Prosodically Constrained Localizers in Classical and Modern Chinese

Abstract

This paper discusses the Chinese localizers in terms of prosody. It argues that the development of localizers in Classical Chinese was a typological change from a synthetic to an analytic language type (Huang, Syntactic analyticity: the other end of the parameters. LSA Summer Institute Lecture Notes. MIT/Harvard, 2005, Xu, Typological change in Chinese syntax. Oxford Press, Oxford/New York, 2006) conditioned on the “multi-syllabic constraint” (Sun, Two conditions and grammaticalization of the Chinese locative. In Xu D (ed) Space in languages of China: cross-linguistic, synchronic and diachronic perspectives. Springer Science, Heidelberg, pp 199–288, 2008) which is a sub-case of the prosodic effects determined by the Nuclear Stress Rule and the newly developed disyllabic foot structure (Feng, Linguistics 6:1085–1122, 2003). Historical evidence is provided to demonstrate the grammaticalization process of localizers with the parallel development of light verbs and light nouns in the history of Chinese.

1 Introduction

Locational phrases such as ‘on NP’, ‘at NP’ and ‘in NP’ need a localizer in Chinese for non-local nouns, for example:

(1)	a.	Shū	zài	zhuōzi	shang
		book	exist	table	on
		‘The books are on the table.
	b.	Tā	cóng	chōutì	lǐ	ná	chū	yī	ben	shū
		he	from	drawer	insider	take	out	one	CL	book
		‘He took out a book from a drawer.
	c.	Tā	cóng	Běijīng	lái
		he	from	Beijing	come
		‘He comes from Beijing.’

In Archaic Chinese, however, no localizers were required as exemplified in (2a–b).

(2)

a.

八佾舞於庭,是可忍也,孰不可忍也。 (論語·八佾 Analects of Confucius)

Bāyì

wǔ

yú

tíng,

shì

kě

rěn

yě,

shú

bùkě

rěn

yě.

8x8

dance

at

court,

this

can

tolerate

Prt.

what

not can

tolerate

Prt

‘(Confucius said of Jishi) Eight rows of eight dancers in the house courtyard—if this can be tolerated, what cannot be tolerated?’

b.

樹吾墓槚, 槚可材也。 (左傳·鲁哀剬 11 年 Zuo Zhuan)

Shù

wú

mù

jiǎ,

jiǎ

kě

cái

yě.

Plant

my

tomb

catalpa,

catalpa

can

good-quality

prt

‘Plant a catalpa tree on my tomb; it can be used as timber.’

c.

Lìushí sì

ge

rén

zai tái

*(shang)

tiàowǔ.

sixty four

CL

people

at stage

top

dance

‘Sixty-four people are dancing on the stage.’

Neither tíng ‘court’ nor tái ‘stage’ are inherently locative nouns and (2c) is ungrammatical without a localizer (shàng) in Modern Chinese but there are no localizers attested in Archaic Chinese in general (Chao 1968; Wang 1958; Chou 1961; Wei 2003; Feng 2014). Actually, from an etymological point of view, the localizers which developed in Medieval Chinese and are used in Modern Chinese, namely, qián ‘ahead, before’, hòu ‘back, behind’, shàng ‘above’, xià ‘below’ and lǐ ‘in’, were presumably all nouns in Archaic Chinese. Paleographic evidence and cognate relationships show that localizers in Medieval Chinese all originated as concrete objects in Archaic or pre-Archaic Chinese. For example, qián had an old meaning, “the front part of a boat (or a toe)”; hòu very likely referred to ‘buttocks’, and lǐ 裏 was certainly a noun meaning ‘inside’ in Old Chinese. Based on the concrete object origin and the general principle that words for abstract notions are derived from lexical content items with concrete meanings, the terms shàng ‘top’ 上 and xià ‘bottom’ 下 may be derived cognitively from tang 堂 ‘high land, court’ and yu 窊 ‘marsh land, lowerlands’.¹ (see Zhang Taiyan’s Wen Shi).

The noun-like usages of pre-localizers such as shàng, xià, etc. can be seen in the following examples.

(3)	a.	至于靡笄之下 (左傳 · 成剬 2 年 Zuo Zhuan)
		Zhì	yú	Mǐjī	zhī	xià
		Reach	to	Miji	‘s	bottom
		‘To arrive at the bottom of Miji Mountain.’
	b.	舍于昌衍之上 (左傳 · 僖剬 29 年 Zuo Zhuan)
		Shě	yú	Chāngyǎn	zhī	shàng
		Reside	at	Changyuan	‘s	top
		‘(Gelu) resides at the top of Changyan.’

The pre-localizers in (3) are used with a genitive marker zhī indicating that they are independent nouns.

With respect to the noun origins of the words under discussion, the question is what motivated them to become localizers in Medieval Chinese. In what follows, I will first discuss the syntax of localizers and then examine the distribution of some spatial nominals (locative nouns) and their grammaticalized usages as localizers in Late Archaic and Medieval Chinese. This paper is organized as follows: Sect. 2 discusses the syntax of localizers. Section 3 argues that the emergence of localizers is prosodically motivated. Section 4 compares parallel developments of prosodically motivated light-verb and light-noun constructions, respectively, in Classical Chinese (Feng 2012). Section 5 is a summary of this paper, synthesizing prosodically motivated syntax in terms of nouns, verbs, light verbs, light nouns and localizers as well.

2 The Syntax of Localizers and the Questions Involved

Huang (2009) proposed a structure for locative PPs in Archaic Chinese, assuming that there is a covert localizer (represented by a capital ‘L’) cross-linguistically.

(4)

The surface structure of (4) is syntactically derived (i.e., [_PP P [_LP NP_i [ _L’ L t_i]] through a head movement caused by the [+strong] feature under L (the LOCAL feature in Huang’s system). This structure, as Huang argues, will generate all Archaic forms exemplified in (2). The locative words like xià mentioned above, however, were grammatically required later on in environments like the following (taken from Peyraube 1994; see also Cao 1999).

(5)

孔子去曹適宋,與弟子習禮大樹*(下)。 (史記 · 孔子世家 Shiji)

Kongzi

qù

Cáo

shì

Sòng,

yǔ

dìzǐ

xí

lǐ

dà

shù

xià

Confucius

leave

Cao

go

Song,

with

disciple

practice

rite

big

tree

under

‘Confucius left Cao and went to Song to practice the rites with his students under a big tree.’

As pointed out by Peyraube (2003), prepositions were required to be overt in the Pre-Medieval Chinese period for non-locative nouns, and so it may seem that the non-existence of (5) is due to the missing preposition. However, PPs are perfectively grammatical with a missing P in Archaic Chinese, as seen in (6).

(6)	子產使校人畜之池。孟子 · 萬章上
	Zichan	shǐ	Xiàorén	xù	zhī	chí
	Zichan	order	Xiaoren	put	it	pool
	‘Zichan orders Xiaoren to put it in the pool.’

The question then is why an overt localizer (cf. xià) in (5) is necessary while it is not so in (6) with respect to the null P structure. In fact there is no example in which a locative phrase is formed with a non-locative noun where both the localizer and the preposition are overtly missing. This situation has motivated Huang (2009) to come up with a new analysis within the following tree structure.

(7)

According to Huang (2009), when the [+strong] feature of a null L is lost in Medieval Chinese, the ‘L’ must be filled with a lexical head (the localizer xià, shàng, lǐ, etc.), as in (7). When the object NP ‘big tree’ moves up to the Spec of LP (for reasons of Case), the surface structure dà shù xià ‘under a big tree’ is derived. The difference between Archaic and Medieval (including Modern) Chinese is therefore formally characterized as a loss of the null localizer (i.e., the strong functional feature) which is replaced by phonetically realized localizers.

The structure of locative PPs provided by Huang in (7) is extremely important for formal analyses of Classical Chinese and it motivates a number of questions as well.

First, a natural question raised by Huang’s analysis is the cause of the loss of [+strong]. What happened to the [+strong] feature and how did it come to disappear around the Han dynasties?

Second, there are cases where both the localizer and the preposition are overtly missing, during and after the Pre-Medieval period. For example (see Li Guo (2013) for more examples of this type).

(8)	a.	遭之塗. (史記 · 管晏列傳 Shiji, ca. 100 BC)
		Zāo	zhī	tú
		Meet	him	road
		‘meet him on the street.’
	b.	Fēijī	yào	luò	zài	jiē	*(shang)		(Modern Chinese)
		aircraft	wants	land	at	street
		‘The aircraft would like to land on the street.’
	c.	請著之竹帛. (史記 · 孝文本紀 Shiji, ca. 100 BC)
		Qǐng	zhù	zhī	zhú	bó
		Please	write	it	bamboo	silk
		‘Please write it on bamboo and silk.’
	d.	Qǐng	bǎ	tā	xiě	zài	zhúzi		*(shang). (Modern Chinese)
		please	ba	it	write	at	bamboo
		‘Please write it on the bamboo.’

The type of counterexamples given in (8) occurred not only in Medieval Chinese, they can also be found in Modern Chinese. For example (see Chu 2004):

(9)	a.	你在黑板寫, 我在書上寫.
		Nĭ	zài	hēibăn	xiě,	wŏ	zài	shū	shàng	xiě
		You	at	blackboard	write,	I	at	book	up	write
		‘You write on the blackboard, and I write on the book.’
	b.	你在黑板上寫, 我在書上寫
		Nĭ	zài	hēibăn	shàng	xiě,	wŏ	zài	shū	shàng	xiě
		You	at	blackboard	up	write,	I	at	book	up	write
		‘You write on the blackboard, and I write on the book.’

The locative phrase zài hēibǎn ‘on the blackboard’ in (9) is perfectively grammatical without a localizer. As a result, it is not always the case that the ‘L’ must be filled with a lexical head by phonetically realized localizers like xià, shàng, etc. Why is this so? It is a mystery not only in modern Chinese but is also directly related to historical syntax of Chinese. This is because, first, it is unclear why the ‘L’ feature in Medieval Chinese (7) can be realized by a monosyllabic lexical-head (cf. shàng, xià, lǐ, wài, etc.) when the locative phrase is an adjunct (10a), but the ‘L’ must be disyllabic when the locative phrase is a predicate, as observed in Sun (2008). For example:

(10)	a.	終日在裏(面)默坐 (朱子語類 : 卷一百一十三 Zhuzi Yulei)
		Zhōng-rì	zài	lĭ	mò	zuò
		all-day	at	inside	quiet	sit
		‘Sit quietly inside all day long.’
	b.	有幾個秀才在里*(面) (警世通言 Jingshi Tongyan)
		Yǒu	jĭ-ge	xiùcái	zài	lĭ	*(mian)
		Have	several-CL	scholar	at	in	*(−side)
		‘There are several scholars inside.’

As Sun pointed out (2008), there is a complementary distribution between monosyllabic localizers and disyllabic localizers in terms of adjunct versus predicate. However, it is unknown why lǐ-mian ‘in-side’ must be used when it serves as a main predicate with zài (10b) while a monosyllabic localizer lǐ ‘inside’ is enough to realize the L-feature in (7) if it is used in an non-predicate (adjunct) position (10a). In other words, why should the L-feature be sensitive to predicate/adjunct and mono-/di-syllabic distinctions, respectively?

Still another mystery remains involving localizers in Modern Chinese as discussed in Feng (2003). There is a grammatical contrast between monosyllabic and disyllabic localizers in nominal structures as in the following:

(11)	a.	書在桌子上(頭)
		Shū	zài	zhuōzi	shàng	(tou)
		Book	at	table	top	(side)
		‘The book is on the table.’
	b.	書在桌子的上*(頭)
		Shū	zài	zhuōzi	de	shàng	*(tou)
		Book	at	table	de	top	(side)
		‘The book is on top of the table.’

The question is: what is the syntactic status of monosyllabic localizers and their corresponding disyllabic localizers?²

All of the questions above call for further explanation and analysis of the mysteries regarding the bizarre behavior of localizers in Chinese historical syntax. In what follows I will propose a prosodic analysis to account for the questions raised above.

3 Prosodically Motivated Localizers

Although the structure of localizers is syntactically generated and universally formed (Huang 2009), the original emergence and further development of such structures would not have taken place without a proper motivation (assuming that prosody is a parametric factor for activating UG operations). What then motivates the change in the first place and what constraints their development later on? To date, there are no adequate explanations for these questions. What I would like to suggest in this paper is the following: It is prosody that gave rise not only to the new grammar of localizers but also to the new light-verb and light-noun syntax, emerging around the same time in Chinese history. Evidence and analyses for the prosodic hypothesis of the development of localizers are given below.

First, as seen in (3), locative words like shàng ‘top’ and xià ‘bottom’ could occur in [N 之下] (‘the bottom of N’) in Archaic Chinese functioning as an independent noun. This situation, however, changed in Medieval Chinese. That is, more and more locative words were adjacent to monosyllabic nouns forming a disyllabic unit ([N xià/shàng]), as shown in (12).

(12)	a.	葬之郐城之下 (左傳·僖剬 33 年 Zuo Zhuan)
		Zàng	zhī	Kuài	chéng	zhī	xià
		Bury	it	Kuai	City	‘s	bottom
		‘Bury him at the bottom of Kuai City.’
	b.	齊梁之兵連于城下 (史記·張儀列傳 Shiji)
		Qí	Liáng	zhī	bīng	lián	yú	chéng	xià
		Qi	Liang	‘s	army	join	at	City	bottom
		‘Qi and Lian’s army joined at the bottom of the city.’

The tendency to change from [N ‘s xià] (‘N’s bottom’) to [N xià] (‘N bottom’) was prosodically motivated (forming a Disyllabic Unit) and modified the syntax (locative words became localizers). This, I argue, is the origin of localizers that changed from locative nouns to a functional category of localizers by gradual loss of their noun properties (Roberts 2007; Roberts and Roussou 2003). Both situations, being a syntactic head of non-locative noun and binding prosodically with a non-locative noun, caused the head (i.e., the locative word xià, shàng, etc.) to be reanalyzed as occupying the L position (13), and finally gave rise to a new category of localizers in the history of Chinese, as seen (14).

(13)

(14)

3

The prosodic effect on grammaticalization of localizers can be seen from the fact that, first, more and more monosyllabic names became disyllabic as seen in Table 1 (taken from Sun 2008), and second, more and more locative words (shàng, xià, etc.) merged with non-locative nouns as seen in (16). Both were in fact required and reinforced by the newly established disyllabic foot structure (Feng 1997) during the Han dynasty.

Table 1 Old Chinese place names in the region from 34° to 36° north and 111° to 116° east (Tan 1982; Sun 2008)

Note that in Modern Chinese, monosyllabic place names are ungrammatical when they are used alone. For example:

(15)	a.	(孔子)已而去魯。(史記 · 孔子世家 Shiji - Kongzi Shijia, ca. 100 BC)
		(Kǒngzi)yǐ’er	qù	Lǔ
		Confucius	shortly	departure	Lu
		‘Confucius left Lu State shortly after.’
	b.	Míngtiān	wǒ	xiǎng	qù	*T ō ng 4
		tomorrow	I	want	go	Tong
		‘I want to go to Tong County tomorrow.’
	c.	Míngtiān	wǒ	xiǎng	qù	T ō ng	Xiàn
		tomorrow	I	want	go	Tong	county
		‘I want to go to Tong County tomorrow.’
	d.	Míngtiān	wǒ	xiǎng	qù	Dàxīng
		tomorrow	I	want	go	Daxing
		‘I want to go to Taxing County tomorrow.’

The prosodically determined grammaticality with respect to the monosyllabic place names in Modern Chinese (15b) is a result of a typological change from moraic foot structure (Archaic) to syllabic foot structure (Medieval), starting as early as the third century BC and accelerating during the Han Dynasty (100 BC; see Feng 1997). As a consequence of the typological change, a clear contrast between a monosyllabic noun with a localizer (xià, shàng, lǐ, etc., i.e. [V + [P [[σ]_PN + xia下]]]]) and a disyllabic noun without one (i.e., [V + P [[σ σ]_PN + NULL]]]]) is seen in the following examples.

(16)

a.

至 [[σ] PN  + 下]

至 [[σ σ] PN  + null ]

吳為鄒伐魯, 至城下。(史記 · 魯國剬世家, Shiji)

信遂追, 北至城阳。(史記 · 淮陰侯列傳 ibid)

Wú

wèi

Zōu

fá

Lǔ,

zhì

Chéng

xià

Xìn

suí

zhuī,

běi

zhì

Chéngyáng

Wu

for

Zou

attack

Lu,

reach

City

bottom

Xin

then

chase,

north

reach

Chengyang

‘Wu attacked Lu for Zou, and arrived at the city wall.’

‘Xin then went after (Xiang Yu), and arrived at Chengyang in the North.’

b.

至城下, 圍其西北。(漢書, Hanshu)

至城门, 遂複言。(漢書, Hanshu)

Zhì

Chéng

xià ,

wéi

qí

xīběi

zhì

Chéng

Mén ,

suí

fù

yán

arrive

city

bottom,

surround

its

northwest

arrive

City

Gate,

then

again

talk

‘To arrive at the city wall…’

‘To arrive at the city gate…’

c.

到城下 (後漢書, Houhanshu)

至城郭。(後漢書, Houhanshu)

Dào

Chéng

xià

Zhì

chéng-guō

Arrive

City

bottom

arrive

City-wall

‘To arrive at the city wall.’

‘(They) arrived at the city wall.’

In each of the historical records (Shiji, Hanshu and Post-Hanshu), there is a xià ‘bottom’ used with monosyllabic cheng, forming a disyllabic locative-denoting noun chéng-xià ‘City wall’. However, when the name of the city is disyllabic (i.e., Chéngyáng), xià ‘bottom’ is not needed and hence is not used. Thus, it is plausible that a reanalysis of the locative words (xià, shàng, lǐ, etc.) as localizers took place in the very environment where the spatial nominals were monosyllabic. The following example confirms this analysis:

(17)

a.

臣始至于境 (孟子·梁惠王下 Mengzi, c.a. 300 BC)

Chén

shǐ

zhì

yú

jìng .

I

just

arrive

at

frontier

‘I just arrived at the frontier.’

a′.

臣嘗从大王與燕王會境上 (史記 · 廉颇藺相如列傳 Shiji, c.a. 100 BC)

Chén

cháng

cóng

Dà

Wáng

yǔ

Yān

Wáng

huì

jìng

shàng

I

ever

follow

Great

King

at

Yan

King

meet

frontier

top

‘I have ever followed the Great King to meet the Yan King at the Frontier.’

b.

是聖人僕也。是自埋于民,自藏于畔。(莊子 · 則陽 Zhuangzi, c.a. 300 BC)

Shì

shèngrén

zhī

pú

yě,

shì

zì

mái

yú

mín ,

zì

cáng

yú

pàn

This

Sages

‘s

servant

Prt.

this

self

burry

in

people,

self

hide

in

field

‘This is a servant of sages. It is the case that one hid himself among the people and in the fields.’

b′.

分散在民間 (論衡 Lunheng, 100 AD)

Fēnsàn

zài

mín

jiān

Scattered

in

people

among

‘(Jupiter) scattered among people.’

Regardless of whether or not the locative expressions shàng ‘top’ and jiān ‘among’ used here are analyzed as nouns or localizers, the fact is that they were not required before the Han Dynasty (221 BC), as seen in (17a–b), but were prosodically necessary in the language used in the Shiji (100 BC) and Lunheng (100 AD). As a result, the necessity of using locative words in the Han languages was originally and primarily a prosodic constraint, and only through further development was a syntactic categorical grammar reanalyzed as localizers in later stages.

The argument for prosodically motivated localizers is further supported by the fact that only in the Nuclear Stress (NS for short) position, as seen in (19), are disyllabic localizers necessary and hence developed. For example:

(18)	a.	Qīutiān	dàyàn	dōu	wǎng	nán	(biān)	qiānyí
		autumn	wild-goose	all	to	south	(side)	migrate
		‘All wild geese migrate south in Autumn.’
	b.	Qīutiān	dàyàn	dōu	qiānyí	dào	nán	*(biān)	le.
		autumn	wild-goose	all	migrate	to	south	side	Asp.
		‘All wild geese migrate south in Autumn.’

As Sun (2008) has pointed out, there is a syllabic contrast in terms of the grammaticality of [direction + localizers] in different syntactic positions. Although the observation is correct, a question remains: What is the determining factor involved here? Actually the complementary distribution of the localizers used between pre-verbal and post-verbal positions is, I would like to propose, a natural consequence of applications of the Government-based Nuclear Stress Rule (NSR) stated as follows.

(19)	The Government-based Nuclear Stress Rule
	Given two sister notes C1 and C2, if C1 and C2 are selectionally ordered (see Zubizarreta 1998), the one lower in selectional ordering and containing an element governed by the selector is more prominent.

Following Liberman (1975), Feng (1995) and Zubizarreta (1998), Feng (2003) proposed that the Nuclear Stress of a sentence is, informally speaking, assigned by the verb to its mutually c-commanded (i.e., directly governed) complement, termed the Government-based Nuclear Stress Rule (G-NSR, for short). Since there is only one primary stress per sentence, only the directly governed complement gets the nuclear stress; the second constituent after the verb (or the complex verb [V + C], etc.) is not allowed due to the lack of a proper stress in the sentence.

According to the G-NSR, the monosyllabic directional noun in (18b) will be analyzed as not being heavy enough to realize the Nuclear Stress (NS) assigned by the verb at the end of the sentence, thus a localizer is naturally used to fulfill the disyllabicity requirement, otherwise, the sentence would be prosodically ungrammatical (or ineffable).

Under this analysis, we can now explain why there is a complementary distribution between monosyllabic localizers and disyllabic localizers in terms of predicate and non-predicate positions, as seen in (10), repeated here as (20).

(20)	a.	終日在裏(面)默坐 (朱子語類 : 卷一百一十三 Zhuzi Yulei)
		Zhōng-rì	zài	lĭ	mò	zuò
		all-day	at	inside	quiet	sit
		‘Sit quietly inside all day long.’
	b.	有幾個秀才在里*(面) (警世通言 Jingshi Tongyan)
		Yǒu	jĭ-ge	xiùcái	zài	lĭ	*(mian)
		Have	several-CL	scholar	at	in	*(−side)
		‘There are several scholars inside.’

Obviously, lǐ-mian ‘in-side’ must be used when it appears in the NS position (10b) while a monosyllabic localizer lǐ‘in’ is sufficient to realize the L-feature (7) because it occupies an adjunct (or non-predicate) position where no NS is required as in (10a). Similarly, the grammatical contrast between monosyllabic non-locative nouns (with a localizer) and disyllabic non-locative nouns (without locaizer) as seen in (9) can also be explained. Compare:

(21)	a.	你在黑板寫,我在書上寫
		nĭ	zài	hēibăn	xiě,	wŏ	zài	shū	shàng	xiě
		2nd	at	blackboard	write,	I	at	book	up	write
		‘You write on the blackboard, and I’ll write in the book.”
	b.	你在黑板寫, *我在書寫
		nĭ	zài	hēibăn	xiě,	wŏ	zài	shū	xiě
		2nd	at	blackboard	write,	I	at	book	write
	c.	*你把字寫在黑板*(上)
		nĭ	ba	zi	xiě	zài	hēibăn	shang
		You	ba	character	write	at	blackboard	up
		‘You write on the blackboard, and I’ll write in the book.’
	d.	你在本上寫吧
		nĭ	zài	běnr	shang	xiě	ba
		You	at	notebook	CK	write	Prt
		‘Would you write on a notebook?’
	e.	*你在本兒寫吧
		nĭ	zài	běnr	xiě	ba
		You	at	notebook	write	Prt
		‘Would you write on a notebook?’
	f.	你在筆記本寫吧
		nĭ	zài	bĭjìběnr	xiě	ba
		You	at	notebook	write	Prt
		‘Would you write on a notebook?’

It is obvious that the missing localizer can only be permitted either preverbally or within parallel sentences. This is so because contrastive stress or parallel prosody overrides the Nuclear Stress assigned only to the complement of the verb postverbally (in a rhetorically stress-neutral situation), and thus, only G-Based NS positions (postverb) are prosodically heavy.⁵

Not only can the grammatical contrast between monosyllabic localizers and disyllabic localizers in Modern and Medieval Chinese be explained naturally based on the theory presented here, but also the monosyllabic non-locative nouns can be explained systematically as well. As seen before, unlike Archaic Chinese, Medieval and Modern Chinese grammar do not allow monosyllabic nouns to be used for location-denoting purposes even if they appear in preverbal position. For example (taken from Sun 2008):

(22)	a.	Tā	xǐhuān	zài	shān	*(shang)	kàn	rì	chū
		he	like	at	mountain	top	see	sun	out.
		‘He likes to watch sunrise on a mountain.’
	b.	Tā	xǐhuān	zài	gāo	shān	(shang)	kàn	rì	chū
		he	like	at	high	mountain	top	see	sun	out
		‘He likes to watch sunrise on a high mountain.’

How can we rule out sentences like (22a)? Based on Huang’s theory, we suggest the following three different operations.

(23)

As Huang pointed out, in Medieval Chinese, the Archaic covert null L feature was lost; however, as seen above, the medieval overt-L (the localizers) was most likely grammaticalized under a prosodic motivation. In the current analysis, it is highly plausible that prosody forced monosyllabic non-locative nouns to be combined with a monosyllabic locative word (shàng, xià, etc.) to fulfill the NSR (19) and/or Disyllabicity. Thus, only through such operations can the locative words be reanalyzed as occupying the L position, giving rise to a new functional category for the localizer.

4 Prosodically Motivated Light Verbs and Light Nouns in Medieval Chinese

The development of localizers is not a sporadic case of prosodically motivated syntax in Classical Chinese. As observed by Feng (2005), the null light verbs in Archaic Chinese were also phonetically realized under prosodic pressure in Medieval Chinese. For example (taken from Feng 2005):

(24)	a.	(顆)夜夢之曰…(‘余, 而所嫁婦人之父也;) (左傳 · 宣剬 15 年 Zuo Zhuan)
		(Kē)	yè	mèng	zhī	yuē…
		Ke	night	dream	it,	say
		‘Ke dreamed of it in the night and said…’
	b.	(相如)與卓氏婚, 饒於财。(史記 · 司馬相如列傳 Shiji)
		(Xiang Ru)	yǔ	Zhuó	Shì	hūn	ráo	yú	cái
		(Xiang Ru)	and	Zhuo	Miss	marry,	rich	at	future
		‘(Xiang Ru) and Miss Zhuo got married and are extremely rich.’
	c.	無友不如己者。(論語 · 學而 Lunyu)
		Wú	yǒu	bù	rú	jǐ	zhě
		No	friend	no	like	self	nom
		‘Do not make friends with those who are not as good as yourself.’

The words mèng ‘dream’, hūn ‘marriage’, and yǒu ‘friend’ are generally used as nouns while they also functioned as verbs taking an object in Archaic Chinese (Takashima 2005). It is assumed that there was a covert light verb DO used in Archaic Chinese (Feng 2005) and only around the time of the Eastern Han (200 AD) did phonetically realized light verbs such as zuò 作 ‘do’, qǐ 起 ‘up’, xing 興 ‘appear’ start to appear. For example:

(25)

a.

其夜作夢, 見有人來。(法苑珠林卷 76 Fayuan Zhuzilin zhuan)

Qí

yè

zuò

mèng,

jiàn

yǒu

rén

lái

that

night

do

dream,

see

have

people

come.

‘(He) he had a dream that night where he saw someone coming.’

b.

仁者何用工巧之人共作婚為? (佛本行集經 · 卷 13 Fo Benhang Jijing)

Rén

zhě

hé

yòng

gōngqiǎo

zhī

rén

gòng

zuò

hūn

wéi?

nice

person

why

use

exquisite

’s

person

together

DO

marriage

Q

Why a nice person would take an exquisite person to marry with?

c.

我不用汝與我作友 (佛本行集經 · 卷 25 Fo Benhang Jijing)

Wǒ

bú

yòng

rú

yǔ

wǒ

zuò

yǒu.

I

not

use

you

with

I

do

friend

‘I don’t need you to be friends with me.’

Why did overt light verbs suddenly appear in the language after the Han? The emergence of overt light verbs in the history of Chinese syntax is a mystery not solved until Xu (2006) Hu (2005) and Feng (2005). However, disyllabic verbal expressions are expected to be favored under the NSR (19), which can be satisfied by any syntactic means. Thus, all overt light-verb expressions, as seen in (25), are located within the NS domain. In fact, making the archaic covert light verbs overt was merely one of many syntactic strategies activated under the prosodic grammar during the Late Medieval Chinese (see Feng 2011 for more prosodic effects on syntax). The historical change of light verbs (from covert to overt) can be seen even more clearly in the examples given in (26).

(26)

a.

不鼓自鳴6 (佛本行集經 · 卷 2 Fo Benhang Jijing)

Bù

gǔ

ér

míng

Not

drum

and

sound

‘No (one) drummed (beat) the drum but it sounds.

b.

c.

時彼大眾…或複騰鈴,或複打鼓。(佛本行集經 · 卷 8 Fo Benhang Jijing)

Bǐ

shí

dàzhòng…

huò

fù

téng

líng,

huò

fù

dǎ

gǔ

That

time

people…

or

again

ring

bell,

or

again

hit

drum

‘At that time people… either rang bells again or beat drums again.’

d.

不久打鼓,明星欲出。 (佛本行集經 · 卷 36 Fo Benhang Jijing)

Bù

jiǔ

dǎ

gǔ,

míng

xīng

yù

chū

Not

long

hit

drum,

bright

star

will

out

‘Not long afterwards, (they) beat the drum when the bright stars came out.’

It is well known that nouns like gǔ ‘drum’ could easily be verbalized as seen in (26a–b). However, by the time of Late Medieval Chinese (i.e., after the Han, ca. 200 AD), the denominalization operation (such as gǔ ‘drum’ used as a verb) was gradually replaced by a process of adding a semantically empty verb like zuò ‘do’ (in the present case) before the noun (i.e., gǔ), making a VO phrase out of a monosyllabic verb (or a noun). What is most important to note here is the fact that all of the sentences above will become prosodically odd without overt light verbs, for example:

(27)	時彼大眾…或複騰鈴,或複*鼓。 (佛本行集經 · 卷 8 Fo Benhang Jijing)
	Bǐ	shí	dàzhòng…	huò	fù	téng	líng,	huò	fù	* g ǔ .
	That	time	people…	or	again	ring	bell,	or	again	*drum.
	‘At that time people… either rang a bell again or beat a drum again.’

This is why there are no examples like (27) in historical documents. Why must there be an overt light verb (or a verb) in each of the sentences in (26b–d)? As suggested above, the light verb construction was a parallel development to the prosodically motivated syntactic changes in Archaic Chinese in an across-the-board fashion.

Parallel development of prosodically motivated light noun (i.e., classifier) constructions (Feng 2012) also exhibits a prosodic motivation: a classifier emerges when the number is monosyllabic (such as qí ‘seven’), while the classifier can be omitted from the [N Num CL] structure if the number is disyllabic (such as shí-bā ‘eighteen’), as seen in (28).

(28)	a.
	b.	七枚熱鐵丸…十八鐵丸。 (法苑珠林 Fa yuan zhu lin)
		qī	méi	rè	tiě	wán…	shí-bā	tiě	wán.…
		seven	CL	hot	iron	ball…	ten-eight	iron	ball…
		‘(there are) seven hot iron balls… (and) 18 iron balls…’

Examples given in (28) once again show that if the numeric element is monosyllabic, a light noun is favored, indicating that prosody may also have affected the development of classifiers in Classical Chinese (Feng 2011), which parallels the development of light verbs and localizers not only in terms of movement lost (Huang 2009, 2013), but also in terms of chronological parallelism (Pan 1982; Shimura 1995; Wu 2003; Feng et al. 2008).

5 Conclusion and Final Remarks

In this paper, I have adopted Huang’s (2009) syntactic and Sun’s (2008) prosodic analyses for Chinese localizers in classical and modern Chinese. I have attempted to show that the typological change of Classical Chinese from synthetic to analytic can be characterized in terms of syntax (i.e., losing movement) motivated by prosody (the Nuclear Stress Rule and the disyllabic requirements).

Under the system of prosodic grammar proposed here, the scenario of localizer development is different from previous studies. First, under the current theory, prior to the [+strong] feature of the null L being lost in Archaic Chinese, pre-localizer xià, shàng, lǐ, etc., were used to fulfill the disyllabic requirement for monosyllabic nouns (whether intransically locational or not) through disyllabicity or in the Nuclear-Stress positions. When more and more shàng, xià elements were used as prosodic place-holders for location-denoting nouns in Medieval Chinese, the pre-localizers (xià, shàng, lǐ, etc.) were reanalyzed as lexical heads in the ‘L’ position of (7). Only then were true localizers born, giving rise to the surface structure of dà shù xià ‘under a big tree’ (5). The difference between Archaic and Medieval (including Modern) Chinese is therefore not simply a loss of the null localizer (i.e., the strong functional feature), but instead the [+L] feature was inherited by (1) a disyllabic place name in non-NS positions and (2) is replaced by a new category of localizer especially in NS positions.

In fact, the prosodic hypothesis given here is not limited to localizer development (Behr 2010; Redouane 2007). The prosodic phenomena among the three functional categories, namely light verbs, classifiers and localizers discussed in this and other papers (Feng 2005, 2012) were all newly created syntactic structures during the typological change from Old Chinese to Medieval Chinese, and each of them requires a systematic and unique analysis in terms of prosodic syntax that activates relevant parametric factors and motivates grammaticalization in the history of Chinese, a fascinating new area for linguistic studies.