 2016. Prosodic Morphology. In Wolfgang Behr, Gu Yueguo, Zev Handel, C.-T.
James Huang, and Rint Sybesma (eds.) Encyclopedia of Chinese Language and Linguistics.
Leiden: Brill.
Feng, Shengli and Jan,Vihan. 2016. The historical evidence for tripartite division of Formal,
Informal, and Elevated Registers: The Airs, the Elegantiae, and the Eulogia of the Odes.
In Barbara Meisterernst (ed.) New Aspects of Classical Chinese Grammar. Wiesbaden: Harrassowitz, 47–74.
Feng, Shengli, Jie,Wang and Mei, Huang. 2008. “An Automatic Feature Checking Algorithm
for Degree of Formalities in Written Chinese 漢語書面語體莊雅度的自動測量”. Linguistic Sciences 語言科學 2: 113–126.
Huang, Kan. 1922/2001. Huang Kan Diary 黄侃日記. Nanjing: Jiangsu Educational Press.
Huang, Kan. 1936/2013. Notes on Paleography. In Huang Zhuo (ed.) Ancient Phonology and
Exegesis in Chinese 文字聲韻训诂笔记. Wuhan: Wuhan University Press.
Jia, Linhua. 2015. A Study on He’Ou Disyllabic Words in Modern Written Chinese. Ph.D. dissertation. Beijing: Beijing Language and Culture University.
Kiesling, Scott F. 2009. “Style as Stance: Stance as the Explanation for Patterns of Sociolinguistic Variation”. In Alexandra Jaffe (eds.) Stance: Sociolinguistic Perspectives. Oxford: Oxford
University Press, 171–194.
Kroch, Anthony. 2000. “Syntactic Change”. In Mark Baltin and Chris Collins (eds.) The
Handbook of Contemporary Syntactic Theory. Oxford: Blackwell.
Labov, William. 2014. Speech by Professor William Labov. In Feng, Shengli and Virginia,Yip
(eds.) A Dialogue on Sound Change Between William Labo and William S.-Y. Wang. Beijing:
Peking University Press.
Lü, Shuxiang. 1962. “On ‘free’ and ‘bound’ 說‘自由’和‘粘著’”. Chinese Philology 中國語文
1: 1–6.
Lü, Shuxiang. 1963. “An Preliminary Research on Mono-disyllabic Combinations in Modern Chinese 現代漢語雙音節問題初探”. Chinese Philology 中國語文1: 11–23.
McCarthy, John and Prince, Alan. 1986. Prosodic Morphology. MS: University of Massachusetts
and Brandeis University.
Pan, Wenguo, Ye, Buqing and Han, Yang. 2004. Studies of Chinese Word-Formation
漢語的構詞法研究. Shanghai: East China Normal University Press.
Wang, Hongjun. 2008. Non-Linear Phonology of Chinese 漢語非線性音系學. Beijing: Peking
University Press.
Wang, Li. 1993. Essays from the Longchong Bingdiao Studio 龍蟲並雕齋瑣語. Beijing: China
Social Sciences Press.
Wang, Yongna. 2009. Study on Grammar Means of Formal Written Chinese.
漢語書面正式語體的語法手段. Ph.D. dissertation. Beijing: Beijing Language and Culture University.
Wang, Yongna. 2012. “The Structural Mechanism of Predicative Coordinating Verbal Phrases
Using the Conjunction hé in Written Chinese 書面語體“和”字動詞性並列結
構的構成機制”. Chinese Teaching in the World 世界漢語教學 2: 188–197.
Prosodic register grammar
163
Wang, Lijuan. 2014. Morphological Prosody in Chinese 漢語的韻律形態. Beijing: Beijing Language and Culture University Press.
Yang, Charles. 2016. The price of linguistic productivity: How children learn to break the rules of
language. Cambridge, MA: MIT Press.
Zhang, Zhongxing. 2002. Gateway to Literary Chinese文言津逮. Beijing: Beijing Press.
Zhu, Dexi. 1982. Teaching Materials on Syntax 語法講義. Beijing: Commercial Press.
7Compound prosodic
word – Sizige 四字格
This chapter discusses the formation of the Compound Prosodic Word, traditionally called Sìzìgé (Four-Character Pattern, FCP in short) in Chinese morphology. It is shown how the double-foot prosody, or the Compound Prosodic
Word is generated through ways of duplicating prosodic words, which results
in two fixed stress patterns, and how the results equip the FCP with expressive
power in formal, informal and elevated registers.
1. What is Sìzìgé?
Sìzìgé ‘four-character pattern’, literally meaning ‘four-character box’, is a popular term in Chinese expressions which is itself formed by four characters. As
seen in Chapter 3, each character in Chinese corresponds to a morphosyllabic
unit, so ‘four-character’ is naturally interpreted as four syllables.
The four-syllable patterns in Chinese are a unique prosodic category in Chinese morphology and literacy, which is not only used as a powerful expressive
diction in verbal communications but also as a powerful rhythmic pattern in
Chinese prosodic phonology. Take the famous exmaple given by Yuen Ren
Chao (1968):
wú fèi bìng niú 無肺病牛
1. no lung disease cow
a) There are no cows that have lung‑disease. (wú((fèi bìng) niú))
b) Sick cows that have no lungs. ((wú fèi ) (bìng niú))
c) Cows without lung‑disease. ((wú(fèi bìng)) niú)
The phrase wú fèi bìng niú has three semantic interpretations with different
syntactic bracketing:
2. a. (A ((B C) D))
b. ((AB) (CD))
c. ((A (B C)) D)
Compound prosodic word – Sizige 四字格
165
However, only the second reading is favorable for native speakers (see Chao, 1968).
This indicates that, if a disyllabic foot grouping can be satisfied in a given string
of syllables, no polysyllabic foot is possible. Note that the favored reading given
in (b) is semantically odd: a sick cow without lungs, while the other two readings
are both semantically more natural: (i) There are no cows that have lung disease
and (iii) cows without lung disease. The odd choice among these three readings
indicates that quadrisyllabic strings must be analyzed as two units (i.e., the foot,
in prosodic terms), even though the choice would result in a semantic anomaly.
What is Sìzìgé then? Now we see that the first answer to this question
would be this: Sìzìgé is a prosodic pattern acquired by Chinese among all levels/classes of native speakers. While this is true, we may also like to realize that
it does not abound in Chinese. Tsou (2014) has recently pointed out that the
Sìzìgé四字格 ‘Four Character (idiomatic) Patterns’ (FCP for short), functioning as figurative speech, are linguistic markers of the language user’s erudition
and elevated cultural status, and it is also a rhetorical flair associated with the
highest form of literary traditions and the language arts since Qin (200–221
BC) and Han (221 BC–300 AD) dynasties, as we have seen in Chapter 2 of this
book about the development of disyllabicity.1 The power of the FCE, like that
of the characters themselves, are abundant not only in Mandarin and all Chinese dialects and minority languages in China, but also in many neighboring
languages such as Vietnamese, Japanese and Korean, as seen from the following
examples (taken from Tsou 2014). For example:
3. a. Japanese: 一期一會 いちごいちえ i chi go i chi e
‘To cherish the opportunity of serving customers.’
b. Korean: 馬腳露出 마각노출 munjeon-ogdab
‘Conspiracy exposure’
c. Vietnamese: 井底之蛙 tinh de chi oa Near-sighted
‘Narrow minded’
d. Thai: 趁熱打鐵ตีเหล็กเมื่อร้อน tee rek mea ron
‘Strike the iron while it is hot，or Make hay while the sun shines.’
e. Min Dialect：鴨仔聽雷 Ah-á thiann luî.
‘talking to a wall’ or ‘talking to a brick wall’
This situation raises wide-ranging questions, for example: What is the nature
and function of the FCP in the Chinese language and society and why might
this be emblematic of the language and culture? Also, do they represent horizontal or vertical transmission of linguistic and cultural traits in Asia? On the
basis of prosodic morphology given in this book and in light of the above
questions, I shall examine the linguistic processes employed as prosodic devices
in the FCP and their functionalities, starting with the facts about the FCP in
Mandarin Chinese.
166 Compound prosodic word – Sizige 四字格
2. Facts about four-character patterns
The FCPs are commonly used to express many different things, varying from
common notions to what in English would be translated with a proverb or
idioms. For example:
4. Informal
a. 你來我往
nǐ lái wǒ wǎng
you come I go
‘back and forth’
b. 連蹦帶跳
lián bèng dài tiào
jump
with hop link
‘jumped up and down, bounced’
c. 稀裡糊塗
hı- li hu-tu‘xili muddle (xi-li as foot template filler)’
‘muddleheaded’
Formal
d. dé
ga-o
wàng
zhòng 德高望重
virtue
high
name
heavy
‘Have a high virtue and a glorious name.’
e. wàn
shòu
wú
jia-ng 萬壽無疆
10-thousand-year age
no
limit
‘a long life’
f. qián
chezhıjiàn 前車之鑒
before
carriage
’s
mirror
‘The preceding carriage’s mirror’ (condensed from “The fall of the
preceding carriage serves as a mirror for the following one.”)
g. jìn
tuì
liǎng
nán 進退兩難
advance
retreat
two
difficulties
‘between a rock and a hard place’
Elevated2
h. ga-o sha-n
yǎng
zhǐ 高山仰止
high mountain look-up Prt.
‘look up to the high mountain > to admire someone greatly’
i. yǐ jiè
méi shòu 以介眉壽
prt. wish
brow age
‘Best wishes for a long life.’
j. yı-n jiàn
bù yuǎn 殷鑒不遠
yin mirror
not far
‘The Yin-dynasty mirror is not far > the former mirror.’
Compound prosodic word – Sizige 四字格
k. jìn
tuì
wéi
advance
retreat
be
‘between a rock and a hard place’
167
gǔ 進退維谷
valley
As seen above, the more ancient the expressions are, the higher the register
will be. The remarkable use of the four-syllable expressions has earned them
different names, each with little semantic differences, of which the typical ones
are Sìzì Géyán 四字格言 ‘four-character aphorism’, Sìzì Chéngyǔ 四字成語
‘four-character idiom’ or simply, Sìzìgé 四字格 ‘four-character pattern’ and so
on. However, different as their names might be, one crucial thing is the same:
they are all composed of four syllables with a rhythmic pattern of 2x2. In what
follows we will use the four-syllable pattern or Sìzìgé as a cover term to explore
why they are so unique in Chinese.
3. Four-character patterns as a product of the prosodic
morphological system
How are four-character patterns generated in Chinese morphology? As seen in
previous chapters (see also, Shih 1986, Feng 1995, Duanmu 2000) (disyllabic
foot) formation has important impacts on Chinese morphology, and the Natural Foot Formation is formulated in (5) (‘σ’ represents syllable).
5.
f
σ
σ
Natural Foot Formation (NFF)
A natural foot in Chinese is grouped by two syllables from left to right and the stray
syllable is attached to the neighboring foot when the number of syllables is odd.
One of the effects of the Natural Foot Formation (NFF) on Chinese grammar is
the morphological process of compounding, which results in the prosodic word
formation determined by the NFF in (5), as seen from the following diagram in (6).
6.
Phrase
Compound
Prosodic Morphology
PrWd
f
σ
σ
Under this system, 80% of dictionary entries are made up of compounds
shaped by a disyllabic or trisyllabic template (i.e., PrWd). Another striking
168 Compound prosodic word – Sizige 四字格
impact on prosodic word formation in Chinese is the combination of two
PrWds (see Feng 1997a), as illustrated in (7).
7.
Minimal Combination of PrWds
PrWd
σ
PrWd
σ
σ
σ
The Minimal PrWd combination, like a combination of two lexical words (cf.
brown sugar), can be further analyzed into two subcategories in natural speech:
one is a minimal phrase (hence forth, PrWd-phrase), and another is a double-feet
compound (hence forth, PrWd-compound), as shown in (8a) and (8b), respectively.
8.
a.
PrWd-Phrase
PrWd
σ
b.
PrWd-Compound
PrWd
σ
σ
PrWd
σ
σ
PrWd
σ
σ
σ
(8a) under this system will generate the forms seen in (4a,b) as PrWd-phrases,
while (8b) will produce the forms listed in (4c-k) as PrWd-compounds.
What is the difference between a PrWd-Phrase and a PrWd-Compound
then? Although it is not easy to make a clear cut between the two, it is detectable that there is a very short break which occurs in the juncture of the two
PrWds in the PrWd-Phrases, while no such break is attested in the PrWdCompound. Compare (bracketing represents rhythmic group, ‘#’ represents a
juncture):
9. 你來 我往
(nǐ lái) (wǒ wǎng)
you come I go
‘Back and forth’
一衣帶水
(yıyı-)
(dài
shuǐ )
one cloth
belt
river
‘a narrow strip of water (be continuous to
each other’
In addition to the ‘short break’ test, the difference between the PrWd-Phrase
and PrWd-Compound can be captured by the stress patterns discussed by
Yu (1989), Hoa (1983), Vigário (2003) and Meredith (1990:154). It is then
Compound prosodic word – Sizige 四字格
169
proposed (Feng 1997a) that there are two stress patterns observed in the PrWdCompounds, as illustrated in (10).
10. Stress Pattern of PrWd-Compound
a. [(02) (13)]
yıyıdài
shuǐ
One
cloth belt
river
‘close relation’
b. [2 0 / 1 3]
xı- li
hútu- (see [1c])
Xi li
muddle (xi-li = foot template filler)
‘muddleheaded’
In Chinese the so-called four-syllable idioms 四字成語 are all uttered with a
pattern of [0213] as in (10a), which are different from the colloquial ones that
have a [2013] stress pattern as seen in (10b). What is important to note here
is the fact that the ones categorized as PrWd-Phrases do not exhibit the same
type of stress pattern. Instead, they have two parallel [w s] units that co-exist as
a pair, as seen in (11) (‘s’ stands for strong while ‘w’ for weak).
11. PrWd-Phrase
[w s # w s]
nǐ lái wǒ wǎng 你來我往
you come I go
‘back and forth’
Whether or not the categorizations in (10) and (11) will accommodate all the
four-character patterns in the language is still an open question; it is, however,
unquestionable that the double-footed rhythm, whether in a PrWd-Phrase or
a PrWd-Compound, is so powerful that it could force (or reanalyze) an ordinary four-character noun phrase into a 2x2 rhythm, creating either a Compound-PrWd in the lexicon, or a parallel PrWd-Phrase in natural speech. For
example:
12. a. yı- yı- dài shuǐ 一衣帶水
[One [cloth belt]] river]
Lit: a river like a belt in width
Re: a narrow strip of water
‘close neighbours, separated by only a strip of water’
b. wú fèibìng niú 無肺病牛
[no [lung disease]] cow]
(i) There are no cows that have lung disease. (wú((fèi bìng) niú));
(ii) Cows without lung disease. ((wú(fèi bìng)) niú);
(iii) There are sick cows without lungs((wú fèi) (bìng niú)).
170 Compound prosodic word – Sizige 四字格
In (12a), the original meaning and structure of the idiom has become unrecognizable for most native speakers with limited education, which is caused by
a mismatch between the syntactic structure and the prosodic structure (i.e.,
the double-footed PrWd-Compound). In (12b) Wú fèibìng niú, on the other
hand, there are syntactically and semantically alternative readings from the foursyllable string of “no lung sick cow (no cows with lung disease)”, but the
favorite reading is the illogical one, i.e., (12b-iii), which shows, once again, how
powerful the [2x2] prosody is in Chinese.
Other languages also provide evidence for the existence of CompoundPrWd in human languages. (See more discussions in Kenstowicz 1995 and Vigario 2003 for arguments on “a prosodic word that dominates two constituents
of the same type = standard prosodic word”).
4. Syntactic structure of four-character expressions
How are Sìzì Gé formed structurally? It is widely recognized that one characteristic of the Sìzì Gé formation is that any major syntactic relations, namely,
Subject+Predicate, Verb+Object, Verb+PP, Adverb+V, Coordinating Structure,
etc. is capable of forming a four-character pattern either as a Compound-PrWd
or a Phrasal PrWd. For example:
13. Subject+Predicate
sha-n qióng shuǐ jìn
mountain end
river exhaustive
‘the end of hills and rivers’
山窮水盡
Verb+Object
zha-ng yá wǔ zhǎo
張牙舞爪
show fang brandish claw
‘bare fangs and brandish claws – make threatening gestures’
verb+PP
yè jı-ng yú qín
work excellence from diligence.
‘Efficiency comes from diligence.’
業精於勤
Adverb+V
jı-ng dia-o xì kè 精雕細刻
refinedly engrave precisely carve
‘work at sth. with the care and precision of a sculptor > work with
great care’
Coordinating
she-ng sǐ cún wáng 生死存亡
life die survive extinct
‘of vital importance’
Compound prosodic word – Sizige 四字格
171
Modifier+N
jǐng dǐ zhı- wa-井底之蛙
well bottom ’s frog
‘a frog in a well > a person of narrow view’
Duplication
zhàn zhàn jı-ng jı-ng戰戰兢兢
fear-fear vigilant-vigilant
‘be in a blue funk’
As seen above, as long as a four-character string can be prosodized into a [2x2]
rhythmic structure, it can be analyzed as PrWd-Phrase (with an [s w/s w] stress
pattern) or a Compound-PrWd (with a [0213] or [2013] pattern), depending
on the stylistic nature of the words involved in the string, which is a topic we
will discuss in the next two sections.
5. The prosodic category of Compound PrWd
According to Derek Attridge (1982), the four-by-four structure (four-beat line) is
the most basic form of all English verses, for most nursery rhymes and folk songs
tend to be composed in this simple meter. Similarly, four-character patterns also
exhibit a similar effect in Chinese since they are not only a prosodic unit in Mandarin Chinese, but also a poetic line in Archaic Chinese. Also, they are analyzed
as a “four-beat line” expression with specific metricalized structure of [2x2] (a
prosodic group [or poetic line] formed by two units followed by two syllables).
Feng (1995, 2002) observes that even if compounds must first be a PrWd,
it is not necessary that every PrWd is a compound. Compounding is only one
of the various ways to realize PrWds. PrWds that are formed by epenthesis, for
example, are obviously not compounds, as follows:
14. Words
‘search’
‘blink’
Medieval Chinese
*ziIm
>
*tʃɣɛp
>
Modern Mandarin
xún-me
zhǎ-me
By identifying the above type of prosodic word in Mandarin, Feng (1995)
argues that the following four-syllable string is also formed by adding epenthetic material to the base form and it should be considered as an independent
prosodic unit of some kind. For example,
15. yı--ga-n-èr–jìng
一乾二淨
one-tidy-two-clean
thoroughly, completely
ta- bǎ fàn chı-de yı- ga-n èr jìng. 他把飯吃得一乾二淨。
he BA food eat-De one-tidy-two-clean
‘He finished the food completely.’
172 Compound prosodic word – Sizige 四字格
Since gan-jing ‘clean’ is a compound or a B(ase)-compound formed by two
root morphemes in (15), the morphemes ‘one’ and ‘two’ are taken from the set
of natural numbers. The four-character patterns in Chinese (as in (15)), as Feng
argued (1995), should be recognized to be a category of Compound-PrWd
(thus C-PrWd) “which are formed by inserting two Place Holders (PH) into
a B-compound (or a PrWd), and each morpheme/word of the B-Compound
(PrWd) gets one place holder in order to form a new foot” (Feng 2002:136).
In other words, Compound PrWds are not necessarily generated by tree structures given in Chapter 5, and they can also be generated by morphological
processes of epenthetization, so that some of them can be formed by adding
extra materials to a base-form (a PrWd) to fulfill the two-foot requirement of
being a Compound PrWd. This is the same operation used to augment an element to a monosyllabic word to satisfy the foot requirement of PrWds as seen
in Chapter 4. The following examples prove that this is indeed the case (taken
from Feng 2002:36): one compound (PrWd) is augmented into two feet and
then satisfies the requirement forming a Compound PrWd:
16.
As seen above, the two place holders can (randomly) make use of natural
numbers such as yı--èr ‘one-two’ in (15) or qı--ba- ‘seven-eight’ in (16a). Example
(16b) is even more convincing: yǎng zhu- 養豬 ‘raise pig’ is a VO phrase and
forms a PrWd given the theory presented in Chapter 4 above. However, dàyǎng
-zhu- 大養豬 ‘greatly raise pig’ violates the prosodic principle for PrWds, thus
being ruled out. However, a semantically empty particle qí (originally meaning
‘it’ in Classical Chinese) is inserted into the PrWd yǎng-zhu-, so that the adverb
dà ‘greatly’ functions like another place holder working together with qı- to
license the two feet.
Thus, Compound-PrWds should be considered as an independent prosodic
unit in the language they occur because they are different from two-compoundword phrases (or two-prosodic-word phrases) even if two two-compound
words can also be used to form a Compound-PrWd.The fact that the insertion
of place holders (extra syllables) can form a two- (rather three- or four-)feet
unit proves that the four-syllable strings are requirement of a prosodic category
of some kind that is in accordance with a morphological category.The prosodic
Compound prosodic word – Sizige 四字格
173
category is Compound-PrWd while the morphological category is traditionally named four-character patterns (FCP).
A crucial question regarding to the nature of the FCP arises as to whether
or not all four-character patterns are Compound PrWds? Note that a [2x2]
four-syllable group which is intentionally created in the language is not necessarily a Compound PrWd. That is, they can perfectly be two-feet phrases (or
prosodic phrases). The question then is on what condition can we say that the
[2x2] expression such as yı- qı-ng èr chǔ 一清二楚, as a group of two PrWds, is
a Compound PrWd, rather than a two-PrWd Phrase which is either/both a
syntactic phrase of two compounds or prosodic category of a two-prosodicword phrase? These questions lead to a more general one: what is the difference
between the prosodic phrase and prosodic word, or what is the nature of the
prosodic constituent formed by two prosodic words? We know that word-word
compounds are composed of two (in Chinese) or more (in other languages)
independent prosodic words in general.The words that form a compound have
the same phonological shape as when they appear in isolation. Thus, phonologically, there are at least two possibilities of the combining results: (a) phonological phrase, (b) compound prosodic word, shown as follows:
17.
a.
b.
ωmax
ω
ω
Then, what are the prosodic characteristics of (17b)? The following definition is what Feng (1998) has proposed and we reformulate, as given in (18):
18. Criterion of Compound PrWd:
There is only one prominent element within the category of Compound
PrWd.
That is, the combination of two PrWds regularly forms a compound prosodic
word whose internal prosodic structure of each PrWd is reorganized prosodically into a one-head (one primary stress) unit. This is consistent with Vigario’s
well-formedness condition on prosodic word and compound (or maximal)
prosodic word, as in (19) (Vigario 2003:423).
19. Well-formedness condition on prosodic word domain
1) A minimal prosodic word has one and only one (word) primary stress.
2) A maximal prosodic word has one and only one prominent element.
3) A unit bearing word stress must be included within a minimal prosodic word.
174 Compound prosodic word – Sizige 四字格
These universal properties, as we have seen in previous chapters, entail that a minimal prosodic word is a prosodic word immediately dominating the next lower
prosodic level, i.e., the foot, while a maximal prosodic word, as we will argue in
this section, can be generated by two types of structures as illustrated below:
20. a.
ωMax
ωw
ωs
ω
b.
ω
σ
W
ω
σ
σ
W
S
σ
S
For both (20a) and (20b), there is only one primary stress in each derivation.
This, as I argue in next section, is what actually happened in Mandarin Chinese.
6. A Prosodic algorithm for two types of the FCE in Mandarin Chinese3
Following Feng (1997b), this section provides an algorithm that derives the
[0213] and [2013] stress patterns with the two types of operations as in (21a)
and (21b), illustrated below:
21. a.
‖
‖
Compound prosodic word – Sizige 四字格
b.
175
Compound-PrWd
PrWd
|
Foot-1
.σ
σ
PrWd
|
Foot-2
σ
σ
In order to capture the prosodic results of stress patterns of the Compound
PrWd, Feng (1997b) comes up with what is called the Stress Adjustment Principle, stated as follows.
22. Stress Adjustment Principle (Feng 1997b:46)4
When two identical feet combine to form a new prosodic unit, their original stress must be adjusted according to the principle of making the strong
stronger and the weak weaker.
When two feet are combined into one unit (i.e., a compound PrWd in this
case), the resulting stress pattern of the new unit follows naturally from an operation of “making the strong stronger and the weak weaker” – an algorithm for
four-character patterns in Chinese. The operations are exemplified as follows.
(1) The [0213] stress pattern is derived from the following processes:
23.
i) Base form (the right-strong pattern is based on Chao 1968, Wang and Feng 2009)
PrWd
[w]³
1
³[s]
2
ii) Compoundization of PrWd (through germination):
C-PrWd
PrWd
σ
w
1
σ
s
2
PrWd
σ
w
1
σ
s
2
iii) Unification through stress adjustment
C-PrWd (Sìzìgé ഋᆇṬ˅
w
PrWd
Surface form
σ
w
1
1
0
[0
σ
s
2
2
2
2
s
PrWd
σ
w
1
1
1
1
σ
s
2
3
2
3]
base form
make strong stronger
make weak weaker
176 Compound prosodic word – Sizige 四字格
(2) The [2013] stress pattern is derived from the following processes:
24.
i) Base form
PrWd
[w]³
1
³[s]
2
ii) Compoundization of PrWd (through augmentation)
C-PrWd
Sìzìgé
σ
σ
PrWd-1
w
1
PrWd-2
s
2
σ
σ
w
1
1
1
1
w
w
1
1
0
0
s
2
2
2
[2
.
s
relative prominence
result
make strong (PrWd-2) stronger
make weak (PrWd-1) weaker
surface form
s
2
3
3
3]
(3) The [2013] can be derived also by a split-and-insertion process:
25.
i) Base form
PrWd
σ[w]
1
[s]σ
2
ii) Compoundization (through split-and-insertion )
PrWd-A
PrWd-B
σ
w
1
σ
s
1
σ
w
2
σ
s
2
iii) Unification by stress adjustment
PrWd [s]
PrWd [w]
σ
w
[1
2
2
[2
σ
(w
(1
σ
s)
2)
0
0
1
1
σ
s
2]
3
3
3]
base form
make strong ones (the outer PrWd) stronger
make weak ones (the inner PrWd) weaker
surface form
Compound prosodic word – Sizige 四字格
177
6. The stylistic-register function of four-character
expressions
A notable fact about four-character patterns is that they function differently
in terms of register (see Chapter 6 about register grammar). First, the PrWdPhrases always exhibit a casual or informal style and are used mainly in everyday
speech, for example:
26. a. ta-men liǎngge [nǐ lái wǒ wǎng], chéng le péngyǒu. 你來我往
they two [you come I go], become asp. friend
‘The two interact back and forth, and become friends finally.’
b. wǒmen xiànzài do-u [yǒu chı- yǒu he-]. 有吃有喝
we now all [have food have drink]
‘Now we all have plenty to eat and drink – very good life condition.’
愛來不來
c. ài lái bù lái
like come not come
Lit：like come then come, not come then not
‘regardless of whether (you come or not)’ or ‘who cares (whether you
come or not)’
d. xı- li hu-tu- 稀裡糊塗
‘xi li muddle (xili= foot template filler)’
ta- zuòshì zǒngshì xı-lihútu-, ràng rén bú fàng xı-n.
He work always muddle-headed, make people not put heart
‘He is always muddle-headed when working and people worry about it.’
(26c) is a typical example showing how colloquial Chinese favors a PrWd
Phrase in casual speech: two separate phrases (i.e.,‘if you like to come then do it,
and if you don’t like to come, then forget about it’) is contracted or condensed
into a four-character pattern favored by the PrWd Phrase prosody, which has
become a powerful expressive pattern: [ài V bù V] where any monosyllabic verb
could be inserted in natural speech.
The last example xı-li hu-tu- is even more convincing for why the PrWd
Compound with a [2013] stress pattern is exclusively used in casual speech.
It is because the pattern can realize, as suggested by Feng (2010c), the swift
jumping rhythm (xuáncha-lü` 懸差律), or the Limerick Prosody (for being like
the prosody of a limerick exhibiting a “swift catchy meter” (Perrine, 1963),
which is most suitable for casual and not serious situations. As a result, the
stylistic effects of PrWd-Phrases are expressively vivid, vigorous and succinct
in style.
Of course, an even more remarkable feature of the four-character pattern in
Chinese verbal culture is this: the double-foot prosody has systematically served
formal and elevated purposes in high verbal communications. Thus, examples
like (4f) and (4j) can only be used on formal and elevated occasions.
There are no detailed studies yet for a distinct stress feature between formal
and elevated styles, and from the current research, we assume that both of them
share the [0213] prosody. Effable and auditable ancient words with classical
178 Compound prosodic word – Sizige 四字格
allusions (4h-k) are used to make the elevated style sound more educated and
erudite, like the learned Backing in French (Dell and Selkirk, 1979). In terms
of Stylistic-Register Grammar (Feng 2012), it is the place where people look
for stylistic differences between formal [0213] and informal [2103] and between
higher or lower educations in Chinese verbal expressions.
7. The prosodic syntax of four-character patterns
How are four-character patterns used syntactically? In terms of syntax, the Parallel PrWds (double-foot prosody) are mainly used as predicates, that is, both
the formal [0213] PrWd-Compound and the informal [2013] PrWd-Phrase
are predicates (27a). When they are used as modifiers for both nouns (27b) and
verbs (27c), the possessive (attributive) marker – de and the adverbial marker –
di must be added respectively, as seen below.
27.
a. 總統和副總統同床異夢，各有各的打算。
zǒngtǒng hé fù zǒngtǒng tóng chuáng yì mèng, gè yǒu gè de dǎsuàn.
President and vice president same bed different dream, each have each ’s plan
‘President and vice president hide different purposes behind the semblance of accord, and each has his own plan.’
b. 他們兩個是同床異夢的搭檔。
ta-men liáng gè shì tóng chuángyì mèng de da-dang dàng.
they two CL. be same bed different dream ’s partner
‘They two are partners of dreaming different dreams.’
c. 他們同床異夢地合作了10多年。
ta-men tóng chuáng yì mèngdi hézuò le shí duo- nián.
they same bed different dream di coordinate Asp. 10 more year
‘They pretend to cooperate for more than 10 years, just like sleeping in
the same bed but dreaming different dreams.’
Besides, it is worthwhile to point out the syntactic constraint that prohibits
transitive verbs (cf. bùwénbúwèn with a [0213] stress pattern) from taking an
object.
28. a. *父親從來不聞不問孩子的功課。
*fùqin cónglái bù wén bú wèn háizi de go-ngkè.
father never not listen not ask children s’ homework.
‘The father has never paid any attention to his children’s homework.’
b. 父親對孩子的功課從來不聞不問。
fùqin duì háizi de go-ngkè cónglái bù wén bú wèn.
father to children’s homework never not listen not ask
‘The father has never paid any attention to his children’s homework.’
Compound prosodic word – Sizige 四字格
179
Instead of taking the object in the canonical (postverbal) position, a preposition dui must be used to introduce the object before the verb, as seen in
(28b). It is still a mystery why this is so even if a suggestion has been made
by Feng (2005) in terms of prosodic syntax (see also the prosody-syntax
interface).
From the above discussion on four-character patterns in Chinese, it is not
unreasonable to conclude that they are a unique category in the language in
the sense that they function as a prosody embodiment, an education reflector,
as well as a persuasive device – a highly cultivated linguistic form in Chinese.
As seen above, the FCP has become an educational symbol in verbal Chinese
for both native and non-native speakers as a result of the expressive powers in
different registers.
8. The four-character poem and the Four-Six Prose
When we talk about the FCP, it is worthwhile to mention that the tetrasyllabic prosody functions as a basic and most powerful device for poetic lines in
the Odes of Archaic Chinese. It is the [1+1]+[1+1] rhythm that works as the
basic prosodic pattern of Odes within a perfectly balanced poetic structure.
It is the most popular and powerful poetic pattern in Chinese literature, as
expected when disyllabic forms became more and more popular in ancient
China, and the poetry-making mechanism has motivated different theories of
poetic prosody5 to explain why there are four, then three and later five, sevensyllable poems in the history but rarely six-syllable lines in regularized poetic
formulations. To limit myself so as not to go beyond the scope of this chapter,
I will argue, following Feng (2010c) that the poetic system of Ancient Chinese
has made use of the disyllabic foot formation and evolved a poetic system conditioned on prosody the process of poetry-making, as illustrated in (29).
29. Condition on poetry-making in Chinese
One syllable cannot form a foot,one foot cannot form a poetic line and a poetic
line cannot form a poem (單音不成步、單步不成行、單行不成詩).
Thus,
a minimal foot = two syllables
a minimal line = two feet (or two prosodic units)
a minimal melodic unit = two lines (a couplet)
a minimal poem = two melodic units (a stanza, quatrain or 絕句)
Given the conditions on the system of Chinese poetic structure, the minimal,
and hence the basic or the optimal poetic foot is disyllabic (required by relative
prominence principle), and the minimal (hence basic and optimal) poetic line
is two feet, and further, the minimal (basic and optimal) couplet consists of two
lines, and again, the minimal and optimal stanza is formed by two couplets. In
other words, two minimal feet make one poetic line, two minimal lines = two
180 Compound prosodic word – Sizige 四字格
feet (or two prosodic units) make a minimal melodic unit = a couplet and two
couplets make a minimal poem = two melodic units (a stanza, quatrain or 絕句).
In musical terms,6 the [2x2] syllable structure is essential and it is minimally
and maximally realized within two lines as a melodic unit with each line consisting of two prosodic units. As a result, two syllables make up one poetic unit,
two poetic units make up one poetic line, two poetic lines form a melodic unit
(or couplet), two couplets form a minimal poem (stanza, quatrain, 絕句) and
finally two minimal poems make a compound or regularized poem (律詩).This
is the basic structure for the so-called new style poem (近體詩) in the later
development of Chinese poetry.7
Given the theory outlined above, most of the known facts about regulated poetry in the history of Chinese literature can be explained systematically. In addition, the so-called ‘parallel prose’ (pián wén 駢文) developed in
Chinese ancient literature can also be accounted for prosodically. This is so
because the parallel prose is derived from the main characteristic style of the
pián wén 駢文, and most importantly the grammar of parallel prose is essentially based on the duplication of a disyllabic unit to versify parallel phrases
between [σσ+σσ] or [σσ+σσ+σσ] antithesis, earning the traditional term
‘Four-Six Prose’ (i.e., prose made by four- or six-syllable sentence-pairs).
For example, (30):
30. mí yí píng yuán 彌迤平原 [σσ+σσ]
smooth and gently sloping, a level plain,
nán chí ca-ng wú zha-ng hǎi 南馳蒼梧張海 [σσ+σσ+σσ]
Southward galloping to the Grann Kola and the Swollen Sea
běi zǒu zǐsè yàn mén 北走紫色雁門 [σσ+σσ+σσ]
Northward racing to the Purple Pass and the Goose Gate. . . . . .
(From Wucheng Fu 蕪城賦 “Rhapsody on the city overgrown with weeds”
by Bao Zhao 鮑照 [416–450 AD])
A question regarding the Four-Six Prose is when did the four-character phrase
become what we have defined as the PrWd-Compound? There is not a clear
answer and this needs more investigation in the future, partially because we
don’t know the stress patterns of the four-character patterns such as mí yí píng
yuán彌迤平原 in Wucheng Fu. Though it is premature to give a conclusive
answer to the prosodic status of the four characters in early Medieval Chinese,
Liu Xie’s commentary on the poetic prosody of the four- and six-syllable group
in the rhapsodic line, as follows:
31. 四字密而不促，六字格（裕) 而非緩。. . . . . . 。
sì zì mì ér bù cù, liù zì gé (yù) ér fe-i huǎn . . .
four character tight and not rash, six character ample and not leisurely,
‘groups of four characters are tightly packed and not rushed, while groups
of six are ample and not leisurely.’
(Liu Xie, Literary Mind and the Carving of Dragons)
Compound prosodic word – Sizige 四字格
181
The word ‘dense’ (密) here is the key to characterizing four-syllable strings: the
elements of this four-character unit are tightly bound together. This implies
that the four-syllable unit has become a single prosodic template. It is the first
time in the history of Chinese that the tetrasyllabic string has been recognized
as a single unit with the prosodic property of being “dense but not hurried”.8
“Tightly packed” indicates how close various “units” are “packed together” in
distance. This idea can be formulated as follows:
32.
“tightly packed” = one category or one unit
{
}
[σσ σσ]
“not rushed” = “there is a little hiatus in between feet”
In other words, the term ‘tightly packed’ refers to an idea of being a single
‘unit’ and the term ‘not rushed’ means “there is a little hiatus in between feet”,
which is a perfect description of a two-feet unit. If we analyze the groups of
four syllables from this perspective, the “four syllables” in the rhapsody of Liu
Xie’s time are composed of tightly packed feet (of course there are still other
types in addition to variations of the latter type). The question that follows, of
course, is what qualities or properties do the two types of categories have? The
answer to this is that it is a prosodic unit (or category). Then what is it?
Recall that four-syllable units can be of different types. What type of category is a four-syllable “two-feet-in-one (= tightly packed and not rushed)”
unit? Liu Xie’s observation of “not rushed” gives a clear hint that this is the
compound prosodic word in prosodic morphology,9 namely:
33.
Compound PrWd
foot
σ
y
а
foot
σ
σ
jiàn shuāng
㇝
䴉
σ
diāo
䴅
one
arrow
two hawk
‘One arrow, two hawks.’
As see above, compound prosodic words are a type of compound word
which is composed of two standard foot (= prosodic word) groups. This is
exactly why they are “tightly packed and not rushed”. “Tightly packed” is the
result of a single unit, while “not rushed” is due to the fact that a prosodic pause
can still occur in between a [2+2] compound.
182 Compound prosodic word – Sizige 四字格
Given this analysis provided by Feng (2010c, 2016), it is highly likely that
the four-character patterns started to be versified first and then prosodicized as
Compound PrWds during the time around the North-South dynasties (420–
589 AD) in Chinese prosodic morphology.
Notes
1 Note that, according to Feng (2010a), the compound prosodic word has not been prosodicized or metricalized until the North-South dynasties (420～589). The reader is referred
to Feng’s 2010a for detailed arguments and evidence.
2 Note that difference between formal and elevated can be characterized as follows: unfamiliar words such as 止，介，鑒，維 are classical morphemes that are not intelligible in
Modern Chinese but they are necessary for elevated register, such as the Harvard motto
Veritas (‘truth’) in Latin, which is not accessible by ordinary people.
3 The FCEs behave differently among different dialects according to their different phonological and tonological systems. This book only focus on Mandarin dialect and leaves for
future studies the other possibilities of the FCEs in other dialects.
4 Hayes (1995:35) proposed a similar principle called Continuous Column Constraint.
5 There are insightful comments on the poetic effects of tetrasyllabic forms, such as “for
rhythms of a grand style, one must consider the four-syllable pattern to be the standard,
and other possibilities are not upright even if they can serve a variety of different styles 雅
音之韻，四言為正，其餘雖備曲折之體，而非音之正也”(摯虞《文章流別論》);
“the four-character line is orthodox, of which refining and smoothing are its essence
四言正體，雅潤為本”(劉勰 《文心雕龍》); and “in poetry, the four-character line is
graceful and gracious, the five-character line is upright and disdainful; the seven-word line
is relaxed and flowing; the three-character line is rigid and falling, the six-character line is
sweet and pleasant; mixed-length lines are like fragrant flowers. The four-character line is
the sound of Daya and thus the pneuma of poetry 詩四言優而婉，五言直而倨，七言
縱而暢，三言矯而掉，六言甘而媚，雜言芬葩，頓跌起伏。四言《大雅》之音也
，其詩中之元氣乎？” 陸時雍《詩鏡總論》
6 The melodic property of poetry is captured by Rhythmic Melody given below:
Rhythmic Melody
Rhythm in Chinese poetic system is licensed by rhyme, which is realized between at least
two lines, that is:
[LR1 #, LR2]where ‘L’ denotes ‘line’,‘R’ is ‘rhyme (at the end of a line)’ and ‘#’ is ‘break between
two lines.’Thus LR1 and LR2 form a minimal melodic unit, i.e., a couplet in poetic terms.
It implies that poetry makes use of the musical properties of language and the intrinsic
property of melody is repetition (thus, there are no one-line poems in classical literature of
Chinese as well as other languages too; see Hollander (2001:11–12), “the one-line poem is
almost always really a couplet, an epigram formed by the title and the line itself ”).
7 This system can be rendered in Chinese as 兩音＝詩步、兩步＝詩行、兩行＝詩聯、
兩聯＝絕句、兩絕＝律詩 (see Feng 2008).
8 Another important piece of evidence for the tetrasyllabic unit comes from the development
of four-syllable idioms. As calculated by Zhang T. (1999), among the 6,593 four-syllable
idioms collected in the Etymological Dictionary of Chinese Idioms《漢語成語考釋詞典》,
68.07% were produced after the Han 漢朝 and mostly in the Tang-Song唐宋and MingQing 明清 dynasties (59.33%), appearing especially in poetry and lyrical songs such as
“qingtian bairi (blue sky and white sun – transparent and good political system)” coined
from the poem “qı-ngtia-n báirì yìng lóutái 青天白日映樓臺” by the Tang poet Yu Han
韓愈.Though future research is needed, the above statistics indicate that the four-syllable
template may have been lexicalized as prosodic word compounding after the Wei-Jin
Compound prosodic word – Sizige 四字格
183
魏晉 Period. I would like to thank one of the anonymous reviewers for providing the
evidence.
Note that eight-syllable lines are too unstable for any footing possibilities, giving rise to
unfavorable usage in poetic lines in Chinese. Since only odd numbers of syllable lines can
avoid multiple footing possibilities while observing the Minimality Condition, only fiveand seven-syllable lines are favorable in Chinese poetry.
9 Shengli Feng (2011) and Marino Vigario (2003). The Prosodic Word in European Portuguese.
Berlin: Mouton de Gruyter, pp. 215–262.
References
Attridge, Derek. 1982. The Rhythms of English Poetry. London: Longman.
Chao,Yuen-Ren. 1968. A Grammar of Spoken Chinese. Berkeley: University of California Press.
Dell, Francois C. and Selkirk, Elisabeth. 1979. “On a Morphologically Governed Vowel
Alternation in French”. In Samuel Jay Keyser (eds.) Recent Transformational Studies in European Languages. Cambridge, MA: MIT Press.
Duanmu, San. 2000. The Phonology of Standard Chinese. Oxford: Oxford University Press.
Feng, Shengli. 1995. Prosodic Structure and Prosodically Constrained Syntax in Chinese. Ph.D. dissertation. Philadelphia: UPENN.
Feng, Shengli. 1997a. Prosodic Structure and Compound Words in Classical Chinese. In Jerome
Packard (eds.) New Approaches to Chinese Word Formation. Berlin: Mouton de Gruyter, 197–260.
Feng, Shengli. 1997b. Interactions Between Prosody, Morphology and Syntax in Chinese 漢語的韻
律詞法與句法. Beijing: Peking University Press.
Feng, Shengli. 1998. “On ‘natural feet’ in Chinese 論漢語的‘自然音步’ ”. Chinese Philology
中國語文 1.262: 40–47.
Feng, Shengli. 2002. “A Formal Analysis on the Origin of Verb-Resultative Constructions
in Chinese 漢語動補結構來源的句法分析”. Yuyanxue Luncong 語言學論叢 Essays on
Linguistics 26: 178–208.
Feng, Shengli. 2005. Prosodic Grammar in Chinese 漢語韻律語法研究. Beijing: Peking University Press.
Feng, Shengli. 2008. “On the parallel developments of trisyllabic foot structure and the QinHan poetry 論三音節音步的歷史來源與秦漢詩歌的同步發展”. Essays on Linguistics
語言學論叢 37: 18–54.
Feng, Shengli. 2010a. “A Theoretical Exploration of Prosodic Syntax”. Yuyanxue Luncong
語言學論叢 [Essays on Linguistics] 39: 204–244.
Feng, Shengli. 2010b. “On Register Mechanism and Its Grammatical Functions”
論語體的機制及其語法功能. Chinese Philology 《中國語文》5: 400–412.
Feng, Shengli. 2010c. “Some Fundamental Principles of Prosodic Stylistics 論韻律文體學的
基本原理”. Contemporary Rhetoric 當代修辭學 1: 25–36.
Feng, Shengli. 2011. “A Prosodic Explanation for Chinese Poetic Evolution”. Tsing Hua
Journal of Chinese Studies 41.2: 223–258.
Feng, Shengli. 2012.“Stylistic-Register Grammar: Correspondence Between Form and Function 語體語法:‘形式-功能對應律’的語言探索”. Contemporary Rhetoric 當代修辭學
6: 3–12.
Feng Shengli. 2016. Prosodic Morphology. In Wolfgang Behr, Gu Yueguo, Zev Handel, C.-T.
James Huang, and Rint Sybesma (eds.) Encyclopedia of Chinese Language and Linguistics.
Leiden: Brill.
Feng, Shengli. 2016. “Parallel Prose and Spatio-temporal Freedom--A Case Study of Creative Syntax in Wuchengfu 駢文韻律與超時空語法--以《蕪城賦》為例”. Lingnan Journal of Chinese Studies 嶺南學報 3: 189–219.
184 Compound prosodic word – Sizige 四字格
Hayes, Bruce. 1995. Metrical Stress Theory: Principles and Case Studies. Chicago: University of
Chicago Press.
Hoa, Monique. 1983. L’Accentuation en Pekinois. Ph.D. dissertation. University of Paris. Paris:
Centre de Recherches Linguistiques.
Hollander, John. 2001. Rhyme's Reason: A Guide to English Verse. New Haven:Yale University
Press.
Kenstowicz, Michael. 1993. “Evidence for Metrical Constituency”. In Kenneth Hale and
Samuel Jay Keyser (eds.) The View From Building 20 – Essays in Linguistics in Honor of Sylvain Bromberger. Cambridge, MA: MIT Press, 257–273.
Meredith, Scott Scott Emery. 1990. Issues in the Phonology of Prominence. Doctoral dissertation.
Cambridge, MA: Massachusetts Institute of Technology.
Perrine, Laurence. 1963. Sound and Sense, Poetry: An Introduction to Poetry. New York: Harcourt
Brace Jovanovich, Inc.
Shih, Chilin. 1986. The Prosodic Domain of Tone Sandhi in Chinese. Ph.D. dissertation. University of California, San Diego.
Tsou, Benjamin K. 2014. “What Is the Emblematic About Chinese QIE’s (Quadrasyllabic
Idiomatic Expressions) and Why? A Sociolinguistic Approach”. Linguistics Seminars at the
Chinese University of Hong Kong, February 25, 2014.
Vigário Marina. 2003 [2011]. The Prosodic Word in European Portuguese. Berlin; New York:
Mouton de Gruyter.
Yu, Min. 1989 [1981]. Collections ofYu Min’s Works in Linguistics 俞敏語言學論文集. Ha’erbin:
He-ilóngjia-ng People’s Publisher.
8Conclusion and final
remarks
In this book, we have discussed various aspects of prosodic morphology in
Chinese, from Archaic moraic foot formation to Medieval disyllabicity, from
prosodic word formation in Medieval Chinese to prosodic word compounding
in Modern Chinese, and finally arriving at prosodic register grammar.
The book starts from ancient prosody not only because the disyllabic foot
formation results from the change of moraic foot formation but also because
ancient scholars had noticed the linguistic phenomena of prosody long before.
To mention a few of the prominent ones, for example, Shen Yue (441–513 AD)
of the Medieval period, when analyzing the verses of his time, held that:
we should make characters of different tonal patterns alternate with each other and
ensure that the poetic couplet thus formed emerge in a relatively prominent manner.
If the first line of the couplet contains characters of the level tone, then the second
line must have characters of the non-level tone.The five characters of each line should
differentiate one from another in terms of low and high pitch levels and the two lines
must have completely different stress patterns.
(Biographies and Commentary of Xie Lingyun from Songshu)
Furthermore, in order to explain ga-oyáng zhı- pí 羔羊之皮 ‘lamb-sheep’s skin,
the skin of lambs’ in Shaonan of the Book of Songs, the philologist Kong Yingda
of the Tang dynasty (574–648 AD) proposed that “the text says lamb with sheep,
because lamb is also a kind of sheep, in order to balance the sentence, so the text
mentions both lamb and sheep”. Here, Shen Yue’s proposal of “level and nonlevel tones” and Kong Yingda’s analysis of xié jù 諧句 ‘to balance the sentence’,
zú jù 足句 ‘to fulfill the sentence’, yuán wén 圓文 ‘to round off the sentence’ all
constitute the well-known metrical study in the history of Chinese literature.
In the past several decades, a large number of Chinese prosodic morphological phenomena have been explored and many new discoveries that merit
attention from general linguistics are emerging. Prosodic morphology is everchanging and rapidly developing to be a new research field. Judging from its
development trends over the past decades, it can be predicted that this field will
develop even faster in the future.
186 Conclusion and final remarks
The significance of study on prosodic morphology has caught attention not
only in Modern Chinese linguistics but also in Chinese historical phonology,
not only in Chinese but also in general linguistics.
Wang HJ (2008) commented on the prosodic register grammar (written
Chinese) by taking it as “a significant” concept. She says, “ ‘Modern Chinese
written language’ . . . is a very important concept” and “(it) fills an important
gap in second language teaching” with “significant viewpoints such as ‘Modern Chinese written language has independent grammar’ ” and “the property
of Modern Chinese written language is the demand on disyllabicity”. Finally,
she concludes, “These provide us with a new, promising research area”1 (Wang,
2008:309–313).
In the study of Chinese historical phonology, Ho, Dah-an, the former Director of the Institute of History and Philology, Academia Sinica in Taiwan, considers, in his recent article (2016:221), that the moraic foot formation and their
prosodic weight we discussed in Chapter 2 “provide new insights from different
perspectives . . . from the lightness and heaviness of moras”, which “differ from
what The Re-Construction of Archaic Chinese Pronunciation claims” in the study of
Archaic Chinese phonology.2 Zhengzhang (2017) confirmed the moraic analysis of Archaic Chinese phonology in his current work:
Prosody has played an important role in Archaic Chinese, and at that time, monosyllabic moraic foot functioned actively.The presence and absence of moras among different types of syllables cause the contrast between short and long syllables (e.g., the
contrast between weak and strong pronouns such as吾ŋa vs.我ŋai) and the alternation between weak and strong interrogative particles such as 邪laa vs. 也laal’’ >
laiʔ), etc.). Moraic prosody has played an important role in Archaic Chinese grammar and sound evolution.
It is worthwhile to note from the chapters presented here that prosodic morphology in Chinese has some interesting features in the sense that western
theories and ancient thoughts go hand in hand. For example, the theory of
Relative Prominence proposed by Mark Liberman (1975) goes perfectly with
Shen Yue’s statements: “we should make characters of different tonal patterns
alternate with each other and ensure that the poetic couplet thus formed
emerge in a relatively prominent manner” and “the five characters of each line
should differentiate one from another in terms of their low and high pitch levels, and the two lines must have completely different stress patterns”. Another
example comes from the facts of prosodic morphology. While McCarthy and
Prince (1986, 1995) invented the theory of prosodic morphology, we found
Lu Zongda and Yu Min’s (1954) study on prosodically differentiated part of
speech in Mandarin Chinese: “kaikai (to bubble up, the first syllable kai is
stressed within the verb reduplication) vs. kaikai (boiling water, the second
syllable is stressed within the adjective reduplication)”3 and based on both we
come up with a new idea of the morphological prosody in Chinese (Wang L.J
2009, 2014). The modern linguistic theories and traditional Chinese studies on
prosody complement each other.
Conclusion and final remarks
187
Prosodic morphology is a subpart of Prosodic Grammar.The work presented
here has been recognized in the field such that prosodic grammar constitutes
one of the important areas of Chinese formal linguistics.4 The fundamental principle of formal linguistics lies in rationalism, while prosodic grammar
not only roots itself in rationalism but also in the reason-based approaches
(i.e., Lǐ bì理必 ‘logical certainty’) held by the scholars of the Qian-Jia Era of
the Qing dynasty5 (which gives the top priorities to logical certainties rather
than materials collected). As a result, prosodic morphology in Chinese, as seen
in previous chapters, is neither a sinicization of western linguistic theories
nor a westernization of Chinese linguistic theories. Its goal is to discover the
enlightening ideas of Chinese traditional thoughts and to reveal the profoundness of western theories with the essence of Chinese traditional insights. In
a word, it is under such guidance and direction that we have made our discoveries and accounted for the facts. The emergence of prosodic grammar is
the product of the combination of the eastern and western academics, and
the future research of this field will move forward unswervingly along this
road. Therefore, the objects of our attention and research include not only
those in general linguistics, such as: What is prosody? What is rhythm? What
is a foot? What is a prosodic word? What is the difference among CP-/TP-/
VP-Prosody and how do they interact? Why does the tone of the language
correspond to its (sentence final) particle? etc. Of more interest to us are the
prosodic matters of the Chinese language, such as: Why did the moraic foot
in ancient times gradually disappear? Why were there no tone-3 (rising) and
tone-4 (falling) together with the lack of sentence final particles in Archaic
(or proto-) Chinese? Why do tone development and the disyllabification walk
side by side? Why did the disyllabification suddenly rise in the Western Han
and Eastern Han dynasties? Why did the distinction between verse and prose
occur in the Medieval Chinese? Why did the ‘Four-Six Parallel Prose’ start
in the Wei-Jin Periods? Since poems are to be recited and proses chanted,
what are the prosodic principles of recitation and chant? Why should we ‘hum’
Ancient Chinese prose but not ‘read’ it?6 Why is it doubtful that the “3+3+4”
rhythm in “The vehicles were rumbling, the war horses were neighing, and
the soldiers have put their bows and arrows on the waist” (a verse line in Du
Fu’s poem Chariots March 兵车行) is a loan rhythm in terms of its prosody?
How can it be that the “stressed-unstressed-unstressed” pattern of wo men de
‘ours’ in Beijing dialect is taken from the Mongolian language? Moreover, why
are there stylistic differences between stressed syllables and unstressed syllables
from the perspective of register (stylistic) grammar? Why is there a prosodic
stylistic correspondence of syllabic length and stylistic elevation, as exemplified
by the monosyllable-disyllable contrast in Chinese and the contrast between
words of Latin origin and words of Anglo-Saxon origin in English? And a
more challenging question was raised by Hu Qiaomu in his letter to Yuan
Ren, Chao on June 12, 1981:
How did Chinese poetry change from even-numbered characters (syllables) in each
line, such as in the Book of Songs and the Poetry of Chu, into odd-numbered
188 Conclusion and final remarks
characters (syllables) after the Western Han and Eastern Han Dynasties? As for
poems of even-numbered characters in each line, except for the prose poems, the classical poems with six characters in each line were always not popular, and there were
no poems with eight characters in each line at all (excluding modern new poetry). As
for the poetic forms with odd-numbered characters in each line, there were basically
limited to poems with five or seven characters in each line (excluding the lyrics of
Ci and Qu), and most of the folk songs have seven characters in each line. After the
emergence of the modern new poetry, the situation changed again. Poems of evennumbered characters in each line are in the majority, and most lines have eight or
more characters (free verses are not taken into consideration).7
He wondered, “whether Ancient Chinese during this period has gone through
any crucial changes”.Today we know that Ancient Chinese was a synthetic language before the Western Han and Eastern Han dynasties. Then it evolved into
an analytical language (Huang 2015, Xu 2006, Feng 2014 and references cited
there). Obviously, the typological change has opened a new perspective for
future researches to look into it deeper and wider than before8 and to explain
the change not only in the form of literature, but also motivate linguistic explorations for prosodic morphology and prosodic syntax, thereby providing linguistic bases for literary forms and genres.
Given prosodically based study of lexicon, morphology, syntax, literature and
the way of thinking9 in both Ancient and Modern Chinese, it is not exaggerated to say that to understand Chinese is essentially to comprehend its prosody.
In other words, the understanding of Chinese grammar is based on the understanding of the grammatical functions of monosyllabicity all the way up to
prosodic morphology and prosodic syntax. This book, of course, only offers
one aspect of the whole picture of prosodic grammar in Chinese. However,
prosodic morphology is the foundation of all related areas among prosodically formed modules of grammar, such as prosodic syntax, prosodic register
and prosodic literature, all important and interesting areas for future research.
Prosodic morphology has now developed to be a crucial part of Chinese linguistics, and has made considerable contributions not only to Chinese linguistics but also to linguistic theories in general (see among others, Liu 2007, Cao
2010, Deng 2010, Hong 2015, Huang M. 2015, Wang LJ 2015, Wang YN 2015,
Zhuang 2015, Zhu SP 2015, Li G 2015, Wang C. 2017, Wang CS 2017). In his
review on the study of Chinese prosody and syntax (2014), Andrew Simpson
asserts that
the continued study of prosody and syntax interactions, whether as a study synchronically across different varieties of Chinese or (quite possibly) as a study diachronically, promises to be a rich and very informative area of future research for Chinese
linguistics, and in ways that Chinese can also make important contribution to the
general theories of human language.
(Simpson 2014:489)
Conclusion and final remarks
189
This is what this book has intended to do and once more, quoting from the
Analects of Confucius: fe-i yue- néng zhı-, yuàn xué ya-n 非曰能之，願學焉 “I am
not saying that I am capable of doing it, but I am learning toward it.”
Notes
1 The original texts read: “馮勝利（2003a, 2004b 提出了一個重要的概念 ——
現代漢語書面語”、“ ‘現代漢語書面語’是一個十分重要的概念”、“馮對於‘現代漢
語書面語’的研究，填補了二語教學的一個重要的空白”、“他不僅提出了‘現代漢
語書面語有獨立的語法’、‘現漢書面語的特點是韻律成雙的要求 ’等重要理論觀點
，而且提出應該編寫專門供二語學習者使用的漢語課本，該課本應該設立大量漢
語字詞‘由單組雙、拆雙為單’的練習。這就開闢出了一片新的、大有可為的研究
領域。” —— 王洪君《漢語非線性音系學》(Feng, Shengli: “Feng (2004b)” cited in
Wang (2008:309)'s original text should be corrected as “Feng (2003b).”)
2 “Sun (2014) and Feng (2012, 2013) provide new insights from different perspectives: the former from co-innovations shared by Chinese and Tibetan, and the latter
from the lightness and heaviness of moras. Their comments differ from what The ReConstruction of Archaic Chinese Pronunciation [上古音新構擬] claims, therefore I suggest
reader to read and compare between these different perspectives, then one should
know what to believe. ” (Da’an He (2016) “Should not make this kind of mistake:
Comment on Baxter-Sagart’s Archaic Chinese reconstruction.” Journal of Chinese Linguistics 44: 221).
3 See Modern Chinese Grammar by Lu Zongda,Yu Min. Commercial Press 2016 [1954].
4 See Studies on Contemporary Chinese Linguistics edited by Li Yuming. 2016:199.
5 The “Necessity of reason” is described in the inference of “Guang (heng) pi si
biao光(横)被四表” made by Dai Zhen (1724–1777) in “The Classics of Emperor Yao”. For
more details, see “The ‘Necessity of Reason’ Raised by the Scholars During the Reigns of
Emperor Qianlong and Emperor Jiaqing of the Qing Dynasty and the Scientific Nature of
Language Research” by Feng Shengli. Chinese Frontier of Language and Literature, 2015(9):
99–117.
6 See Zhu Guangqian (2013) “The Sound and Rhythm of Prose 散文的聲音節奏” In:
Chatting Literature《談文學》. Peking University Press.
7 See A Collection of Linguistics Essays by Wu Zongji, The Commercial Press: p. 246.
8 Perry Link (2013/2007) writes: “Is there any way to tell whether wǔyán 五言 ‘Five-character (syllable) verse’ and qı-yán 七言 ‘Seven-character (syllable) verse’ are ‘default’ preferences? In an article called ‘On the Natural Foot in Chinese,’ Feng Shengli has come up with
an ingenious method for doing exactly that.” (An Anatomy of Chinese: Rhythm, Metaphor,
Politics. 2013:51/53/58–58/60–61/67)
9 Y. R. Chao (1976) writes in Aspect of Chinese Sociolinguistics. Stanford: Stanford University
Press:271): “I venture to think that if the Chinese language had words of such incommensurable rhythm as nán-nǚ 男女 ‘male and female’, tia-ndì 天地 ‘heaven and earth’, zhì-miù
智繆, there would never be such far-reaching conceptions as Yı-n-Yáng 陰陽 ‘yin and
yang’, and Qián-Ku-n 乾坤 ‘heaven and earth, the universe’.”
References
Cao, Wen. 2010. The Prosodic Realization of Focus Stress in Chinese. 漢語焦點重音的韻律
實現. Beijing: Beijing Language and Culture University.
Deng, Dan. 2010. Study on Prosodic Word in Chinese 漢語韻律詞研究. Beijing: Peking University Press.
190 Conclusion and final remarks
Feng, Shengli. 2003a. “The Prosodically Restrained Written Language and the Audiolingua Approach 韻律制約的書面語與聽說為主的教學法”. Chinese Teaching in the
World 世界漢語教學 1: 87–97.
Feng, Shengli. 2012. “Evidence of Monosyllabic Feet in Old Chinese: New Ways of Studying Old Pronuncation from the Perspective of Prosody”. Studies in Historical Linguistics
歷史語言學研究 5: 78–90.
Feng, Shengli. 2013. New Perspective of Research on Old Chinese Phonology. In Shi, Feng
and Peng, Gang (eds.) Eastward flows the Great River: Festschrift in Honor of Prof.William S-Y.
Wang on his 80th Birthday. Hong Kong: City University of Hong Kong Press, 71–84.
Feng, Shengli. 2014. “Historical Syntax of Chinese”. In Huang C.-T. James, Li Y.-H. Audrey
and Andrew Simpson (eds.) The Handbook of Chinese Linguistics. Chichester, UK: Wiley
Blackwell, 537–598.
Ho, Dah-an. 2016. “Such errors could have been avoided: Review of Old Chinese: A New
Reconstruction”. Journal of Chinese Linguistics 44.1: 175–230.
Hong, Shuang. 2015. Minimal Word in Chinese 漢語的最小詞. Beijing: Beijing Language and
Culture University Press.
Huang, C-T. James. 2015. “On Syntactic Analyticity and Parametrical Theory”. In Li Audrey,
Andrew Simpson and Dylan Tsai Wei-Tien (eds.) Chinese Syntax in a Cross-Linguistic Perspective. New York: Oxford University Press, 1–50.
Huang, Mei. 2015. Monosyllabic Word Used in Disyllabic Template in Chinese. 漢語嵌偶單音詞.
Beijing: Beijing Language and Culture University Press.
Li, Guo. 2015. Prosodically Constrained Wh-Question and Their Syntax in Archaic Chinese
韻律制約的上古漢語疑問詞語及句法. PhD Dissertation. Hong Kong: The Chinese
University of Hong Kong.
Liberman, Mark. 1975. The Intonation System of English. Ph. D. dissertation, MIT, Cambridge,
MA.
Link, Perry. 2013. An Anatomy of Chinese: Rhythm, Metaphor, Politics. Cambridge, MA: Harvard
University Press.
Liu, Xinqiang. 2007. Study on Modern Chinese Rhythm 現代漢語節奏研究. Beijing: Beijing
Language and Culture University Press.
Lu Zongda and Yu Min. 1954/2016. Modern Chinese Grammar 現代漢語語法. Beijing:
Zhonghua Book Company 2016.
McCarthy, John and Prince, Alan. 1986. Prosodic Morphology. MS, University of Massachusetts
and Brandeis University.
Simpson, Andrew. 2014. “The Phonology and Syntax Interface”. In Huang C.-T. James,
Audrey Li and Andrew Simpson (eds.) The Handbook of Chinese Linguistics. Chichester,
UK: Wiley Blackwell, 465–492.
Shen,Yue. 487/1987. The History of Song. 宋書. Shanghai: Shanghai Guji Press.
Sun, Hongkai. 2014. “Co-Innovation of Sino-Tibetan Languages 漢藏語系語言的共同
創新.” Minzu Yuwen 2: 3–13.
Wang, Changsong. 2017. “A Study on Noncanonical wh Particles From the Syntax-Prosody
Interface – a Case Study of shenme in ‘V shenme (V)/(NP)’ Construction”. Studies in
Prosodic Grammar 2.1: 73–100.
Wang, Chi. 2017. “Prosodic Constraints on the Predicate of Causative Constructions in
Mandarin Chinese”. Studies in Prosodic Grammar 2.1: 101–135.
Wang, Hongjun. 2008. Non-linear Phonology of Chinese (Enlarged Edition) 漢語非線性音系學
(增訂本). Beijing: Peking University Press.
Wang, Lijuan. 2009. Study on the Morphological Function of Disyllable in Mandarin Nouns and
Verbs 從名詞、動詞看現代漢語普通話雙音節的形態功能. Ph.D. dissertation. Beijing:
Beijing Language and Cultural University.
Conclusion and final remarks
191
Wang, Lijuan. 2015. Morphological Prosody in Chinese 漢語的韻律形態. Beijing: Beijing Language and Culture University Press.
Wang,Yongna. 2015. Disyllabic Words Used in Quadrisyllabic Templates 漢語合偶雙音詞. Beijing: Beijing Language and Culture University Press.
Xu, Dan 徐丹. 2006. Typological Change in Chinese Syntax. Oxford; New York: Oxford University Press.
Zhengzhang, Shangfang. 2017. “Prosody in Chinese Dialects and Old Chinese”. Studies in
Prosodic Grammar 2.1: 1–2.
Zhu, Guangqian. 2013. Chatting about Literature 談文學. Beijing: Peking University Press.
Zhu, Saiping. 2015. Four Character Expressions in Chinese 漢語的四字格. Beijing: Beijing
Language and Culture University Press.
Zhuang, Huibin. 2015. Syntactic Words in Chinese 漢語的句法詞. Beijing: Beijing Language
and Culture University Press.
Index
Page numbers in italic indicate a figure on the corresponding page.
Adjectival Ordering Restriction (AOR)
121, 123, 127, 129
adjusting distances 137, 140
affixation 1, 10, 26 – 27, 32, 91 – 92, 108
amalgamation principle 137, 145,
148 – 150, 159
anaphoric rules 124 – 125, 127
Anglo-Saxon 139, 187
Archaic (or Proto-) Chinese 27, 43, 45
Archaic Chinese 14, 38 – 39, 44 – 45, 47,
67n19 – 20, 171, 179, 185 – 187, 189n2
augmentation 81, 176
bimoraic 41, 49, 71, 116
bimoraic feet 14 – 15, 42 – 43, 49, 52, 57,
61n20, 131
bimoraic syllable 49, 61n20, 116, 131
branching 39 – 40, 42, 44, 46, 56
casual 86, 129 – 130, 136, 141, 150, 177
categorical 107, 117 – 118, 139
Chinese phonology 3, 164
Chinese Poem 1, 179 – 180, 182n6, 182n8,
187 – 188
Classical Chinese 3, 6, 14 – 20, 24 – 25,
32, 44 – 45, 48 – 53, 55 – 59, 59n1, 59n3,
62n22, 63n26, 63n30, 68, 74 – 75, 77, 117,
142 – 145, 147 – 148, 152, 159, 160n4,
182n6
cliticize 47, 62n22
coda 28, 38, 40, 42, 60n10, 60n12, 61n17,
62n22, 87n9, 87n13
colloquial expression 75 – 76, 137, 139 – 140,
141, 143, 149, 155, 159, 169, 177
communicative distances in face-to-face
situations 137 – 138
complement 24, 57, 59n6, 62n23, 63n28,
92 – 93, 96, 109, 152, 186
compound 4, 6 – 7, 9, 10 – 11, 14 – 19, 20 – 26,
28 – 29, 30 – 37, 48, 50 – 59, 59n2, 59n4,
59n6, 59n8, 60n15, 61n21, 63n27,
63n29 – 30, 68, 73, 74 – 77, 81, 84 – 86,
85n11, 88n13, 90, 92 – 99, 100 – 111,
115 – 118, 120 – 129, 132 – 133, 158 – 159,
164, 168, 171 – 173, 180 – 181
compounding 185
compound PrWd 4, 50 – 52, 85 – 86, 94 – 99,
110 – 111, 122, 133, 159, 168, 168 – 173,
175, 178, 180, 181, 182
consonant cluster simplification
26 – 28
constituent 2, 39, 44 – 45, 47, 59n3, 63n27,
71 – 72, 170, 173
contrasting concepts 100
coordinating compound 29, 54, 101 – 103
coordinating structures 25, 52 – 56,
102 – 103, 138, 170
Coordination Construction (CC) 92
covert light verb 129
dative construction 139
degree of colloquialism 137, 140 – 159
degree of formality 137, 140, 159
degree of informality 140
de-phrasalization 56 – 67
diglossia 141 – 143, 159
distance 137 – 137, 140 – 141, 143, 146,
160n6
distance grammar 語距語法 140
distance-word 語距詞 140
disyllabic compound 14, 22, 32, 51, 55, 68,
117 – 118
disyllabic foot 5 – 6, 14 – 15, 38 – 39, 42 – 44,
46 – 49, 51 – 52, 55 – 58, 61n20, 62n22,
62n25, 68, 78, 92, 97 – 99, 116, 165,
179, 185
Index
disyllabicity 14, 33, 35 – 36, 48, 53, 55 – 56,
58, 62n22, 68 – 69, 70, 76 – 77, 79, 81, 83,
132, 165, 185
disyllabic word 1, 3, 6, 10, 32, 35 – 36, 69,
74, 76 – 77, 79, 88n15, 98 – 99, 136, 153,
155 – 156, 160
edge alignment (ALIGN) 2, 71, 87n7, 87n12
elevated 12, 76, 87, 136 – 141, 146, 150 – 152,
160n1, 161n14 – 15, 164 – 167, 177 – 178
emphatic forms 43
epenthesis 75, 87n12, 91, 171
epistemic stance 140 – 141
exclamation 47, 91
extrametricality 108
focus 47
Focus Prosody Correspondence
Principle 43
foot 2, 4 – 9, 10 – 15, 20, 35, 38 – 39, 40 – 49,
50 – 52, 55 – 58, 61n20, 61n21, 62n22,
62n25, 63n27, 63n30, 68 – 69, 70 – 71, 75,
77 – 79, 83, 85, 87, 92, 95 – 99, 100 – 111,
111n1, 115 – 116, 136, 154, 158, 165 – 170,
172 – 175, 177 – 182, 183n8, 185 – 187
Foot Assignment Principle (FAP) 95,
97 – 98, 101 – 103, 106
Foot Binarity 14, 41, 49, 57, 115 – 116
foot-directionality 5, 7, 10, 95, 99, 100, 106
Foot Formation Rule 5, 10, 14, 15, 20,
42, 48, 50 – 52, 57, 58, 61n21, 62n22,
63n27, 68 – 69, 75, 77, 95, 110 – 111, 167,
185 – 186
Foot Formation Rule in Classical Chinse 48
formal 10, 86, 136 – 141, 143, 145 – 146,
150 – 152, 160n1, 161n8, 164, 166,
177 – 178, 182n2
formal chinese 142, 144 – 145, 148 – 149,
153 – 155, 158
formal Chinese writing system 142
formal expression 144, 147 – 149, 151, 152,
154, 160n7
Form-Function Correspondence Principle
(FFCP) 137 – 138
four-character pattern 164, 166 – 167,
169 – 173, 175, 177 – 180, 182
four character poem 179
Four-Six Prose 179 – 180, 187
four syllable idioms 169, 182n8
full tone syllable 84, 90, 110, 131
genre 141, 188
germination 175
Government-based NSR 109
193
head-to-head adjunction 126
heavy 38 – 39, 42, 45, 47
heavy syllable 39, 41, 43
Hé-ŏu Cí 合偶詞 11, 136 – 137, 141, 146,
152, 154, 159
high-respectful/dominating 140
human register competence 141
Idiomatized Prosodic Word 20, 49 – 52,
57 – 59
infixation 133
informal 76, 137 – 141, 143, 149, 151,
161n15, 164, 166, 177 – 178
informal chinese 144
interpersonal stance 140
isomorphism 50
Jù-xíng 正式體句型 136
Juxtaposition 37
Latinate Origin 139
leftward-footing 100 – 102
lexemes 127
lexical compound 122, 126 – 129, 132 – 133
Lexical Integrity (Hypothesis) 17, 59n3, 98,
123 – 124, 128
lexicalization 36, 51 – 52, 57 – 58, 82,
103 – 104
light syllable 40, 43
Limerick Prosody 177
lingua franca 32
literary Chinese (文言文) 142 – 143
Literary Revolution (文學革命) 142
loan words 3, 7, 70, 74 – 77, 111
logographic 87n4
low=modest/honorific 140
maximality condition 131 – 132
May-Fourth Movement 141 – 145
Medieval Chinese 15, 20, 25 – 28, 37 – 38, 40,
42 – 43, 57, 59n9, 60n10, 60n11, 61n21,
62n22, 76, 82 – 83, 91, 171, 180, 185, 187
merge 26, 28, 50
metrical theory 38, 40, 44
minimality condition 116, 131, 183n8
minimality constraint 133
Minimal PrWd (MinWd) 115 – 116,
133, 168
Minimal Word condition 122
minimal word effect 10, 115 – 118, 129 – 130
Minimal Word Theorem 132
Modern Chinese 1 – 2, 14 – 15, 26, 59, 59n3,
60n15, 61n20, 61n21, 63n29, 69, 76, 78,
85, 117, 138 – 139, 141 – 146, 148, 150,
194 Index
155, 157 – 158, 160n1, 160n4, 182n2,
185 – 186, 188
Modern Mandarin 37, 59n3, 61n21, 171
modern written chinese (白話文) 142 – 143
modification-head structure 122 – 124
modifier head compound 29, 54, 56, 58,
59n6, 98, 105
monomoraic syllable 131
monosyllabic 1 – 3, 5 – 7, 15, 23, 35, 39, 42,
46, 48 – 49, 51 – 52, 57, 61n20, 63n25, 68,
70, 72, 75, 77, 93, 98, 109, 130 – 131, 136,
138, 147, 150, 158, 177
monosyllabicity 48, 68, 70, 81, 188
monosyllabic myth 12, 68 – 69, 84, 86
monosyllabic word 10, 21 – 23, 29, 33,
44 – 45, 47, 55 – 56, 61n21, 76 – 80, 95, 102,
136, 152, 155 – 157, 159, 172
mora 2, 4, 38, 41, 43, 44, 49, 57, 61n21, 71,
115 – 116, 131, 134n5, 186
moraic foot 38 – 39, 43 – 44, 57,
185 – 187, 189
morpheme 2 – 7, 15, 68, 70 – 74, 75 – 77,
79 – 82, 84 – 87, 87n8, 87n10, 87n12,
87n15, 91 – 92, 95, 107 – 108, 111, 133n2,
136, 143, 155 – 156, 158 – 159, 172, 182n2
morpheme edge (MEdge) 71 – 72
morphemization 73 – 74, 77
Morpholization 83
morphological category (Mcat) 71
morphological prosody 10, 186
morphology 1, 3, 6 – 7, 10 – 11, 20, 25, 32,
48, 68 – 71, 73 – 75, 79, 81, 84 – 86, 92,
99, 100 – 101, 103 – 104, 106, 110 – 111,
111n1, 115 – 116, 131 – 132, 164, 167, 188
morphosyllabic 81, 87n4, 92, 164
Morphosyllabic Constraint (MC) 2 – 3, 68,
74 – 75
Morphosyllabicity Rule (MR) 39, 52,
68 – 69, 70 – 71, 74
near-intimate, far-distant/formal 140
neutralization 18, 110
neutral-tone 15
neutral-tone syllable 84
node 39 – 41, 46, 56, 62, 70, 72, 95, 97,
127, 129
Non-Coincidental-Distinction
Approach 143
non-elevated 140
none-to-two (annotation) 21 – 22
Non-head Stress Rule 62n23
nuclear stress (NS for short) 56 – 57, 83
Nuclear Stress Rule 45, 118
Old Chinese 14, 15, 20, 25 – 33, 35, 37 – 43,
45, 48, 51, 57, 59n9, 60n9, 60n11, 61n18,
62n22
one­to-two (annotation) 21, 23
Onomatopoeia 92
onset 40 – 42, 61n18, 72
open syllable 27, 37, 60n9
ordinary monosyllabic words 136
ordinary speech(zhìyán 質言) 103, 136,
145, 152
parallel prose 180, 187particle 16, 39, 75,
87n9, 130 – 131, 172, 186 – 187
phonological attrition 26, 32
phonological word 79, 103
phrasalization 91
Phrase Structure Constraint 59n3
phrase-structure rules 49, 52, 63n26
piānyì fùcí 偏義復詞 18, 30, 31, 58
place holders (PH for short) 172
P-Morphology 95, 99, 100
polysyllabic Words 3, 4, 32, 73, 75
postcoda 28, 38, 60n12
prefixations 110
primary stress 45 – 46, 56, 173 – 174
Principle of Auditory Intelligibility(PAI for
short) 137, 147, 159
Principle of Prosodic Register 154
Principle of Shaping by Prosody (PSP for
short) 137, 147, 159
prosodic constraints 100, 103, 145
prosodic domain 46, 132
prosodic grammar (PG for short) 99, 109,
136, 150, 154, 159 – 160, 187 – 188
Prosodic Hierarchy 4, 6, 14, 49, 50, 57,
115, 116
prosodic literature 188
prosodic minimality 79, 81
prosodic morphology 1, 2, 4, 6, 7, 10 – 11,
14 – 15, 25, 48, 50, 69 – 70, 77, 79, 94, 96,
98 – 100, 103, 106, 111 – 116, 133, 150,
155, 159, 165, 167, 181 – 182, 185 – 188
prosodic phonology 6, 39, 96, 100, 111, 164
prosodic register 188
prosodic register grammar (PRG for short)
142, 150, 154, 160, 185 – 186
prosodic-syntactic phenomena 132
prosodic Syntax 178 – 179, 188
prosodic template 51, 155, 181
prosodic Unit 1, 7, 35, 38, 44, 46, 52, 136,
171 – 172, 175, 180 – 181
prosodic weight 42 – 43, 61n19, 108 – 109,
131, 186
Index
prosodic Word (PrWd for short) 4, 6 – 7, 11,
14, 20, 48 – 52, 63n25, 63n27, 68, 81, 86,
90, 92, 94, 106, 110, 115 – 118, 154, 157,
166, 168, 170 – 174, 181 – 183, 185, 187
prosodic Word Formation 7, 68, 90, 92, 106,
167 – 168, 185
prosodic word requirement 14, 50
prosody-syntax interface 179
Proto-Chinese 27
PrWd-compound 168 – 169, 171, 177 – 178
PrWd-phrase 168 – 169, 171, 177 – 178
Qiàn-ŏu Cí 嵌偶詞 136 – 137, 141, 146,
150 – 151, 154 – 157, 159
quadrisyllabic 133n2, 165
quasi-prefixes 107
quatrain 179 – 180
rationalism 187
reanalyzed (re-categorized) 87n10, 107,
133n4, 139, 142, 144
redundancy 84, 86
reduplication 6, 19, 90, 92, 99, 129,
131 – 133, 159, 161, 186
register: Elevated Registers 161n14,
164; register-forming capacity 150;
register syntax 136; register system 86,
136 – 137, 143
Register Coherence Principle 137, 154
register grammar(RG for short) 10,
136 – 139, 141 – 143, 145, 150, 156,
159 – 160
register property 130
regularized poem 180
relative prominence 176, 186
relative prominence principle 4, 44, 54, 56,
133n1, 179
Relative Prominence Rule 39
resyllabification 3, 72, 87n9, 87n12
rhyme 20, 27, 39, 40 – 42, 61n18, 61n21,
171, 182n6
Rhythmic Melody 182n6
rhythmic pattern 9, 60n16, 104, 131,
164, 167
right-strong 131, 175
rightward-footing 99 – 100
root morphemes 3, 70 – 71, 75, 77, 87n9,
87n11, 87n12, 107, 172
rule out 47, 139
semantic disambiguation 28 – 29
semantic prefixes 107
Sentential Prosodic Rule 45
195
sister node 46
Sìzì Gé 11, 94, 164 – 165, 167, 170, 175
sonorant 43
spoken language 139, 143
stance 140 – 141
Stance Theory 140 – 141
standard foot 6, 44, 47, 48, 51 – 52, 61n21,
63n25, 95 – 96, 102, 158, 181
standard prosodic word 63n25, 170
stanza 179 – 180
Stress Adjustment Principle 175
stressed 43, 45, 85, 186 – 187
stress pattern 164, 168 – 169, 171, 174 – 178,
180, 185 – 186
stronger 4, 46, 55 – 56, 175
style 138 – 143, 146, 148, 149 – 151, 160,
177 – 178, 180, 182n5
stylistic-register 86, 143, 177 – 178
stylistic-register effects 143
stylogrammar 139 – 141
stylo-word 140
subordinating structures 53 – 56
super-foot 96 – 98, 102, 131
super-foot formation 102, 131
super-heavy syllables 40, 42
super-PrWd 96
SVO language 44 – 45, 62
swift jumping rhythm 177
syllable structure 1, 15, 27, 28, 35, 37 – 42,
57 – 59, 60n9, 61n20, 90, 98, 180
syllable structure simplification 37, 39
syntactic atom (X0) 127, 129
syntactic compound 19, 30, 82, 84, 87n11,
126 – 129, 132
system of distance 141
Templatic Constraint 117, 133
tetrasyllabic prosody 179
thou lowest 139
three-way stylistic registers 140 – 141
tone: departing tone 26 – 28, 60n13, 62n22;
Entering Tone 60n13; four-tone system
38, 58, 61n20; full-tone syllable 84, 90,
131; high-rising tone 28, 62n22; level
tone 27, 60n13, 134n5, 185; Rising Tone
27; tone-bearing syllables 38, 57
Tone Sandhi 61, 133 – 134n5
Tone Sandhi domain 61
triglossia 140
trisyllabic: trisyllabic compounds 7, 94,
98, 101 – 102, 133n2; trisyllabic foot
62n25, 97, 116; trisyllabic super-foot 97;
trisyllabic template 167
196 Index
trisyllabic maximality constraints 132
truncation 91
two-rule system of Chinese morphology
69, 81, 84
two-to-two (annotation) 22
unstressed 43, 187
verb reduplication 129 – 133, 186
verb-resultative complement 57, 59n6,
92 – 93, 109
vernacular 20 – 21, 47, 141 – 144, 148
vernacular vocabulary 143
very elevated 140
vowel 3, 27, 37, 43, 60n12, 61n17, 61n21,
72, 108
Warring States Period 23, 25, 33, 36,
44, 59n1
weak: weak and strong 39, 186; weakened
38, 43, 130, 131; weakening 57; weaker
4, 5, 43, 175 – 176; weak form 134; weak
node 40, 46, 56; weak syllables 39
weight-sensitive language 44, 61n19
wh-expression 44 – 46
word formation 1, 6 – 7, 10, 48, 68, 74,
90, 92, 95, 106, 115, 133, 167 – 168,
185
word formation of compounding 133
Word Formation Rule 15, 20, 50, 57,
63n27
write what you say (寫口) 142
written language (wényán 文言) 136, 139,
143, 152, 186
Xiéshēng 諧聲 26
Yuti Grammar (YG for short) 137