See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326416384

Word order and sentence structure in Mandarin Chinese: new perspectives

Thesis · March 2018

CITATIONS READS
3 8,469

1 author:

Anna Morbiato
Università Ca' Foscari Venezia
22 PUBLICATIONS   24 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A multi-dimensional approach to word order in Mandarin Chinese: functions, factors, restrictions View project

Acquisition of Chinese by Italian L1 learners View project

All content following this page was uploaded by Anna Morbiato on 27 July 2018.

The user has requested enhancement of the downloaded file.

Thesis submitted in partial fulfilment of the requirements for the degree of
Cotutelle Ph.D. in Asian and African Studies (Ca’ Foscari University of Venice)
and in Linguistics (The University of Sydney)

Word order and sentence structure

in Mandarin Chinese: new perspectives
SSD:L-OR/21

Supervisors
Prof. Magda Abbiati
Ca’ Foscari University of Venice, Department of Asian and North African studies
Prof. William Foley
The University of Sydney, Department of Linguistics

Graduand
Anna Morbiato
Matriculation Number: 819933

Academic Year
2017/ 2018
Table of Contents

Acknowledgments .......................................................................................................... 1

Abstract........................................................................................................................... 3

Abbreviations ................................................................................................................. 5

1. Introduction ............................................................................................................. 7

1.1. Overview and research objectives ................................................................... 10

1.2. Theoretical assumptions ................................................................................. 17

1.3. Methodology and linguistic data..................................................................... 19

1.4. Structure of the thesis ..................................................................................... 21

1.5. Limitations of the study .................................................................................. 23

2. Grammatical Relations .......................................................................................... 25

2.1. Overview .......................................................................................................... 25

2.2. Methodology .................................................................................................... 27

2.3. Literature review and terms of the debate ......................................................30

2.4. GR-sensitive constructions: the case of MC ................................................... 41

2.4.1. Relativisation site...................................................................................... 43

2.4.2. Reflexivisation........................................................................................... 46

2.4.3. Imperatives ............................................................................................... 55

2.4.4. Diathesis and passive ................................................................................ 56

2.4.5. Topic extraction out of relative clauses .................................................... 72

2.4.6. Word order permutations ......................................................................... 76

2.4.7. Co-reference constructions ....................................................................... 78

2.4.8. Control constructions and voice switch.................................................... 79

 2.4.9. Finite and non-finite constructions .......................................................... 81

2.4.10. Raising constructions .............................................................................. 87

2.4.11. Topic extraction ....................................................................................... 91

2.4.12. Conjunction reduction ............................................................................. 93

2.4.13. Floating and quantifier float .................................................................... 94

2.1. Interim summary ............................................................................................ 99

3. Constituent Structure .......................................................................................... 104

3.1. Overview ........................................................................................................ 104

3.2. Methodology .................................................................................................. 105

3.3. Noun phrases................................................................................................. 110

3.3.1. Indivisibility ............................................................................................ 110

3.3.2. Fixed order ............................................................................................... 111

3.3.3. Substitution/replaceability as a whole .................................................... 113

3.3.4. Required elements ................................................................................... 114

3.3.5. Movement/distribution ........................................................................... 115

3.3.6. Coordination ............................................................................................116

3.3.7. Summary .................................................................................................. 117

3.4. Verb phrases .................................................................................................. 118

3.4.1. Indivisibility ............................................................................................ 120

3.4.2. Fixed order ............................................................................................... 121

3.4.3. Substitution/replaceability as a whole ................................................... 124

3.4.4. Required elements .................................................................................. 132

3.4.5. Movement/distribution .......................................................................... 134

3.4.6. Coordination ........................................................................................... 137

3.4.7. Summary ................................................................................................. 142

3.5. Interim summary .......................................................................................... 142

4. Argument Structure ............................................................................................. 144

 4.1. Overview ........................................................................................................ 145

4.2. Event structure and argument structure ...................................................... 146

4.2.1. Salient aspects of event structure and its encoding ............................... 148

4.2.2. Event structure and aspect ..................................................................... 150

4.2.3. Causativity and its semantic representation ........................................... 151

4.2.4. Aspectual and causal shifts ..................................................................... 154

4.3. Methodology and framework of analysis ...................................................... 157

4.4. Verb classes and argument realisations in MC ............................................. 160

4.4.1. Verbs denoting states, conditions, or properties ................................... 160

4.4.2. Psychological state predicates ................................................................. 171

4.4.3. Verbs of involuntary activities ................................................................ 174

4.4.4. Verbs of existence, appearance, disappearance ..................................... 175

4.4.5. Verbs of posture ...................................................................................... 176

4.4.6. Verbs of motion....................................................................................... 180

4.4.7. Verbs of action on objects ....................................................................... 182

4.4.8. Verbs of change of possession ................................................................ 184

4.4.9. Measure verbs ......................................................................................... 186

4.5. Argument alternations and aspectual/causal shifts ..................................... 187

4.5.1. Locative inversion and other argument inversions................................ 187

4.5.2. Contrastive versus stative reading .......................................................... 188

4.5.3. The stative-inchoative alternation .......................................................... 189

4.5.4. Causativity and the BA-BEI realisation patterns ................................... 193

4.6. Interim summary .......................................................................................... 198

5. Information Structure ........................................................................................ 202

5.1. Overview ....................................................................................................... 202

5.2. Methodology and framework of analysis ......................................................203

5.3. IS notions and terminological issues ........................................................... 206

 5.3.1. Topic and the sentence-initial position ................................................. 208

5.3.2. Aboutness vs. frame ................................................................................ 210

5.3.3. Information and cognitive status of topic referents ................................ 211

5.3.4. Comment ................................................................................................. 215

5.3.5. Focus ....................................................................................................... 216

5.4. IS in Mandarin Chinese ................................................................................ 219

5.5. Topic in MC .................................................................................................. 220

5.5.1. Topic as a frame setter ............................................................................ 223

5.5.2. Locatability as the cognitive status of topic referents ............................228

5.5.3. On the semantic relation between topic and comment: Qualia structure
242

5.5.4. Topic-comment structures as embedded structures .............................. 247

5.6. Focus and comment in MC ........................................................................... 256

5.6.1. Narrow, predicate and sentence focus patterns ..................................... 258

5.6.2. Focus in context: native speakers’ evaluation and corpus data ............. 261

5.6.3. Thetic (sentence-focus) sentences .......................................................... 267

5.6.4. Predicate focus (topic-comment sentences where comment=focus) .... 272

5.6.5. The principle of end focus....................................................................... 275

5.7. Word order freezing phenomena ..................................................................286

5.8. Interim summary .......................................................................................... 293

6. Conclusions .......................................................................................................... 295

7. References ............................................................................................................ 301

The limits of my language are the limits of my world.

― Ludwig Wittgenstein

To J. C.
Acknowledgments

My most profound thanks and heartfelt expressions of gratitude go to my PhD

supervisors, prof. Magda Abbiati, Department of Asian and North African Studies at
Ca’ Foscari University of Venice, and prof. William Foley, Department of Linguistics
at the University of Sydney. Throughout the years, Magda has given me over and above
her sound academic guidance. She has ceaselessly shared my passion and keenness for
the linguistic subjects of which we spoke, and has gifted me with much invaluable
advice, both professional and personal. I recall with joy our many zest-filled talks on
the subtle points of syntax, which were energetic and passionate, certainly a little
confusing to others but a delight for the both of us. William Foley, who warmly
welcomed me as a cotutelle student in my second year of Ph.D., provided me with
inestimable support throughout my path. It was an honour for me that such an
outstanding linguistic scholar agreed to better my work through his lavish experience.
I always trusted in his guidance: each time I was stuck in a thorny problem, he was
always there to show me to the perfect reference or reading to solve the irksome matter.
He allowed this research to be my own work but steered me in the right orientation
whenever I needed it. His knowledge, insights and passion for linguistics have been
among the most inspiring aspects of my doctorate path.

It also stands for me to share deepest gratitude to all the experts and scholars who
have helped during my PhD path. A special thanks to prof. Sun Chaofen (University of
Stanford), prof. Tao Hongyin (UCLA) and prof. James Tai (National Chung Cheng
University), who all generously made time for meeting me, and who provided
insightful advice on some crucial topics of my thesis. Much appreciation and thanks
also to prof. Denis Creissels (University of Lyon) and prof. Hilary Chappell (EHESS,
Paris), my masterclass instructors at the Summer School of Linguistic Typology 2016,
who helped me untangle some of the most problematic knots of my research. Then, I
acknowledge and thank Prof. Robert Van Valin (Heinrich Heine University in
Düsseldorf), prof. Helen De Hoop (Radboud University Nijmegen), and prof. Huang
Yan (University of Auckland) with whom I had the chance to chat and get invaluable

1
feedback on the quality of my analysis. Besides, I deeply thank Weilun Lu, Adriano
Boaretto, Carlotta Sparvoli, and Bianca Basciano, for their precious comments on my
research.

I would also like to acknowledge all of the native speakers who patiently and
enthusiastically gave their time to take part in the surveys and acceptability checks for
the hundreds of sentences I have analysed for this research, including Weijian,
Chenbei, Chenlei, Mus, Yufei, Weiqin Fan, Yichen Zhang, Zihui Qiu, Eli Chen, Chris
Liu, Bin Wang, Lainey, Linda, Edward, Xiaowei, Sophia Wu, Naomi Duan, Crystal,
Yulu, Hao, Alina, Jingyi, Teresa Tang, Siyu, Sherry, Qi, Xinlei, Jack, Chalmers Zhong,
Venus Hu, Nancy Wu, Chaoyuan, Jiayu Wu, Ziyu Ye, among others. Without their
passionate participation and input, the surveys could not have been successfully
conducted.

Last, from my heart special thanks goes to all those who made my PhD days beautiful
in Sydney, including Alba, Benedetta, Chris, Daniela, Darius, Elena, Elisabetta, Emma,
Gav, Kate, Juliet, Lina, Rhys, Rossella, Valentina, Toby, and all those who have been
there supporting me through good and bad times. Finally, I must express my very
profound gratitude to Giacomo, Nadia, Giorgio, Stefania, Simonetta, and Marta, for
providing me with unfailing support and continuous encouragement throughout my
years of study, and through the process of researching and writing this thesis. This
accomplishment would not have been possible without all of you. Thank you most
sincerely.

2
Abstract

Word order (WO) is one of the most fascinating and investigated topics in Mandarin
Chinese (MC) linguistics, and many accounts have been proposed on different WO
patterns and constructions. However, despite the large amount of research, several
WO related issues remain rather controversial. Crucially, no unified consensus exists
on the relationship between WO and the different dimensions of the language (i.e.
semantics, syntax and pragmatics), and on how these levels interact with each other.
The present thesis’s aim is twofold: (1) identify the categories that are useful to account
for WO patterns and variations in MC; (2) examine in greater depth the syntactic,
semantic and pragmatic factors that influence word order in MC, as well as how they
interact and impose constraints on possible WO variations. The novelty of the
approach lies on three aspects: (i) a typological, comparative perspective that benefits
from cross-linguistic investigation of WO phenomena in other languages; (ii) a bottom
up approach that employs cross-linguistically validated typological tools (e.g., GR tests,
or constituenthood tests) aimed at conducting the analysis on a language-internal
basis, and (iii) an empirical approach: the analysis avails itself of natural linguistic
data, mainly drawn from corpora, and relies on acceptability checks with native
speakers. Overall, the thesis highlights that WO patterns and constructions are
determined by the interplay of different factors and constraints. It also highlights that,
for the sake of clarity and ambiguity avoidance, WO constraints are hierarchically
organized, and WO freezing phenomena occur to allow disambiguation of participants
in the described event.

3
4
Abbreviations

The present thesis employs the Leipzig Glossing Rules for text glossing and
abbreviations (available at https://www.eva.mpg.de/lingua/pdf/Glossing-Rules.pdf).
Abbreviations include:

ARG argument

ABS absolutive case

BA 把 ba marker
BEI 被 bei marker
C comment
CD commuicative dynamism
CHG sentence-final modal particle 了 le
CL classifier
DE modification marker (的 de: noun modifier, 地 de: verb modifier or 得 de:
complement marker)
DET determiner
EXP experiential aspect
MOD modal particles
NUM numeral
NP nominal phrase
NPST non past
NUM numeral
QNT quantifier
Q question particle
PFV perfective aspect marker 了 le
PROG （正）在 (zhèng) zài progressive aspect marker
PST past tense marker
Q 吗 ma interrogative particle

5
REL relative clause
SM subject marker
T topic
TM topic marker
V verb

6
1.Introduction

Word order is one of the most fundamental aspects of grammar: it can be described as
referring to the temporal or linear sequence of words in an utterance/sentence and is
the necessary outcome of one of the universal design features of all languages, namely
linearity. To convey a message, speakers cannot but utter one linguistic element at a
time, and each element precedes and follows another. In every language, word order
allows speakers to specify the relations among states and events, as well as their
participants and settings. Nonetheless, languages may differ with regard to the extent
to which they rely on surface order in the encoding of meaning, as well as to the range
of permissible orders they exhibit: in this respect Mandarin Chinese1 (henceforth MC)
constitutes a rather interesting case in point.

Word order is one of the most fascinating and investigated topics in MC linguistics:
over the past decades, it has captivated the interest of linguists working within
different theoretical frameworks and has posed several challenges for existing
linguistic theories. The crucial role word order plays in MC information encoding is
captured by the following renowned quote by Chao Yuen-ren in A Grammar of Spoken
Chinese:

It is often said that all Chinese grammar is syntax, all Chinese syntax is word order,
and therefore all Chinese grammar is word order (Chao 1968: 260).

The role played by word order in the grammar is to a significant extent due to the
isolating nature of the language, which relies very little on inflectional or derivational

1In the present thesis, the term Mandarin Chinese (MC) refers to the standard language of the RPC, called 普通话
pŭtōnghuà ‘common language.’ It refers to a formal, educated variety of the Beijing dialect, and belongs to the
Mandarin group of Chinese languages (Sinitic family, see Chappell et al. 2007 for further discussion).

7
morphology to encode linguistic information. It is a general cross-linguistic tendency
for word order to contribute signalling the role of the participants in the described in
a sentence, as “word order is one of the primary devices languages offer speakers to
express who does what to whom” (Gershkoff-Stowe and Goldin-Meadow 2002:377).
However, while inflectional languages also rely on morpho-syntactic markers (e.g.,
agreement or case marking) to single out the role of event participants, MC relies
almost uniquely on the relative order between the verb(s)/predicative element(s) and
their arguments, as well as on the intrinsic semantic features (selectional restrictions)
of the verb. Additionally, the sequence of elements in the MC sentence contributes to
encoding cognitive/information status of referents (in terms of topicality, givenness,
definiteness, aboutness, in the sense of Chafe 1976), the temporal settings and
sequence of the described events and states, as well as other types of linguistic
information. In short, MC is ‘‘one of those languages that rely heavily on word order
as an underlying marking feature for meaning’’ (Ho 1993:138). Furthermore, MC word
order displays several features that have posed a challenge to linguistic accounts based
on, for example, the Greenbergian tradition of word order universal tendencies
(Chappell et al. 2007), as briefly discussed in the next section. Finally, a number of
topic-comment related phenomena, including so-called Chinese-style topics, hanging
topics, double nominatives, pseudo-passives, disposal constructions etc., have
captivated the attention of scholars for decades, and were initially argued to be
peculiar to MC. As a result, MC has been described as an example of a topic-prominent
language, or as a language where syntax plays a less relevant role as compared to
discourse in determining the structure of the sentence. In Huang’s words, “the unusual
character of word order in Chinese has [...] contributed to a continuing debate on the
‘true’ nature of word order in Chinese dating from the 1970s” (2013:84).

In the past decades, a considerable amount of research was conducted to determine

the nature and the restrictions of word order related phenomena, resulting in a rich
and voluminous body of literature on the topic. Among the most investigated
constructions are instances of underspecification of arguments (pseudo-passives,
equi-NP deletion, zero anaphors and topic chains), argument alternations and
inversions (topicalisations, BA and BEI constructions, locative inversions), patterns
that encode the information structure of linguistic elements (topic-
comment structures, hanging topics, double nominatives and other word order

8
permutations), and so-called conceptual principles (e.g., the principle of
temporal sequence, and the whole-before-part principle, among others). However,
despite the large amount of research, several word order related issues remain rather
controversial: such issues will be briefly presented in the next section. Furthermore,
as will be highlighted throughout the present work, accounts of the same phenomenon
proposed by different scholars often diverge significantly, and some suffer from biases
connected to theory- or framework- internal assumptions, 2 and do not consider
research results from large-scale cross-linguistic typological research conducted on
typologically different languages, often referred to as ‘exotic languages’. Crucially, no
unified consensus exists as to the relationship between word order in MC and the
different dimensions of the language (i.e., semantics–argument structure, syntax–
constituent structure, pragmatics-information structure). Specifically, no systematic
and comprehensive analysis has yet been proposed with respect to how these levels
interact with each other, what restrictions each level displays, and what hierarchy
holds between such restrictions: in short, how each part of the grammar contributes
to determining the choice of word order patterns and constructions and hence the final
sequences of elements in the sentence.

The present thesis aims at providing a fresh look at word order permutations in
Mandarin Chinese by exploring all the above-mentioned dimensions and how they
interact (Siewierska 1988:29), with the aim of providing a more complete and coherent
account of word order phenomena. Specifically, it explores the four possible modalities
that determine the final sequence of elements in the sentence, namely grammatical
relations, constituent structure, argument structure, and information structure, as
well as how these modalities interact and impose constraints on possible word order
permutations. It does so by providing a new perspective to word order investigation;
the novelty of the approach lies on three aspects: (i) a typological, comparative
perspective that benefits from cross-linguistic investigation of WO phenomena in
other languages; (ii) a bottom up approach that adopts cross-linguistically validated
typological tools (e.g., grammatical relations-sensitive tests, or constituenthood tests)

2For a review of inhadequacies of accounts proposed within the main theoretical frameworks with respect to
control, coreference, zero anaphora, and reflexives in MC see for example Huang (1994).

9
aimed at conducting the analysis on a language-internal basis, and (iii) an
empirical approach: the analysis avails itself natural linguistic data, mainly drawn
from corpora, and relies on acceptability checks with native speakers. The next
sections provide a more detailed outline of the state of the art, the research objectives
and methodology of the present work, as well as its limitations. Investigating word
order in MC is a difficult task, not because of the lack of research on the topic, but
exactly for the opposite reason. MC word order has been a topic of major research for
linguists for decades. It is impossible to do justice to the wealth of the studies on this
matter, and therefore much pertinent work will be left unmentioned.

1.1. Overview and research objectives

As briefly outlined above, the present thesis looks at word order permutations in
Mandarin Chinese and seeks to determine (1) how and in what terms word order can
be described, and (2) how different components of linguistic organisation determine
the final sequence of constituents in a MC utterance/sentence, as well as how these
components interact in determining the availability of different word order patterns
and constructions. This section clarifies the motivations underlying these research
questions by providing a brief overview of the state of the art and highlighting several
controversies and issues with existing accounts of MC word order.

In order to effectively account for the sequence of elements and their arrangement in
the sentence, an adequate set of categories and notions needs to be adopted, which
capture how and at which level a specific word order permutation is motivated. In the
literature, MC is often described with notions pertaining to different levels of linguistic
organisation, including (i) subject and object (as an SVO language); (ii) topic and
comment (as a topic-prominent or discourse-oriented language); and (iii) iconic
principles (of temporal sequence, scope etc.). The following subsections briefly present
the most salient aspects and issues of such accounts.

10
(i) MC as an SVO language. In line with the Greenbergian tradition3 of word order
correlations, Mandarin Chinese is most often described with respect to the relative
order of the notions of subject and object, i.e. in terms of grammatical relations (GRs).
Specifically, the most widely accepted description among linguists is that the
unmarked order in MC is SVO.4 The characterisation of MC as SVO captures a number
of regularities that can be observed in the language. The MC equivalent of an English
transitive clause like (1) presents the same sequence of words as an English standard
transitive SVO sentence (adapted from Paul 2015:21):

1. ‘She has cleaned the room.’

他 打扫 房子 了。
tā dǎsǎo fángzi le
3SG sweep room CHG5

However, many scholars have noted that a characterisation of MC in Greenbergian

terms as a SVO language poses several problems, which we summarise below.

(i.i) The first inconsistency is observed in relation to Greenberg’s typology of word

order correlates. Such correlates postulate word order regularities with respect to the
basic transitive sentence order: if a language is SVO, it also displays head-initial
structures like auxiliary - verb, preposition - NP, noun - relative clause, verb – adverb,
intensifier – adjective, and so on. However, Standard Mandarin and other Sinitic
languages “present a perplexing case for syntactic typology since they display in

3 Greenberg (1966) proposed a typology based on the relative order of the subject, object and verb, resulting in a
six-way division of languages into: SVO, SOV, VSO, OVS, VOS and OSV, and identified a number of sub-regularities
have “which are taken to warrant expression in the concept of the 'word order type'”. (Siewierska 1988:8)
4A debate is found in the literature with respect to the proposal of a historical shift of MC towards an SOV language,
although most scholars now agree on SVO as being the basic, unmarked word order in MC (for discussion see, for
example, Paul 2015, Ch. 2, or Chappell et al. 2007).
5 This thesis differentiates between the post-verbal aspectual marker 了 le (PFV) and the sentence-final particle 了
le (CHG). The perfective aspect particle 了 le marks the perfective state of an action, indicating that the action is
completed, and is placed immediately after the verb. The sentence-final particle 了 le, on the other hand, is used
“to affirm the message and make the listener aware of its importance or relevance to the immediate situation” (Yip
Po-Ching and Don Rimmington 2004:318) and to “acknowledge some change in the picture of things.” (Tong and
Pollard 1982:142). For a critical analysis of the two different les, we refer the reader to Chappell (1988) or Sun
(1996).

11
general head-final characteristics for their NP structure but a mixture of head-initial
and head-final ordering for their VPs” (Chappell et al. 2007:2). This issue has been
extensively discussed in the literature: this thesis does not engage with this topic, and
only reports the table by Chappell et al. (2007:2) summarizing the inconsistences
displayed by Standard Mandarin with respect to the Greenbergian word order
correlates, referring the reader to Chappell et al. (2007) for discussion.

Table 1.1 - “The perplexing case of Mandarin: a typologically hybrid

language”

Head-final structures Head-initial structures

consonant with SOV order consonant with SVO order
Adjective – Noun Verb – Object
Numeral – Classifier – Noun Auxiliary – Verb
Demonstrative – Classifier – Noun Verb – Modifying adverbial
complements of manner, result
and degree
Relative Clause – Noun Preposition – NP
Genitive – Noun Complementiser – S
Adverb – Verb
Intensifier – Adjective
Standard of comparison – Adjective
Prepositional Phrase – Verb Exception: A subset
of locative constructions take the
form Verb – Prepositional Phrase

In short, MC and other Sinitic languages “pose somewhat of a challenge: they do not
conform to either of the two main alignments, as Dryer (2003), among others, has
observed” (Chappell et al. 2007:2).

(i.ii) The second issue relates to an assumption underlying the above account, i.e. that
linearisation is expressed in terms of grammatical (syntactic) relations such as subject
and object. However, grammatical relations (henceforth GRs) in Mandarin have been
a topic of heated debate in the last decades of the past century, partly due to the
absence of overt GR-specific morphological markers (e.g., case marking or agreement).

12
Although many scholars have dealt with this issue, the question of the definition of a
grammatical notion like subject remains unclear, and scholars hold divergent—and
sometimes opposite, views (see for example LaPolla 1990, 1993 and Bisang 2006).
Meanwhile, typological studies on a variety of previously under-described languages
have shed new light on the status and role of GRs crosslinguistically, highlighting a
language-specific and even construction-specific nature of GRs (see Bickel 2010 and
Witzlack-Makarevich and Bickel 2013). These findings provide a new perspectives and
approaches to the issue of GRs in MC, which can explain the terms of the debate and
the different accounts of GRs proposed by different scholars. The issue of GRs in MC
is the topic of Chapter 2.

(i.iii) MC sentence structure is often regarded as having a constituent structure that is

similar to that of English, namely comprising NPs – noun phrases, in some recent
development of the minimalist framework also referred to as DPs, and VPs – verb
phrases, i.e., constituents comprising the verb and the inner object(s). This account
can be expressed by the following representation, as well as by the features of what
Lambrecht (1987) calls SVO sentences.

2. Traditional constituent structure representation of the basic sentence structure.

3. Lambrecht (1987) features of SVO sentences:

a. A transitive clause with at least two arguments involved/expressed;
b. Within the unit, phrasal structures are construed as dependents of (or
dominated by) a larger structure—the sentence;
c. The unity held among constituents exists only between the verb and its
objects, as indicated by the VP (V NP) complex.

However, constituenthood tests reveal that the evidence for the existence of a VP in
many languages of the world is rather weak (Bresnan 2015). Moreover, corpus analysis
on MC conversation (Tao and Thompson 1994, Tao 1996) reveal that contrary to the
13
notion that the basic syntactic structure of a sentence comprises an NP and a VP (as in
the representation above), a Mandarin sentence in spoken discourse displays different
constituent patterns. According to Tao (1996), the most frequently occurring structure
is X+V, where X is a nominal/referential expression; among this structure type, in 42%
of the cases X is a patient/undergoer/theme argument (what would be described as
OV pattern). Moreover, corpus data show that “while transitives tend to reduce the
number of arguments that are fully specified, the majority of non-transitives sustain
the lexical coding of the one argument associated with them” (Tao 1996:19). Similar
statistical data on PAS (Preferred Argument Structure) in MC are observed in
narrative texts (see corpus research conducted by Lin 2009).

Comparing [2] with the major speech units (…), we can see immediately that there
are profound differences between the conceived syntactic units and actual speech
patterns. First of all, the idealized syntactic template, SVO, rarely appears in natural
discourse. As we have seen, full clauses with a transitive verb account for less than
3.2% of all the intonation units in our data, and this includes both high transitivity
and low transitivity (non-canonical, for most grammarians) clauses. If non-full
clauses are considered to be performance variations or even errors, as followers of
the competence-performance dichotomy might argue, we would encounter a
situation where the overwhelming majority of speech units are non-standard and
have little to do with what grammarians are describing in theory. Evidently, the call
for a syntactic framework which would enable the description of a language a little
bit closer to natural data is justified. (Tao 1996:180)

In other words, statistical corpus data suggest that the verb might establish similar
relationships with all its arguments, while the most frequent pattern is X-V, where X
is one of the verbs’ arguments. On the other hand, the representation in (2) involves a
closer relationship between the verb and its inner argument(s) than to its outer
argument (i.e. the subject of the sentence). To sum up, the constituent structure of MC,
although most often taken to parallel that of English, calls for a more-in-depth analysis.
This is the focus of Chapter 3.

(i.iv) Lastly, as Siewierska (1988:8) notes, the Greenbergian typological word order
evaluation involves “linearization patterns representing what is commonly referred to
as the ‘basic order’ [… which] is typically identified with the order that occurs in

14
stylistically neutral, independent, indicative clauses with full noun phrase (NP)
participants, where the subject is definite, agentive and human, the object is a definite
semantic patient, and the verb represents an action, not a state or an event”. In other
words, linearisation patterns refer to prototypical transitive clauses in the sense of
Hopper and Thompson (1980) just like (1). She further notes that “basic order” is often
equated with “dominant order” and implies statistical prevalence. However, as
mentioned above, statistical data from corpus studies on dominant orders in MC
reveal a different picture. The statistical relevance of high transitivity actor-action
sentences was first challenged by Chao (1968:70), who claimed that “in Chinese, the
proportion of applicability of the actor-action meanings is still very low, perhaps not
much higher than 50 percent”; in other words, the other 50% of sentences have
meaning other than actor and action. Since then, some statistical analyses have been
carried out with respect to basic sentence types in MC: as mentioned above, in Tao and
Thompson’s (1994) corpus study on Modern Chinese spontaneous conversation,
single argument clauses accounted for 61% of all sentence types; only 19% are double-
argument clauses, and 20% are clauses without any argument. In other words, the
Greenbergian SVO pattern accounts for one fifth of clauses in MC conversational data,
whereas the majority (61%) of transitive clauses in Mandarin conversations contain
only one overt argument, while only 19% transitive clauses have two overt arguments.
Similar results are provided in the analysis by Lin (2009) on PAS in narrative texts:
she observes that “clauses with zero or one lexical argument are common, whereas
clauses with two lexical arguments are rare”, regardless the verb valency (mono- or
transitive) and text type (conversation, narrative or written text, see discussion in
section above). Undoubtedly, a description of word order in MC should satisfactorily
account for this data. Therefore, the mapping between the argument structure of the
verb and the argument realisation patterns found in the language need further
investigation. Argument structure and argument realisations and alternations are the
focus of Chapter 4.

(ii) MC as a topic-prominent language. MC is typically quoted as an example of

topic-prominent, discourse-oriented, or topic-comment language (see Li 2005,
Chapter 3 for discussion), whereby the topic is the first element in the clause and is
followed by its comment. In his Grammar of Spoken Chinese (1968), Chao stated that
all clauses in MC are topic-comment structures (and specifically, that the meaning of

15
subject and predicate in MC is that of topic and comment). In other words, the
structure of the clause has two parts: the topic, which points to a specific referent, and
the comment, which supplies some information about that referent. Chao (1959)
argued that word order is not determined by and does not affect the interpretation of
actor vs. non-actor; he saw the clause as analogous to a function in logic: the argument
is an argument of the function, and the truth value is unaffected by its position in the
clause (1959:254). Lü Shuxiang (1979:72-73) also argued that “subject” and “object”
can both be filled by any semantic role and are to a certain extent interchangeable (i.e.
they are not syntactically constrained but are more loosely defined notions). Since Li
and Thompson’s (1976, 1981) typological distinction between subject-prominent and
topic-prominent languages, the sentence-initial position in MC has been associated
with the notion of topic, and with information structural properties such as givenness,
aboutness, definiteness, etc. (Chafe 1976). Scholars have offered very different
accounts with respect to how the notion of topic interacts with that of subject and with
the basic sentence SVO order: this issue has been the topic of heated debate in the
second half of the past century. Most scholars now agree that topics (or at least a sub-
portion of them) (i) are base-generated, i.e., are not the result of left-dislocation
processes, and (ii) do not need to bear selectional restrictions with respect to the
verb/predicative element. However, theories differ with respect to how topic is defined,
both in structural and in cognitive (information-structural) terms; moreover, no clear
consensus has been reached with respect to how the information structure component
of the language (topichood, givenness, contrastiveness etc.) interfaces with other levels
of the grammar. Chapter 5 is devoted to systematically exploring the information
structural component of MC grammar, with particular reference to the notions of topic,
comment and focus, as well as the interaction with other types of grammatical
constraints.

(iii) MC as an iconically motivated grammatical system. Several scholars have

identified and investigated some recurrent patterns in the sequence of elements in a
MC sentence, which have to do with the temporal or causal sequence or
semantic/temporal/spatial scope of elements and events in the sentences. Such
regularities are often captured through so-called word order principles (e.g., the
principle of temporal sequence and general-preceding-particular). This thesis will not
specifically engage with this topic. However, Chapter 5 will present and discuss the

16
tendency of MC to encode the whole before the part, the instantiations of this tendency,
its impact on word order constraints, and its interaction with the role of topic as a
frame setter.

1.2. Theoretical assumptions

Traditionally, word order is seen as tightly connected with the syntactic level of
linguistic organisation. The original meaning of the word syntax comes from the
Ancient Greek σύνταξις, syntaxis, meaning “putting together in order, arranging”; in
Matthews’ (1981:1) words, it studies how words “are arranged to show connections of
meaning within the sentence”. However, as Siewierska (1988:29) observes in her
seminal work Word Order Rules, “studies reveal that word order is dependent on an
array of syntactic, semantic, pragmatic and even phonological factors”: understanding
word order in a language involves accounting for “the diverse range of factors involved,
consider[ing] how they interact with each other, and determin[ing] to what extent this
interaction is constant” within a language. In fact, the past decades have witnessed a
growing interest in how other dimensions of linguistic organisation determine word
order in a language. These include: the constituent structure component, with studies
on configurational vs. non-configurational languages (see Baker 1997, King 1995); the
argument structure component, and the available argument realisations and
alternations – how arguments map into the final structure of the sentence (see Levin
and Rappaport Hovav 2005 for discussion); the information structure component, i.e.,
how the cognitive status of sentence elements (given vs. new etc.) determines the
position and relative order of elements in the sentence (see Güldemann et al. 2015 for
discussion).

Accordingly, this thesis aims to address what factors and functions shape word order
in MC, what levels of linguistic organisation they belong to, and how these levels
interact with respect to each other. The underlying theoretical assumption for this
approach is that grammar is composed of different levels (Danes 1966, Siewierska
1988, Lambrecht 1994), which are interdependent and interrelated. This assumption
is at the basis of theoretical approaches like LFG (Lexical Functional Grammar) or
RRG (Role and Reference Grammar). Such approaches conceive the structure of the
17
clause as a domain in which the different components of grammar—syntax,
morphology, prosody, semantics, and information structure, compete and interact
with each other, and are regulated by universal principles and language-specific
constraints. This interconnection is well explained by Lambrecht:

the most promising but perhaps also the most difficult approach to grammatical
analysis is one in which the different components of grammar are seen not as
hierarchically organized independent subsystems but as interdependent forces
competing with each other for the limited coding possibilities offered by the
structure of the sentence. I take a linguistic theory of high explanatory value to be
one in which these forces are not only analysed in isolation but also in their multiple
dependence relations to each other. In such a theory the grammatical structures
found in particular languages would then be seen as language-specific
manifestations of the interplay between the different components of grammar.
(Lambrecht 1994:10-12)

In line with the observations discussed above, the present work examines each of the
components of the grammar (semantic—verbs and their argument structure;
syntactic—GRs and constituent structure; and pragmatic—topic vs focus and
information structure). Moreover, it investigates the interface between such
components, to understand and describe how their interplay shapes the final sequence
of words and constituents in the sentence, and what constraints each component
applies to word order.

A tightly connected aspect concerns the formalisation of linearisation patterns and

constraints within different linguistic frameworks. Theories differ on how and to what
extent all these constraints and functions of word order are accounted for and deemed
to interact with each other within the overall system of the grammar. For example,
different frameworks rely on different means to integrate the syntactic structure (SVO
order) with the topic-comment structure. As Siewierska (1988:1-2) summarises, in
models of grammar which view order as an abstract underlying property of sentences
such as Chomsky's Transformational Grammar (TG) and Minimalist Programs,
“surface orderings are established via a number of rules: […] ordering is predictable
from the properties of words and their grouping relations must stipulate how
unordered strings of words are converted into well-formed sentences/utterances”. On
18
the other hand, “in models of grammar that adopt a multi-level approach […] the
linearisation rules may in principle be distributed over the whole derivation of a
sentence, or be confined to one, two or more levels.” This thesis will not deal with the
problem of formal representation of linearisation rules within different theoretical
frameworks, and evaluation of the suitability of different frameworks is beyond the
scope of this work. However, some observations will be made in the concluding
remarks as to what a linguistic framework should look like to account for MC in light
of the analysis in the present thesis. With respect to the formal representation of word
order patterns, constituent structure, argument structure and argument realisations,
general representational conventions are adopted; the logical structure of verbs is in
turn represented using the Role and Reference Grammar (RRG) conventions.

1.3. Methodology and linguistic data

The aim of this study is to investigate each level of the grammar on a strictly language-
internal basis, to “capture all of the features of a language without imposing features a
language shows no evidence for” (Van Valin and LaPolla 1997:22). The importance of
such an approach has received increasing attention by linguists, especially typologists
researching typologically diverse languages (sometimes regarded as ‘exotic
languages’). The underlying shared insight is that, in fact, it is empirical
generalisations that result from research on meaning across typologically diverse
languages that provide the basis for cross-linguistically viable theories of meaning.
Under this position, adopted in this thesis, one of the desiderata for a cross-
linguistically viable theory is that it captures empirical generalisations and the
convergence in meaning without assimilating the morphology and syntax of one
language to that of another.

The key challenge for typological metalanguages is that they need to be able to
describe structures across languages in an empirically responsible way, i.e. without
forcing languages into Procrustean beds. (Bickel 2015:6)

Accordingly, the study adopts a bottom up, typological approach, and avails itself of (i)
cross-linguistically validated typological tools (e.g., GR tests, or constituenthood tests);

19
(ii) natural linguistic data – mainly drawn from corpora; (iii) acceptability checks with
native speakers for examples quoted in the literature (online survey). These three
components will be briefly presented below.

Cross-linguistically validated tests. In line with the approach described above,

the present work seeks to motivate the use of linguistic categories and notions on the
basis of linguistic evidence; accordingly, it employs an array of typological tools and
tests, which help grounding the analysis on a more solid empirical basis and avoiding
imposing theory-internal assumptions. This is the case in Chapter 2, which employs
an array of tests based on GR-sensitive phenomena and constructions to establish
what notions are suitable for the description of patterns and constructions in MC, with
a specific focus on GRs like that of subject and object. The same approach is adopted
in Chapter 3, where standard constituenthood tests are employed to determine the
constituent structure in MC and what hierarchical relationships hold among
constituents.

Natural language data. The analysis avails itself of natural linguistic data of
different types, comprising both narrative and conversational texts, and covering
different genres and topics. These include:

(i) sentences drawn from corpora of natural linguistic data/dictionaries, such as: (I)
the PKU corpus of Modern and Classical Chinese, hosted by the Centre For Chinese
Linguistics, Peking University (http://ccl.pku.edu.cn); (II) Lü Shuxiang’s (1980) 现代
汉语八百词词典 (800 Words in Contemporary Chinese); (III)in-print and on-line
dictionaries.

(ii) Corpora of spontaneous language production. These include: (I) Tao and
Thompson’s (1994) corpus of spontaneous conversation interaction, i.e.,
transcriptions of twelve ordinary audio-recorded naturally-occurring interactional
conversations among native speakers of Mandarin; topics include everyday-life
experiences, education, traveling (for further details, see Tao 1996:29-30). (II) Ho’s
(1993) corpus of narrative/expositional text, i.e., transcriptions of interviews with over
20 Mandarin speaking informants. The native speakers comprised students and
teachers, social workers, restaurant workers and religious personnel among others;

20
the interviews represent a variety of genres and discourse types (i.e., narrative,
exposition, and procedural (for further details, see Ho 1993:14-15).

(iii) Transcriptions of interviews and dialogues collected by the author. These include
transcriptions of 7 videos containing interviews of contemporary Chinese artists
transcribed by the author (hereafter referred to as ART VIDEO): all artists are MC
native speakers from different parts of China and talk about their artistic production
and the works displayed in a Chinese art exhibition in 2017 in Vicenza, Italy.

Acceptability check with native speakers. When presenting the different

phenomena and discussing them against the background of the reference literature,
examples used by the various scholars are reported. However, since native speakers
perceive many examples in the literature as artificial or created ad hoc, each such
example has been checked against their intuition. Specifically, examples quoted in the
literature in Chapter 2 have been submitted to a group of 37 MC native speakers. For
most sentences, a scale of acceptability was proposed, with a value of acceptability
from (1 to 10). Whenever needed, relevant context was provided. The statistical
approach adopted in the design of the test captures the variation in acceptability rates
among different individuals and with different contexts provided (especially for
different word order patterns). This is tightly connected to the role played by context
and world knowledge in the interpretation and disambiguation processes by MC native
speakers when decoding a message. Different scholars have highlighted the role of
context in sentence disambiguation. The impact of context in acceptability judgement
was also emphasised by Fan and Kuno (2013:220-4): given the same sentence, “[e]ven
the same speaker might judge it sometimes acceptable, and other times marginal or
awkward. This must be due to the differences among speakers in their ability to place
the sentence in contexts”: this factor is considered when discussing statistical data.

1.4. Structure of the thesis

The present thesis is structured as follows:

Chapter 2 looks at the grammatical relations (GRs) component and seeks evidence
that justifies their employment in the description of the language. The adopted
21
methodology draws from the typological approach developed by Witzlack-Makarevich
and Bickel (2013), which provide a framework to investigate language-specific
grammatical relations in various (possibly typologically different) languages. This
approach involves examination of a range of GR-sensitive constructions, which
include: relativisation; reflexivisation; passivisation; topic extraction; equi-NP
deletion; floating; finiteness; control; raising.

Chapter 3 examines the constituent structure component in MC. By adopting standard

constituenthood tests, as discussed in approaches like Siewierska (1988) and Pavey
(2010), it seeks to establish whether there is clear-cut evidence for the existence of
constituents such as the NP (noun phrase) and the VP (verb phrase). Constituenthood
tests employed in the analysis include indivisibility (or uninterruptability), fixed order,
replaceability as a whole (or substitution), required elements (or omissibility),
movement (or distribution), coordination.

Chapter 4 explores the argument structure component: specifically, it is devoted to a

preliminary investigation of predicating elements (mainly verbs and classes of verbs)
their argument structure, along with how these arguments map into the sentence.
Adopting an approach similar to Levin (1993) for English verbs, the chapter presents
salient aspects of a preliminary qualitative corpus analysis of a range of verb classes in
MC and argument realisation patterns. Specific attention is given to patterns involving
aspectual and causal shifts, which typically affect word order. Examined verb classes
include: verbs denoting states, conditions and properties, verbs of psychological states,
verbs of existence, appearance, and disappearance, verbs of involuntary activities,
verbs of posture, verbs of perception, verbs of motion, verbs of action on objects, verbs
of measure, verbs of change of possession.

Chapter 5 looks at the Information Structure (IS) component of the language, and
specifically the possible permutations of the order of elements in the MC sentence. IS
notions generally associated to different positions in the sentence, such as that of topic,
comment and focus are critically examined with respect to their definition, function,
cognitive and information status, and restrictions. Subsections are devoted to further
exploring the semantic relation between the topic and its comment, and an account is
proposed in terms of Pustejovsky’s Qualia Structure (1991, 1998). Finally, the

22
interaction between the IS component and other components of the grammar with
respect to word order is explored, and an account in terms of word order freezing
(Mohanan and Mohanan 1994) is proposed to capture some word order restriction
phenomena displayed by MC.

Chapter 6 draws the conclusions for the present study, reviews its limitations and
proposes areas of further research.

1.5. Limitations of the study

A very plausible limitation of this study is that its scope is very broad. Mandarin
Chinese has been a major case of study for the refinement of theories of grammars and
linguistic categories over the past 70 years. It is impossible to do justice to the wealth
of the studies on MC grammar, and therefore much pertinent work is left unmentioned.
Providing a thorough literature review is a challenging task, given the vast amount of
research in the topics this thesis touches upon, including grammatical constructions,
topic-comment and information structure, argument realisations and alternations,
and constituency. Hence, when confronted with the necessary choice among the
various studies with regard to a specific issue, only few of which could be covered given
the available space, we chose the more influential and pertinent to the discussion.
Moreover, each chapter is self-contained and examines a specific issue: thus, a brief
literature review is proposed for each chapter, which is specifically relevant to the topic
under discussion. References to other related studies are also mentioned, when
relevant.

Moreover, while the focus of the thesis is word order and sentence structure, the
present study does not (or does only marginally) engage with phrase-internal order,
which however constitutes a very interesting research domain, in that it exhibits some
features that parallel the structure of the sentence (e.g., the whole-part or frame-
setting order that characterises topics – discussed in Chapter 5, can be also observed
in NP-internal modifiers, see Kirkpatrick and Xu 2012). In addition, this thesis does
not discuss word order with respect to the textual/discourse level, which again is a very
interesting research domain, but which would require much a wider discussion.

23
View publication stats

With respect to Chapter 4 on verb argument structure and mapping, the proposed
analysis avails itself of very limited data, as compared for example to the work done by
Levin (1993) on verb classes in English. A thorough, statistically valid analysis would
require examination of a much broader range of verbs and verb classes, as well as a
more thorough account of available argument realisations and alternations. While
definitely interesting as a research domain, this type of analysis necessarily lies beyond
the scope of the present thesis and calls for further investigation.

Finally, the encompassing and synthetic perspective adopted in this work constitutes
part of its limitation. Because of the broad scope and overall perspective this thesis
aims to adopt, the study is unable to cover all arguments with the necessary depth.
However, this limitation is in turn seen as the potential of the present work: this thesis
aims at laying a basis, highlighting issues, and problems with current theories, onto
which further research lines can be developed.

24