Available online at www.sciencedirect.

com

ScienceDirect
Lingua 311 (2024) 103814
www.elsevier.com/locate/lingua

Potential objects and transitivity variations: A comparable

corpus-driven study of Mandarin Chinese Verb-Object
compounds
Menghan Jiang a,⇑, Chu-Ren Huang b,c,⇑
a
Chinese Language Centre, Shenzhen MSU-BIT University, Shenzhen, China
b
Department of Chinese and Bilingual Studies, The Hong Kong Polytechnic University, Hong Kong, China
c
HK PolyU-PekingU Research Centre on Chinese Linguistics, Hong Kong, China

Received 10 December 2023; revised 3 September 2024; accepted in revised form 3 September 2024;

Abstract

This study proposes a new approach to the transitivity variation of the Mandarin Verb-Object (VO) compound by
focusing on the role of the potential objects instead of the incorporated object. This new approach highlights the inter-
actions between language change and variation. Previous approaches to the transitivity variation of VO compounds,
especially Mandarin Chinese VO compounds, have typically focused on the purported blocking effect of the incorpo-
rated object (O). Recent corpus-driven studies, however, have attested that Mandarin VO compounds can take objects.
This study compared clausal contexts in two Chinese varieties when VO compounds take direct objects and showed
that the same noun phrase tends to require a context of higher transitivity to function as a direct object in the Mainland
variety, complementing the previously established tendencies that Taiwan Verb-Object compounds have higher object-
taking abilities. Both changes and variations can be interpreted as lexical constructionalization, supplemented with
insights from noun incorporation and lexical diffusion. Argument advancement after compound formation represents lex-
ical constructionalization, motivated by a tendency to encode transitivity with less compositionality. The contrasting con-
textual and lexical transitivity trends in the two varieties support the shifting encoding of transitivity generalization,
illustrating different stages of lexical constructionalization.
Ó 2024 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar
technologies.

Keywords: Language Variations; Transitivity of Mandarin Chinese Verb-Object Compound; Comparable Corpus-Driven
Approach; Noun Incorporation; Lexical Diffusion Theory; Lexical Constructionalization

⇑ Corresponding authors at: Chinese Language Centre, Shenzhen MSU-BIT University, No. 1 International University Park Road,
Dayun New Town, Longgang District, Shenzhen, Guangdong Province, PR China (Menghan Jiang) and GH513, Core A, The Hong
Kong Polytechnic University, Hung Hom, Hong Kong SAR, China (Chu-Ren Huang).
E-mail addresses: [email protected] (M. Jiang), [email protected] (C.-R. Huang).

https://doi.org/10.1016/j.lingua.2024.103814
0016-7037/Ó 2024 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
2 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

1. INTRODUCTION

1.1. Verb-object compounds

The Verb-Object compound (hereafter ‘VO compound’) in Mandarin Chinese consists of two constituents with a syn-
tactic/grammatical relation of a verb and its direct object (Li and Thompson, 1981), such as 投资 touzi invest-capital ‘to
invest’) and 签约 qianyue sign-contract ‘to sign a contract’. Previous studies (e.g., Huang, 1984; Li, 2012) proposed that
because the verb in a VO compound already assigns case/theta roles to its object—such as 投 tou ‘invest’ and 签 qian
‘sign’ taking the objects 资 zi ‘capital’ and 约 yue ‘contract’—the compound cannot take an additional direct object. This
theory aligns with the observation that many of these compounds can take an object-like argument in other oblique posi-
tions, such as an object in a preverbal prepositional phrase (hereafter ‘PP’). This is called the pseudo-transitive con-
struction in the literature (e.g., Chang et al., 1990).
The term ‘pseudo-transitive’ indicates lower transitivity for this kind of construction (i.e., compounds taking object-like
arguments in oblique positions). Recent studies have observed that, although their numbers are relatively small to begin
with, an increasing number of VO compounds can now take another external object and yield the VO1 + O2 construction
(e.g., Diao, 1998; Wang, 1997). Attested examples include 投资房地产 touzi fangdichan invest-real estate ‘invest in real
estate’ and 签约唱片公司 qianyue changpian gongsi sign contract-record-company ‘sign contract with record company’.
The increase in transitivity of Mandarin VO compounds, especially the ability to take direct objects, poses a challenge
to many current syntactic accounts, particularly those that assume abstract case (e.g., Qin and Duanmu, 2017). Tradi-
tional studies explored why VO compounds, despite taking two arguments, typically do not allow direct object positions
(Huang, 1984; Li, 2012). More recent approaches have considered lexical semantics and the argument structure,
accounting for the non-canonical realization of the second argument (e.g., Huang, 1985, 1990; Her, 1996). In addition,
morpho-syntactic studies on compound verb formation have typically focused on the properties of the incorporate object
(O), and had paid little attention to the direct objects taken by the VO compounds(e.g., Ling, 1999; Liu, 1993; Rao, 1984;
Yang, 2001; Zhu and Sheng, 2008).

1.2. Variations of world Chinese and transitivity variations

Previous studies on the transitivity of Mandarin VO compounds generally adopted diachronic change perspectives,
e.g., questioning whether these compounds were becoming more or less transitive and if pseudo-transitive usages were
intermediate stages of transitivity change, etc. However, recent studies have shown that the object-taking ability of VO
compounds varies across different Mandarin varieties (Jiang and Huang, 2018, 2022). Few studies have integrated both
change and variation perspectives.
Mandarin and Taiwan Mandarin are the most frequently studied varieties of World Chinese or Modern Mandarin for
two reasons: 1) they are two of the most important varieties of Mandarin Chinese, in terms of both the population of
native speakers and second language learners; 2) their 38-year isolation from each other and ongoing socio-political
differences create fertile ground for variation and distinct usage contexts (Huang et al., 2014). It has been observed that
earlier studies on Mainland-Taiwan variations focused on lexical differences (e.g., Diao, 1998; Wang, 1997), relying
mainly on researchers’ introspection and observations from selected sources such as newspapers or novels. The avail-
ability of comparable corpora and advanced computational/statistical analytical tools now enables more extensive and
comprehensive investigations, uncovering subtle grammatical variations in syntactic behaviors that involve preferences
of specific syntactic or semantic selections, such as argument types or polarity (e.g., Lin et al., 2019; Hou et al., 2022).
Motivated by the theoretical construal of languages as complex self-adaptive systems (Beckner et al., 2009), Xu et al.
(2022) applied machine learning methodology to the study of light verb variations between Mainland and Taiwan
Mandarin.

1.3. Language variations and language Changes: A theoretical perspective

Studies on language variation and change seem to be orthogonal by nature: variation studies focus on synchronic
contrasts, such as differences that are geographical and demographically relevant. Studies of language change, on the
other hand, emphasize diachronic contrasts, particularly differences in the same language over time. However, both
perspectives have raised questions about the interaction between variations and changes. For instance, are language
variations the results of differences in language changes in different contexts, or do language changes aggregate vari-
ations or result from contact with other varieties?
 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814 3

By putting language changes and variations within a temporospatial continuum and taking the languages as complex
systems approach, both changes and variations can be seen as self-organizing adaptations recorded from either a tem-
poral or spatial perspective. This can be accounted for by some recent theories, for instance, the variation in transitivity
can be understood as lexical constructionalization (Traugott and Trousdale, 2013, 2014), which accounts for language
variation as usage-based changes. Lexical constructionalization examines how the meanings of words interact with the
grammatical patterns they occur in through time, leading to the emergence and development of new constructions in a
language. Through lexical constructionalization, new constructions can emerge as words adapt to fit into existing gram-
matical patterns. This process involves regularizing and conventionalizing the association between a word and a speci-
fic grammatical context. For example, increased use of a verb in a particular syntactic frame may result in a new
construction. The phenomenon of VO compounds increasing in transitivity exemplifies the reduction in compositionality,
as discussed by Traugott and Trousdale (2013, 2014). Accordingly, higher transitivity indicates a higher degree of lex-
ical constructionalization, with the gradient of transitivity reflecting the ongoing process of constructionalization. Thus,
the variation in transitivity between Mainland and Taiwan may indicate their different status in the process of lexical
constructionalization.
Noun incorporation also views transitivity variation through a complex systems lens. This theory (e.g., Baker, 1988;
Mithun, 1984, 1986) offers an encompassing framework for verbal compound formation. That is, the noun O1 can be
incorporated with the verb stem V, forming a compound verb VO1. With its new lexical status, the incorporated verb
can then be either transitive or intransitive depending on its predicate-argument structure. A transitive VO1 yields the
VO1 + O2 construction. From a diachronic change perspective, the speed of the application of morpho-lexical rules
is well known to vary among different lexical items. However, when considering language variation, interesting issues
arise, such as if two varieties undergo incorporation in parallel and exhibit the same degrees of transitivity across the
lexicons. Would one variety spearhead the change and consistently show a higher degree of incorporation over the
other variety? Given the languages as complex systems view, the various generalizations of noun incorporation as
either language changes or variations can be interpreted as synchronic or diachronic projections of the complex inter-
actions in the temporospatial continuum. This theoretical perspective would also project lexical diffusion (e.g., Wang,
1969, 1977, 1979) in the temporospatial continuum, potentially leading to either some type of change or some type
of variation.
Hou et al. (2022) empirically attested the view that language changes and variations can be modeled uniformly in the
temporospatial continuum. Their study on longitudinal data of Hong Kong and Mainland Mandarin showed that, with
several grammatical features, when mapped on the two-dimensional time–space continuum, neither changes nor vari-
ations were linear. Instead, what remained constant throughout the process were the distances between the two vari-
eties at any given time and the relative distance between the same variety at any two consecutive temporal points. The
study also demonstrated that the same set of grammatical features could capture both changes and variations.

2. METHOD

2.1. Research questions

Transitivity has long been a critical issue in linguistics (Babby, 1975; Hopper and Thompson, 1980; Lee, 2010;
Voigtlander and Bartholomew, 1972, among others). Traditionally, transitivity is defined at the lexical level, as a property
of verbs that determines whether a verb can take objects or not (Robins, 1964; Richards et al., 1985). As a result, verbs
have often been categorized into transitive verbs, capable of taking a direct object (e.g., ‘kick’), and intransitive verbs,
incapable of taking an object (e.g., ‘cry’). Hopper and Thompson (1980) expanded this concept with a usage-based the-
ory, viewing transitivity as a clausal continuum measurable by various factors, assigning each verb a transitivity value.
Adopting the traditional definition of transitivity, Jiang and Huang (2018, 2022) have examined the transitivity fre-
quencies of VO compounds in Mainland and Taiwan Mandarin, comparing the percentage of tokens in which each
VO compound takes objects. Their findings revealed a notable disparity: Taiwan VO compounds exhibited a higher
propensity for transitive usage, with a mean transitivity value of 0.3381, compared to 0.284 in Mainland Mandarin. A
Z-test confirmed that Taiwan VO compounds were significantly more inclined toward transitive usages. Moreover, they
argued that this frequency variation suggested different developmental stages at which the two varieties were located in
the process of transitivization, with Taiwan VOs transiting from intransitive to transitive faster than their Mainland coun-
terparts. In this study, we aim to explore transitivity variations from an alternate perspective: by investigating the types of
objects taken by VO compounds in both variants, following Hopper and Thompson’s (1980) approach. That is, we
examine the contextual conditions that facilitate the object-taking ability of VO compounds in Mainland and Taiwan (re-
gardless of individual verb transitivity). Through the extraction and meticulous examination of all post-compound objects
and their distributional differences in terms of semantic and syntactic features, this study has two objectives:
4 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

1) To find out if there are any differences in the grammatical contexts for VO compounds taking direct objects in Tai-
wan and Mainland Mandarin, especially in terms of transitivity.
2) To model such variations, if any, and explore the possible correlation between contextual variation and the tran-
sitivity variation of VO compound verbs.

Based on the outcomes of these two objectives, we address the following two research questions:

RQ1: Can the variations in the direct object-taking abilities of Mandarin VO compounds be attributed to general ten-
dencies of language change?
RQ2: If so, can the general tendencies and regional variations be accounted for in terms of noun incorporation, lex-
ical diffusion, or lexical constructionalization?

2.2. Data collection

The main corpus utilized is the Annotated Chinese Gigaword Corpus (Huang, 2009), the largest available annotated
corpus covering both varieties. The corpus contains over 1.1 billion characters, including more than 700 million charac-
ters from the Taiwan Central News Agency and 400 million characters from Mainland Xinhua News Agency. A total of
171 VO compounds were selected based on previous surveys (e.g., Luo, 1998; Qian, 2011). The list was further pruned
to ensure that each VO compound was attested in Taiwan and Mainland Mandarin for valid comparison. Sixteen (16)
VO compounds, such as 叫板儿 jiaobanr ‘to challenge’ and 试水 shishui ‘to try’, were only attested in the Mainland cor-
pus and were excluded. The remaining 155 VO compounds served as the basis for data extraction and comparison.
For each VO compound, 1,000 tokens were extracted from each variety of the corpus.1 We further excluded 46
words used intransitively in both corpora.2 In total, for the 109 VO compounds, over 580,000 tokens were extracted from
both Mainland and Taiwan corpora. We manually confirmed all the transitive usages for each VO compound, resulting in
over 280,000 transitive tokens for all the VO compounds in both varieties, including 112,870 tokens from Mainland and
175,579 tokens from Taiwan.
For the 109 VO compounds well-attested with transitive uses in both varieties, a maximum of 200 tokens of transitive
uses were randomly collected for each verb. In cases where the total transitive uses were less than 200, all examples
were sampled for the VO compounds. The selection followed the principle that it should cover different uses of each VO
compound. At this stage, a total of 13,206 examples (5,832 from the Mainland corpus and 7,374 from the Taiwan cor-
pus) were selected for analysis.

2.3. Data annotation

In this study, we focused on transitivity variations between Taiwan and Mainland Mandarin by scrutinizing the dis-
tinctive distributional patterns concerning the semantic and syntactic features of the direct objects associated with
VO compounds. We annotated each VO compound with nine features, which were selected based on Hopper and
Thompson’s parameters and previous studies (e.g., Bresnan et al., 2007; Gao, 1998; Luo, 1998; Masolo et al.,
2002). The features covered semantic (e.g., the animacy of the taken object), syntactic (e.g., part of speech of the
object), and discourse levels (e.g., whether it appears in a parallel structure). The detailed annotation schema is shown
in Appendix A. The main annotator was a trained expert in Chinese linguistics. In instances of ambiguity, two additional
experts were consulted to ensure quality and reach a consensus on annotations when necessary.

3. TRANSITIVITY VARIATIONS OF VO COMPOUNDS BETWEEN MAINLAND AND TAIWAN MANDARIN

3.1. Overview of annotated data

An overview of the annotated data is shown in Table 1, including the number of tokens and percentage for each fea-
ture in each variety. We then conducted a model-based analysis (i.e., Mixed-effect Logistic Regression Model) and sup-
plemented it with an inferential analysis (i.e., Chi-square test) based on these data to explore variation differences in VO
compounds.

1
For a VO compound with fewer than 1,000 tokens in the corpus, all available tokens were included.
2
Given the 1 billion word scale of the Chinese Gigaword Corpus, a small number of exceptional uses are attested. We set the
threshold for extraction at three to ensure data quality.
 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814 5

Table 1
The descriptive results for both varieties.
Feature Mainland Taiwan
(Number of tokens
/Percentage)
POS: the syntactic type of the object Common noun or noun phrase (NP) 4719 5695
0.809 0.772
Deverbal noun 215 325
0.037 0.044
Event noun 227 135
0.039 0.018
VP 266 564
0.046 0.076
Clause 157 311
0.027 0.042
Quantity 103 132
0.018 0.018
Time 145 212
0.025 0.029
Headline No 4808 6539
0.824 0.887
Yes 1024 835
0.176 0.113
Complexity of the object No 3233 4151
0.554 0.563
Yes 2599 3223
0.446 0.437
Aspectual marker No 5594 7295
0.959 0.989
Yes 238 79
0.041 0.011
Polarity Negative 681 1024
0.117 0.139
Neutral 5082 6109
0.871 0.828
Positive 69 241
0.012 0.033
Proper noun No 3822 4963
0.655 0.673
Yes 2010 2411
0.345 0.327
Animate No 4800 6247
0.823 0.847
Yes 1032 1127
0.177 0.153
Pronominality No 5649 7177
0.987 0.973
Yes 183 197
0.03 0.027
Structural parallelism No 5758 7274
0.987 0.986
Yes 74 100
0.013 0.014
6 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

Table 2
The results of the logistic regression model.
Estimate Std. Error z value Pr(>|z|)
(Intercept) 1.127 0.2323 4.851 1.23e-06 ***
Polarity negative 9.807 934.176 0.010 0.991624
Polarity neutral 0.016 0.132 0.116 0.907553
Polarity positive 1.566 0.290 5.394 6.90e-08 ***
POS clause 14.661 998.369 0.015 0.988283
POS deverbal noun 0.492 0.192 2.572 0.010105 *
POS event noun 1.708 0.265 6.437 1.22e-10 ***
POS NP 0.615 0.172 3.576 0.000350 ***
POS quantity 1.111 0.244 4.552 5.30e-06 ***
POS time 0.734 0.236 3.113 0.001853 **
POS VP 0.097 0.191 0.508 0.611553
Proper_noun yes 0.168 0.055 3.025 0.002490 **
Animate yes 0.082 0.074 1.108 0.267651
Pronominality yes 0.388 0.117 3.311 0.000931 ***
Structual_paralism yes 0.026 0.172 0.153 0.878780
Headline yes 0.646 0.060 10.861 < 2e-16 ***
With_modifier yes 0.024 0.046 0.516 0.605678
Aspectual_marker yes 1.435 0.153 9.380 < 2e-16 ***

3.2. Feature-based modeling

We first applied a mixed-effect logistic regression model (GLMM) (Pinheiro and Bates, 2000), implemented in the R
package lme4 (Bates et al., 2015; Bates and Maechler, 2009).3 The model specifications are as below:
Formula: ‘source
polarity + POS + proper noun + pronominality + structural parallelism + headline
+ animacy + complexity + aspectual marker + (1 j item)’
In our model for the VO compound, the random-effect factor ‘item’ represented the different verbs in the V position.
The dependent variable was ‘source’, denoting the Mandarin variety (Taiwan or Mainland), while the independent vari-
ables included the nine annotated features. The features were polarity of the context (‘polarity’), part of speech of the
object (‘POS’), whether the object was a proper noun or not (‘proper noun’), whether the object was a pronoun or not
(‘pronominality’), whether the VO compound appeared in structural parallelism (‘structural parallelism’), whether the VO
compound appeared in the headline (‘headline’), whether the object referred to an animate or inanimate entity (‘ani-
macy’), whether the object had a modifier or not (‘complexity’), and whether the VO compound took an aspectual marker
or not (‘aspectual marker’). We excluded 10 words predominantly found in one variety and focused on 99 items for mod-
eling. To ensure sufficient data for simultaneous comparisons, most variables were rendered as binary values: yes and
no. The results of the regression model are shown in Table 2.
We conducted Chi-square tests to address random effect variance in the logistic regression models, representing
variations among individual words across different variants. Each VO compound was analyzed using a Chi-square test
to assess the significance of the co-occurrence of each factor with each variety using IBM SPSS v. 22. Variables were
considered significantly different at p < 0.01 (with Bonferroni correction). The logistic regression model and the Chi-
square tests demonstrated closely aligned results as expected but also revealed some significant contrasts. For exam-
ple, Taiwan’s preference for taking VP and clause as objects, which, though not significant in regression analysis, was
observed descriptively and in Chi-square tests. Among all VOs, 33 words demonstrated significant variation in the syn-
tactic type of the object taken by the VO compound. Among these 33 words, 15 words (e.g., 投身 toushen ‘plunge into’,
插手 chashou ‘intervene’) revealed Taiwan’s tendency to take event-denoting objects (VP and clause), while the Main-
land counterpart preferred common NP objects, aligning with findings on light verb variations in Xu et al. (2022). Con-
versely, only 4 words (e.g., 立足 lizu ‘base upon’, 抱怨 baoyuan ‘complain’) exhibited the opposite tendency. Another
contrast involves object complexity. According to the Chi-square tests, VOs in Taiwan were more likely to take complex
objects.

3
This model was chosen to accommodate the inclusion of various verbs in our dataset, with their variability being a crucial factor.
Consequently, the model treats the verb as a random variable to account for this variability.
 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814 7

Table 3
The statistical results of both models.
Feature Mainland Taiwan
POS: the syntactic type of the object Common noun or NP + 
Deverbal noun + 
Event noun + 
VP  +
Clause  +
Quantity + 
Time + 
Proper noun + 
Pronominality + 
Headline + 
Complexity of the object  +
Aspectual marker + 
Polarity (positive)  +

We combined both models and presented the results in Table 1. Among all the nine features included in the model,
seven exhibited significant differences between the two varieties in both statistical tests. In Table 3, a “+” stands for sig-
nificant overuses of the factor, while a “-” stands for significant underuses of the factor.
The ‘POS’ feature, which represented the grammatical categories of the object, showed significance in both tests. In
particular, ‘event noun’ and ‘common NP’ were preferred in Mainland Mandarin. Mainland VO compounds were more
likely to take event nouns and common NPs as objects, as illustrated in examples in 1)-2). In addition, quantity objects
(as in 3)) and time objects (as in 4)) were also more prevalent in the Mainland corpus.

1) 我们 的 工作 是 观察 而 不是 插手 比赛。
women de ɡonɡzuo shi ɡuɑnchɑ er bushi chɑshou bisɑi
We DE job is observe but not interfere game.
chɑ:insert; shou:hand
‘Our job is to observe rather than interfere with the game.’
--- Mainland data

2) 进一步 插手 台湾 问题
jinyibu chɑshou Tɑiwɑn wenti
further interfere Taiwan issue
‘to further intervene in Taiwan’s issues’
---- Mainland data

3) 投资 约 2亿 元
touzi yue 2 yi yuɑn
invest approximately 2 billion yuan
tou:invest
zi:caption
‘investing approximately 200 million yuan’
---- Mainland data
8 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

4) 贝克汉姆 同 曼联队 续约 5 年。
Beikehanmu tong Manliandui xuyue 5 nian
Beckham with Manchester United renew contract five years.
xu: renew
yue: contract
‘Beckham renews his contract with Manchester United for 5 years.’
---- Mainland data

In contrast, Taiwan VO compounds tended to take Verbal Phrases as objects, a pattern rarely observed in Mainland
Mandarin and potentially considered ungrammatical by native speakers. This tendency is illustrated in examples 5)-10):

5) 警察 插手 经营 约堡 市中心 区 低级 旅馆。
Jingcha chɑshou jingying Yuebao shi zhongxin qu diji lüguan
Police interfere manage Fort City downtown area low-level hotels.
‘The police intervening in the management of low-level hotels in the downtown area of Fort City.’
---- Taiwan data

6) 已 游说 日本 接手 举办 这项 比赛
yi youshui Riben jieshou juban zhe xiang bisai
have lobbied for Japan take over organize this competition
jie: take
shou: hand
‘has lobbied for Japan to take over the organization of this competition’
---- Taiwan data

7) 希望 政府 能 接手 处理 后续 问题
xiwang zhengfu neng jieshou chuli houxu wenti
hope government can take over handle subsequent problem
‘hoping that government could take over and handle the subsequent issues.’
---- Taiwan data

8) 建议 日本 插手 斡旋 高棉 问题
jianyi Riben chashou woxuan Gaomian wenti
suggest Japan intervene mediate Cambodian issue
‘suggested Japan to intervene in the Cambodian issue’
---- Taiwan data

9) 中意 前往 波兰 投资
zhongyi qianwang Bolan touzi
prefer to head to Poland invest
zhong: meet
yi: wish
‘prefer to invest in Poland’
---- Taiwan data
 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814 9

10) 寄望 争取 更多 的 选票 支持
jiwang zhengqu gengduo de xuanpiao zhichi
hope strive for more DE vote support
ji: place hope
wang: hope
‘hoping to gain more support from voters’
---- Taiwan data

As demonstrated in the preceding sections, the syntactic nature of the objects taken by VO compounds can be sum-
marized as follows: Taiwan VO compounds predominantly select event-denoting objects, whereas those in Mainland
China prefer common NPs as objects.
The ‘headline’ feature also exhibited significant overuse in Mainland Mandarin, suggesting a preference for VO com-
pounds to appear in headlines, unlike their Taiwan counterparts.
The results regarding proper nouns indicated that Mainland VO compounds preferred to take proper nouns as
objects, such as 任职多伦多大学 renzhi Duolunduo Daxue ‘serving at the University of Toronto’.
In addition, the results also demonstrated a preference for taking pronouns as objects, such as 抱怨他们 baoyuan
tamen ‘complain about them’.
The results concerning aspectual markers suggested a greater tendency for Mainland VO compounds to associate
with markers such as 了 LE (perfective marker), 着 ZHE (imperfective marker), and 过 GUO (experiential marker). For
instance, in Example 11), 失信 shixin in Taiwan appeared in 已经失信一次 yijing shixin yici ‘already lose trust for one
time’ without the aspectual marker GUO, while Mainland speakers tended to use GUO in similar contexts.

11) 他 已经 失信 一次。
Ta yijing shixin yici
He already break trust once.
shi: lose
xin: trust
‘He has already broken promise once.’
---- Taiwan data

We have also explored the complexity of the object by examining the presence or absence of modifiers. Although not
significant in the regression model, chi-square tests revealed that VO compounds in Taiwan Mandarin tended to occur
with complex objects. Furthermore, the presence of intricate modifiers in some constructions, as illustrated in Examples
(12) and (13), was exceedingly rare in Mainland Mandarin.

12) 任职 周人参 所 经营 的 百利行 电动 玩具 店

renzhi Zhou Renshen suo jingying de Bailihang diandong wanju dian
hold position Zhou Renshen operate DE Balilin electric toy store
ren: hold
zhi: position
‘took a job at the Balilin Electric Toy Store operated by Zhou Renshen’
---- Taiwan data
10 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

13) 这篇 演说 值得 媲美 举世 著名
Zhe yanshuo zhide pimei jushi zhuming
pian speech worth compare world famous
This pi:
compare
mei:
beauty
的 英国 首相 丘吉尔 半个 世纪
de Yingguo shouxiang Qiuji’er ban ge shiji
DE British Prime Churchill half century
Minister
前 在 西点军校 那篇 演说。
qian zai Xidian na pian yanshuo
ago at Junxiao that speech.
West
Point
Military
Academy
‘This speech is worthy of comparison to the world-renowned address delivered by the British Prime Minister
Churchill at West Point Military Academy half a century ago.’
---- Taiwan data

Notable variations were also attested in terms of ‘polarity’ preferences. Specifically, several VO compounds exhibited
strong preferences and implications of positive polarity in Taiwan Mandarin, while their Mainland counterparts did not
show such preferences. For example, the same VO compound 插手 chashou tended to appear in positive contexts
in Taiwan, as illustrated in Example 14), while Mainland 插手 more frequently collocated with negations and with con-
texts with perceived negative evaluations (e.g., foreign interference in Taiwan issue in example 2)).

14） 各个 部门 热切 希望 他 可以 插手 运营,
Gege bumen reqie xiwang ta keyi chashou yunying
Various department eager hope he can intervene operation,
帮助 恢复 市场 繁荣。
bangzhu huifu shichang fanrong
help recover market prosperity.
‘Various departments are eager for him to intervene in operations, and help to restore market prosperity.’
---- Taiwan data

4. INTERPRETING THE DIFFERENTIATING FEATURES IN TERMS OF DEGREE OF TRANSITIVITY

In the previous section, we conducted a descriptive study of the variations in semantic/syntactic properties of the
attested objects of VO compounds based on nine pre-selected features, using various statistical tests. In this section,
we interpret these results in terms of degrees of transitivity. This interpretation is based on Hopper and Thompson’s
(1980) two parameters of transitivity of the potential object NP, demonstrating that VO compounds exhibit varying
degrees of transitivity across the two language varieties.
We adopted the measurement of transitivity degrees by examining two parameters for NPs established by Hopper
and Thompson (1980): individuality and affectedness of the potential object. Individuality pertains to the distinctness of
the patient from the agent, as well as its distinctness from its background. According to Hopper and Thompson (1980),
proper, animate, concrete, singular, count, and referential referents are considered more individuated and the affected-
ness of an object is defined as the degree to which an action is transferred to a patient and depends on how completely
that patient is affected.
Among all the differentiating features, four contrasting features related to the POS of the object may be viewed as
paired: the Mainland variety’s preference for deverbal and event nouns, and Taiwan variety’s preference for verb
 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814 11

phrases and clauses. These contrasts pertain to individuality, as deverbal and event nouns exhibit higher individuality
than phrases and clauses. In sum, these four features show that Mainland Mandarin VO compounds prefer a higher
individuality for individuated objects, thus requiring a higher degree of transitivity.
Another contrasting feature related to POS was the Mainland variety’s preference for common nouns as objects, a
preference not observed in the Taiwan variety. Nouns in general, and common nouns in particular, are typically consid-
ered more referential and individuated than other alternative categories. This preference for common noun objects indi-
cates a higher degree of transitivity.
The other contrasting feature related to POS was the Mainland VO compounds’ greater preference for quantity
phrases as objects (e.g., 投资一千万 touzi yi qianwan ‘invest 10 million’). Note that these phrases ‘measure out’ the
events. Similar to changes or movements, measuring out requires objects to be individuated units, although these units
are typically not directly affected. Tenny’s (1994) Measure-out Constraint (MoC), for instance, stipulates that measuring
out is an essential property of internal objects, meaning that NPs involved in measuring out are moderately affected but
highly individuated. In Mandarin, a direct object can typically be replaced by a numeral phrase that refers to the property
of the referential noun measuring out the event, such as 一千万 yi qianwan ‘10 million’, which quantifies the unex-
pressed goal and measures out the event simultaneously (Huang and Shi, 2016; Li, 2016). In comparison to the lack
of measure-out properties, objects that can be measured out are more individuated and more affected than the ones
that cannot be. Thus, the tendency to “measure out” also marks direct object NPs in the Mainland variety as higher
in transitivity.
Another contrasting feature was that Mainland VO compounds had a stronger tendency to take proper nouns as
objects. Proper nouns, as proposed by Hopper and Thompson (1980), indicated higher individuality and, thus higher
transitivity in the Mainland. Similarly, the last interpretable contrasting feature was the preference of Mainland VO com-
pounds for pronouns as objects. According to Hopper and Thompson (1980), a definite and referential object is often
perceived as more completely affected than an indefinite and non-referential object, thus indicating a higher transitivity
in Mainland usage when a definite and referential pronoun is employed.
The remaining contrasting features of the VO compounds with attested objects did not appear to have direct impli-
cations for transitivity and will not be discussed further in this paper. Neither polarity nor occurrence in a headline was
relevant to transitivity. Additionally, the post-verbal position of Mandarin temporal phrases did not affect verb transitivity
(Huang and Shi, 2016; Li, 2016). Lastly, the tendency to use aspectual markers more frequently seemed to be the gen-
eral tendency of the Mainland variety and was specific to VO.
In this section, we interpreted the contrasting behaviors of direct objects of VO compounds in two varieties of Man-
darin Chinese: Mainland and Taiwan. We demonstrated that eight contrasting tendencies could be explained in terms of
the transitivity of object NP: individuality and affectedness. All eight contrasting tendencies showed that direct objects
taken by Mainland Mandarin VO compounds exhibited higher transitivity than their Taiwan counterparts. The remaining
contrasting grammatical tendencies were neutral regarding transitivity.
In sum, the contrasting grammatical tendencies between Mainland and Taiwan Mandarin varieties can be general-
ized in terms of transitivity. For VO compounds that take direct objects, the Mainland variety exhibited higher transitivity
in its attested objects compared to Taiwan. Note that the results from earlier studies showed that Taiwan VO com-
pounds had higher transitivity than their Mainland counterparts concerning the ability to take direct objects (Jiang
and Huang, 2018, 2022). We propose that these seemingly orthogonal results are two sides of the same coin. Taiwan
VO compounds exhibit higher lexical transitivity, allowing them to accommodate a broader range of NP objects. Con-
versely, Mainland VO compounds are lexically less transitive, which limits the types of direct objects they can take.
Thus, when examining all VO compounds that take direct objects, the more selective set of direct objects in the Man-
darin variety occurs in grammatical contexts with higher transitivity. On the other hand, the wider range of direct objects
taken by VO compounds in Taiwan shows lower transitivity overall. In these two varieties, the variety with lower tran-
sitivity verbs requires contexts of higher transitivity for direct objects, and vice versa for the other variety.

5. DISCUSSION

In this section, we discuss how the variation in transitivity among VO compounds from different varieties reflects
their position within the evolutionary process continuum (RQ 2), by adopting noun incorporation theory, lexical diffusion,
and lexical constructionalization theory. Having established in the previous section that direct objects of Taiwan VOs
tend to be more transitive, our focus is to determine whether this tendency is influenced by other linguistic factors or
primarily due to the default assumption that Taiwan VOs undergo the process of transitivization more rapidly than
Mainland VOs.
12 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

5.1. Incorporation of compounds

Our analysis of the variations in direct-object-taking behaviors of VO compounds in two varieties of Chinese can
shed light on the theory of noun incorporation (Baker, 1988; Mithun, 1984, 1986). Noun incorporation is widely observed
in the morphological constructions of many languages and involves combining a noun stem with a verb stem to create a
derived verb stem with augmented semantic content. This process typically has two stages: scope narrowing and argu-
ment advancing. The direct-object-taking behaviors of VO compounds can be examined from the perspective of argu-
ment advancing.
Mithun (1984) noted that some verb-noun incorporated constructions allow an additional argument. In certain noun-
incorporated constructions, when a transitive verb incorporates its direct object, an instrument, location, possessor (or
other affected arguments) may assume the vacated object role, thereby advancing the argument (Mithun, 1984). A good
example in Mandarin is the verb-noun incorporation of the verb 插 cha ‘to insert’ and the noun 手 shou ‘hand’. The incor-
porated compound 插手 chashou ‘to interfere/mess (with)’ may take an oblique object (e.g., 地方事务 difang shiwu ‘local
affairs’) and the affected oblique argument is then advanced to become a direct object. The advanced direct object is
affected by the ‘intervention’ event, and noun incorporation facilitates this advancement.
Mithun’s (1984) proposition of elevating an affected argument to the direct object position represents an evolutionary
step following the original verb scoping process. These dual stages illustrate the developmental trajectory of noun incor-
poration. Thus, we argue that the differences in transitivity between Mainland and Taiwan varieties reflect distinct stages
in the evolutionary process of VO compounds. Given that Taiwan VO compounds exhibit significantly higher transitivity,
we assert that these compounds have progressed further in the process of transitivization compared to their Mainland
counterparts. The higher transitivity evident in Taiwan VO compounds implies their advancement in the transitivization
process, surpassing their Mainland equivalents. This observation highlights the distinction in the developmental path of
VO compounds between the two linguistic varieties.
Given the differences in argument advancement between the two varieties of Mandarin, we would like to determine if
differences in their argument selection tendencies provide insights into the motivations for incorporation. Recall that the
attested contrasting tendencies include structural parallelism, presence in headlines, the semantic polarity of the con-
struction, and usage of an aspectual marker. Regarding incorporation: 1) Structural parallelism may favor the use of
VO1 + O2 to maintain alignment with the typical V + O structure used in parallel contexts; 2) The brevity of headlines
may favor VO1 + O2 to avoid longer construction without VO incorporation; 3) Polarity focus on the verb (i.e., V + O)
could promote incorporation; 4) Aspect-taking may suggest that incorporation is motivated partly by avoiding separable
compounds (or ionization) before incorporation (e.g., V-asp-O).
In sum, our study of grammatical tendencies showed no emergent morpho-syntactic patterns favoring direct-object-
taking. This suggests that while noun incorporation could still play a role, morpho-syntactic conditions were not the pri-
mary cause of the changes in transitivity. Further exploration is needed to better understand the transitivity variations.

5.2. Lexical diffusion

In the previous section, we interpreted the variation in VOs using the theory of noun incorporation. However, this
theory did not provide a clear explanation for the contrasting tendencies of direct-object-taking VO compounds in the
two varieties. In this section, we find that lexical diffusion theory offers a suitable model for accounting both for the dif-
fering tendencies in the two varieties and for the idiosyncratic exceptions.
The lexical diffusion theory, originally proposed by Wang (1969, 1979) to study the process of sound change, posits
that sound changes occur abruptly at the phonetic level but progress gradually at the lexical level. This theory signifies
that individual words change at different rates or times. Specifically, the process of change is described as follows:
During the early phase of change, only a small subset of relevant morphemes is affected. Some affected morphemes
may change from X to Y directly, while others initially exhibit both X-pronunciation and Y-pronunciation. Over time, the
X-pronunciation is gradually suppressed in favor of the Y-pronunciation (Wang, 1969).
While the term ‘lexical diffusion’ has often been commonly associated with phonology, recent research has extended
this concept to syntactic changes (e.g., Cheng, 1990; Iyeiri, 2010; Nevalainen, 2006; Ogura, 1993; Tottie, 1991; Yue-
Hashimoto, 1993; Zhang, 2000). Previous studies have agreed on two key generalizations: 1) The gradual dissemina-
tion of the new syntactic form across the lexicon, meaning that the new syntactic form spreads incrementally rather than
affecting all contexts simultaneously, with some contexts adopting it earlier than others (e.g., Iyeiri, 2010; Nevalainen,
2006; Ogura, 1993; Tottie, 1991); 2) The presence of syntactic alternation as an essential concept in syntactic change,
signifying that before replacement occurs, different syntactic alternations coexist (e.g., Cheng, 1990; Yue-Hashimoto,
1993; Zhang, 2000).
 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814 13

The concept of lexical diffusion can be applied to the direct-object-taking behaviors of VO compounds. The degree of
transitivity varies not only among different verbs but also across regional varieties. VO compounds display idiosyncratic
patterns among distinct verbs and varieties. Additionally, we have observed that different syntactic alternations exist for
the VO1 + O2 construction, meaning that for most VO compounds, their intransitive and transitive usages co-exist. This
aligns with the concept of “competing change” in lexical diffusion theory (Yue-Hashimoto, 1993), where both old and
new forms of the same lexical item co-exist and compete over an extended period before one form prevails.
1) The verb-complement construction with the post-verbal preposition (VC construction): 有恩于后辈 you’en yu hou-
bei grateful_for_descendant ‘kind to descendants’;
2) The adverbial-verb construction with the pre-verbal preposition (AV construction): 不对后辈有恩budui houbei
youen not_for_descendant_grateful ‘not kind to descendants’;
3) The external object can be inserted between V and O1: the separation construction: 感前辈的恩gan qianbei de en
thank_predecessors_DE_gratitude ‘be thankful to predecessors’;
Moreover, contrary to overall tendencies, specific VO compounds may exhibit variations in the selection of construc-
tions that do not align with overall tendencies. For example, the same VO prefers different constructions in Mainland
China and Taiwan: in Taiwan, 求助 qiuzhu ‘ask for help’ favors the Adverbial-Verb construction 向政府求助 xiang
zhengfu qiuzhu to_government_ask for help ‘seek help from the government’, whereas, in Mainland China, it shows
a preference for the Verb-Complement construction 求助于政府 qiuzhu yu zhengfu ask for help_to_government ‘seek
help from the government’.
In summary, we argue that variations in the degree of transitivity follow the predictions of lexical diffusion theory. The
ability of VO compounds to take direct objects manifests idiosyncrasies across different verbs and language varieties.
The emergence of direct object-taking by VO compounds does not extend uniformly across all contexts. During this
transformative process, both intransitive and transitive forms coexist, and VOs from various regions may exhibit varying
preferences for different syntactic alternations.

5.3. Lexical constructionalization

In the previous section, we found that lexical diffusion theory effectively models the differing tendencies in two vari-
eties, and accounts for idiosyncratic exceptions. However, lexical diffusion does not provide a direct account of the dif-
ferent grammatical tendencies. In this section, we argue that lexical constructionalization theory can integrate insights
from the two aforementioned accounts and provide a comprehensive interpretative framework, along with a theory of
transitivity.
The theory of lexical constructionalization focuses on how individual words (lexemes) evolve and become associated
with specific grammatical constructions over time. The fact that VO compounds in Mandarin are acquiring the ability to
take direct objects can be viewed as a process of lexical constructionalization, i.e., VO compounds becoming associ-
ated with the V + Direct_Object construction. Under this framework, the variations between Mainland and Taiwan Man-
darin regarding the object-taking ability of VO compounds offer us an opportunity to view different stages of lexical
constructionalization.
The increase in transitivity observed in VO compounds, exemplified in Mandarin Chinese, reflects a broader trend
discussed by Traugott and Trousdale (2013, 2014), where languages evolve towards reduced compositionality. Note
that the earlier blocking account, which explained why VO compounds could not take direct objects, presupposed a cer-
tain degree of compositionality required for VO compounds to fulfill the direct object slot of the verbs. When VO com-
pounds become less compositional, they are treated as bona fide lexical verbs capable of taking direct objects.
However, as Traugott and Trousdale (2013, 2014) suggested, this shift is not necessarily unidirectional, as construction-
alization also results in a new compositional relationship between VO compounds and their direct objects. Our two stud-
ies on variations between Mainland and Taiwan VO compounds in fact indicate that transitivity plays a central role in this
process. Furthermore, our analysis in the current paper suggests that this process entails not merely an increase in tran-
sitivity but a shift of transitivity from the clausal context to the VO compounds since more transitive VO compounds are
paired with lower-transitivity clausal contexts (Taiwan) and vice versa, where languages are evolving towards reduced
compositionality. Given the shared earlier stage where VO compounds did not take direct objects, we hypothesize that
Taiwan Mandarin VO compounds are at a later stage of lexical constructionalization than Mainland VO compounds. In
this account, the shift of transitivity from clauses to compounds represents a transfer of grammatical information from a
phrasal to a lexical unit, making it less compositional.
Integrating insights from the three theories, we propose the following comprehensive account of the transitivity vari-
ations of Mandarin VO compounds from the perspective of both language change and language variation. We view both
language changes and variations as aggregated tendencies of lexical items. For language change, the focus is typically
on the cumulative effect and regularity of the result over time concerning the entire lexical group. Thus, language
14 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

change is often described as a rule for the formation of a construction. For language variation, the emphasis is typically
on the aggregation of lexical tendencies that differentiate language varieties, leading studies to focus on specific lexical
items showing significant divergence. Since lexical changes are fundamental to both language change and variation,
understanding the cause of changes in lexical tendencies is crucial for explaining both phenomena.
We propose that, at the lexical level, Mandarin VO compounds are undergoing lexical constructionalization, enabling
them to take direct objects. While it might be expected that this constructionalization would be morphologically signaled
by noun incorporation, many incorporated VO compounds cannot take direct objects. Specifically, our study demon-
strated complementary tendencies between the two varieties regarding the transitivity of VO compounds, and the tran-
sitivity of the grammatical contexts of object-taking VO compounds. Adopting the hypothesis that lexical
constructionalization instantiates the general tendency toward less compositionality (Traugott and Trousdale, 2012),
these complementary transitivity sites suggest a replacement change (e.g., Blythe and Croft, 2012). That is, given
the eventive information of the directly affected participant, the compositionally encoded transitivity information at the
clausal level (before incorporation) is gradually shifting to a less compositional site of the incorporated compound.
As a replacement change and following the nature of lexical rules in general, we know that the speed of shifting can
vary lexically. By lexical diffusion (Wang, 1979) and following the S-curve model of replacement changes (Blythe
and Croft, 2012), we interpret the variation in direct-object-taking abilities between two varieties as reflecting the inter-
mediate stages preceding the snowball effect of regular changes. The variations, on the other hand, provide empirical
evidence that VO compounds are in the process of lexical constructionalization.
This integrated lexical constructionalization account, based on the grammatical encoding of transitivity, offers several
advantages over traditional accounts. First, it eliminates the need for ad hoc stipulations regarding the object-taking (in)
ability of VO compounds. The eventive information remains consistent, with differences only in the site of encoding,
whether at the clause or within non-compositional compounds. Second, it provides a unified explanation for the trend
toward decreasing compositionality, which a morpholexical account fails to explain—specifically, why some incorpo-
rated compounds are more likely to take direct objects than others despite undergoing the same compounding process.
Third, it also avoids the issue of sustainability of the tendency. That is, if the compositional tendency is interpreted pri-
marily at the morpholexical level, then at one point one language could become extremely non-compositional, impeding
further change. Fourth, it accounts for how variations occur without significant consequences of the shared lexical
forms. That is, the VO compounds are morphologically identical in the two varieties, and parallel among different lexical
items, regardless of their differences in their grammatical behaviors. This account requires that the compound formation
(i.e., noun-incorporation) process be completed and the compound forms fully lexicalized to serve as non-compositional
targets for shifting transitivity information. The delinking of morpholexical processing of noun incorporation and the shift
of transitivity plays a key role in accounting for Mandarin Chinese VO compounding transitivity variations. This analysis
also underlines the lexical nature of constructionalization, emphasizing that a construction is lexical in the sense that it
features a unique form-meaning pairing, which in turn indicates that changes in form and meaning do not need to occur
in parallel.

6. CONCLUSION

In this study, we explored the direct-object-taking behaviors of VO compounds by focusing on the potential objects
instead of the incorporated objects. We achieved two key objectives. We found that the contextual transitivity of direct-
object-taking VO compounds differs between Mainland and Taiwan Mandarin in a way that complements the stronger
lexical transitivity in Taiwan. Our comprehensive corpus-based study examined contrasting tendencies between direct-
object-taking VO compounds in these two varieties. The goal was to determine whether the contexts of attested tran-
sitive uses differ, given the slight behavioral variations in direct-object-taking. Our study identified nine significantly dif-
ferent tendencies. The analysis revealed that eight tendencies can be accounted for with the generalization that taken
direct objects appear in clausal contexts with higher transitivity in Mainland Mandarin. We argued that this generalization
is compatible with the previously established contrast that Taiwan VO compounds exhibit higher lexical transitivity com-
pared to their Mainland counterparts. That is, the lower lexical transitivity of Mainland VO compounds means that the
direct objects they take occur in contexts with higher transitivity. Conversely, the higher lexical transitivity of Taiwan VO
compounds allows their direct objects to exist in lower transitivity contexts. Thus, the transitivities of VO compounds and
the clausal contexts of their direct objects vary in opposite directions in the two varieties of Mandarin.
We addressed the two research questions based on the results obtained from our two objectives. We found that vari-
ations in the direct object-taking abilities of Mandarin VO compounds can be attributed to general tendencies of transi-
tivity changes. That is, by adopting lexical constructionalization, supplemented with noun incorporation and lexical
diffusion, we observed that VO compound variations reflect the general trend towards encoding transitivity with lower
compositionality. In particular, noun incorporation theories predict behaviors such as VO compounds taking an addi-
 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814 15

tional direct object during the argument advancement stage, following morphological combination due to scope narrow-
ing. However, there is no clear explanation for the contrasting tendencies of direct-object-taking VO compounds in the
two varieties. We found that lexical diffusion theory provides a useful model for understanding different tendencies in
these varieties and idiosyncratic exceptions, but it does not directly explain the different grammatical tendencies. Finally,
we found that lexical constructionalization can integrate insights from the previous accounts and provide a comprehen-
sive interpretative framework alongside a theory of transitivity. Under the reduced compositionality trend, the direct-
object-taking behaviors of Mandarin VO compounds can be explained by the shifting of transitivity information from
the clausal context to the incorporated compound, i.e., from a high compositionality unit to a low compositionality unit.
Given that argument advancement is a later stage and considering the trend towards lower compositionality, the reli-
ance on high transitivity contexts for direct objects in Mainland Mandarin suggests it is in an earlier stage of lexical con-
structionalization compared to Taiwan Mandarin.
Our research significantly contributes to accounting for VO variations between Mainland Mandarin and Taiwan Man-
darin by proposing a novel perspective of focusing on the properties of the potential objects instead of the incorporated
objects. Our approach not only effectively identifies distinctions among various Chinese varieties but also delves into the
nature and motivation of these variations from a theoretical perspective. Furthermore, we proposed and validated sta-
tistical models for analyzing linguistic variations using comparable corpora, thus paving the way for enhanced synergy
between corpus linguistics and theoretical linguistics. Lastly, we combined insights from noun incorporation and lexical
diffusion theories into a lexical constructionalization framework to account for the complex transitivity variation data in
two Mandarin varieties, showing how general tendencies of change are maintained despite lexically driven variations.

CREDIT AUTHORSHIP CONTRIBUTION STATEMENT

Menghan Jiang: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Resources,
Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualiza-
tion. Chu-Ren Huang: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision,
Conceptualization.

Data availability

Data will be made available on request.

ACKNOWLEDGEMENT

This work was supported by 2023 International Chinese Language Education Research Project “Study on Variations
in Ditransitive Constructions under the Globalization Perspective” [23YH81D]. It was also partially supported by the
PolyU-PekingU Research Centre on Chinese Linguistics. We sincerely thank the editors and reviewers for their valuable
comments and suggestions, which have significantly improved the quality of this paper. Additionally, thanks to Professor
Kathleen Ahrens for her comments and contributions on earlier version of this work. All errors are our own.

APPENDIX A. ANNOTATION CRITERIA

1. POS: the syntactic type of the object taken by a VO compound

a. Common NP: 获赠一部手机huozeng yibu shouji receive_one_CL_phone ‘receive a phone’;
b. Event noun: 投身大游行toushen da youxing plunge into_big_parade ‘join the parade’;
c. Deverbal noun: 插手电商的运营chashou dianshang de yunying intervene_e-commerce_DE_operation ‘be
involved in e-commerce operation’;
d. Clause: 投诉澳娱分红不合理tousu aoyu fenhong bu heli complain_STDM_dividend_BU_fair ‘complain that
the dividends of STDM are not unfair’;
e. VP: 插手经营家族企业chashou jingying jiazu qiye intervene_manage_family_enterprise ‘intervene the man-
agement of family enterprise’;
f. Quantity: 驻军十六万zhujun shiliuwan station_troop_160000 ‘station 160,000 troops’;
g. Time: 执教五十年zhijiao wushinian coach_fifty_years ‘coach for fifty years’;
16 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

2. Proper noun: the taken object is a proper noun or not:

a. Yes: 移民美国 yimin Meiguo immigrate_America ‘immigrate to America’;
b. No: 移民 发达 国家 yimin fada guojia immigrate_developed_country ‘immigrate to developed country’;
3. Pronoun: whether the VO takes a pronoun or not
a. Yes: 帮忙他 bangmang ta do a favor_3 SG ‘do him a favor’
b. No: 帮忙家人bangmang jiaren do a favor_family ‘do family a favor’
4. Structural Parallelism: Whether the VO + O construction has parallel structure or not
a. Yes: 他们坚持把关审核, 强化责任, 规范操作, 配套协调tamen jianchi baguan shenhe qianghua zeren guifan cao-
zuo peitao xietiao_insist_guarantee_check_strengthen_responsibility_standardize_operation_support_coordina
tion ‘They insist on guaranteeing the checking, strengthening the responsibility, standardizing the operation,
and supporting the coordination’)
b. No: 把关服务质量baguan fuwu zhiliang guarantee_service_quality ‘guarantee the quality of service’)
5. Headline: Whether the constructions appear in a headline
a. Yes: marked with 标题biaoti ‘title’
b. No: without the mark
6. Polarity: the semantic polarity of context
a. Positive: 各个部门热切希望他可以插手 gege bumen reqie xiwang ta keyi chashou every_department_intent_h
ope_3 SG_can_intervene ‘Every departments are intent on his interventions’
b. Neutral: 她很少插手妹妹的事ta henshao chashou meimei de shi SG_rarely_intervene_little sister_DE_thing ‘She
rarely intervenes in the business of her little sister.’
c. Negative: 各方势力经常插手流通环节, 造成市场秩序十分混错乱gefang shili jingchang chashou liutong huanjie,
zaocheng shichang zhixu shifen cuoluan all_parties_force_often_intervene_lin_flow_cause_market_order_very_
disruption ‘The parties often intervene in the flow of links, resulting in the disruption of market order.’
7. Complexity of the object: Whether the object of VO has a modifier
a. Yes: 媲美这个世界一流名声大噪的新产品pimei zhege shijie yiliu mingsheng dazao de xinchanpin rival with_this_
world-class_famous_DE_new_product ‘rival with this world-class famous new product’)
b. No: 媲美中亚pimei ZhongYa rival with_Central_Asia ‘rival with Central Asia’
8. Aspectual Marker: Whether a VO is affixed with zhe.le.guo
a. Yes:
a) ASP.le: 出版了一套系列丛书chuban le yitao xilie congshu publish_LE_one_CL_series_books ‘have pub-
lished one series of books’;
b) ASP.zhe: 媲美着世界闻名的技术pimei zhe shijie wenming de jishu rival with_ZHE_world_famous_DE_tech
nology ‘rival with world-famous technology’;
c) ASP.guo: 执教过中国队zhijiao guo Zhongguo dui coach_GUO_Chinese_National_team ‘have coached the
Chinese National team’)

b. No: 出版系列文集chuban xilie wenji publish_series_collected_works ‘publish serials’)

9. Animacy: the animacy of the object:
a. Animate (中意这个徒弟zhongyi zhege tudi like_this_apprentice ‘like this apprentice’)
b. Inanimate (中意这道菜zhongyi zhe dao cai like_this_CL_dish ‘like this dish’)
 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814 17

APPENDIX B. VO COMPOUND LIST

致函 zhihan ‘to write a letter’ 签约 qianyue ‘to sign a contract’

从事 congshi ‘to engage in’ 赠书 zengshu ‘to gift books’
变身 bianshen ‘to transform’ 媲美 pimei ‘to rival, to be comparable to’
涉嫌 shexian ‘to be suspected of’ 投诉 tousu ‘to lodge a complaint’
获赠 huozeng ‘to receive as a gift’ 领军 lingjun ‘to lead, to be at the forefront’
涉足 shezu ‘to get involved in’ 入职 ruzhi ‘to start a job, to join a company’
出席 chuxi ‘to attend’ 参演 canyan ‘to participate in a performance, to take
part in acting’
致信 zhixin ‘to send a letter’ 投资 touzi ‘to invest’
致电 zhidian ‘to make a phone call’ 驻军 zhujun ‘to station troops, to be stationed’
得罪 dezui ‘to offend’ 致谢 zhixie ‘to express gratitude, to give thanks’
驰誉 chiyu ‘to be renowned’ 接轨 jiegui ‘to integrate, to align’
取悦 quyue ‘to please, to delight’ 入境 rujing ‘to enter a country, to go through
customs’
进军 jijin ‘to march into, to advance’ 满意 manyi ‘to be satisfied, to be pleased’
调任 diaoren ‘to transfer to another post’ 登场 dengchang ‘to make an appearance, to go on
stage’
插手 chashou ‘to intervene’ 携手 xieshou ‘to join hands, to work together’
放眼 fangyan ‘to have a broad view’ 把关 baguan ‘to check, to inspect’
立足 lizu ‘to establish oneself’ 失信 shixin ‘to break one's promise’
借道 jiedao ‘to pass through, to take a 启航 qihang ‘to set sail, to start a journey’
detour’
定都 dingdu ‘to establish a capital’ 避难 binan ‘to take refuge, to seek asylum’
接手 jieshou ‘to take over’ 献计 xianji ‘to offer advice, to give suggestions’
寄语 jiyu ‘to send one's regards or wishes’ 联手 lianshou ‘to join forces, to collaborate’
聚会 juhui ‘to gather, to have a meeting or 移民 yimin ‘to immigrate, to migrate’
party’
投身 toushen ‘to devote oneself to’ 撤军 chejun ‘to withdraw troops’
称霸 chengba ‘to dominate, to rule over’ 接壤 jierang ‘to border on, to be adjacent to’
现身 xianshen ‘to appear, to show up’ 垂爱 chui’ai ‘to be deeply loved, to be highly
esteemed’
迁居 qianju ‘to move residence’ 领航 linghang ‘to lead the way, to guide’
动员 dongyuan ‘to mobilize’ 过目 guomu ‘to glance over, to skim through’
寄望 jiwang ‘to have hopes, to expect’ 转行 zhuanhang ‘to change professions, to switch
careers’
聚焦 jujiao ‘to focus’ 撤兵 chebing ‘to withdraw troops’
取信 quxin ‘to gain trust, to be believed’ 观光 guanguang ‘to sightsee, to tour’
称雄 chengxiong ‘to be unrivaled, to be dominant’ 失望 shiwang ‘to be disappointed’
出土chutu ‘to unearth, to excavate’ 助阵 zhuzhen ‘to assist, to support’
出兵 chubing ‘to send troops’ 转岗 zhuangang ‘to transfer jobs, to change positions’
关心 guanxin ‘to care about, to be concerned 参展 canzhan ‘to participate in an exhibition, to exhibit’
about’
执教 zhijiao ‘to teach, to coach’ 出丑 chuchou ‘to make a fool of oneself, to disgrace
oneself’
献身 xianshen ‘to dedicate oneself to’ 除名 chuming ‘to expel, to remove from a list’
插足 chazu ‘to intervene, to get involved in’ 触电 chudian ‘to get an electric shock, to be
electrified’
18 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

Appendix B (continued)

改名 gaiming ‘to change one's name’ 登台 dengtai ‘to go on stage, to make a public
appearance’
驰名 chiming ‘to be famous, to be renowned’ 对话 duihua ‘dialogue, to have a conversation’
求证 qiuzheng ‘to seek verification, to seek 发力 fali ‘to exert effort, to make an effort’
confirmation’
起诉 qisu ‘to sue, to file a lawsuit’ 放手 fangshou ‘to let go, to release one's grip’
挑战 tiaozheng ‘to challenge’ 分羹 fengeng ‘to share the benefits, to have a share of
the spoils’
出版 chuban ‘to publish’ 奉命 fengming ‘to be ordered, to receive a mission’
登陆 denglu ‘to land, to arrive’ 告破 gaopo ‘to solve a case, to crack a case’
移情 yiqing ‘to transfer affection, to be 攻关 gongguan ‘to tackle a key problem, to overcome
emotionally moved’ difficulties’
帮忙 bangmang ‘to help, to lend a hand’ 毁容 huirong ‘to disfigure, to ruin one's appearance’
会师 huishi ‘to meet, to assemble’ 结盟 jiemeng ‘to form an alliance, to ally’
揭秘 jiemi ‘to reveal secrets’ 捐资 juanzi ‘to donate money, to contribute funds’
牵手 qianshou ‘to hold hands’ 联网 lianwang ‘to connect to the internet, to network’
迁怒 qiannu ‘to transfer one's anger’ 联姻 lianyin ‘to form a marriage alliance, to marry
into a family’
对阵 duizhen ‘to confront, to be pitted against’ 露脸 loulian ‘to show one's face, to make an
appearance’
提名 timing ‘to nominate’ 露面 loumian ‘to appear in public, to be seen’
定名 dingming ‘to name, to christen’ 落选 luoxuan ‘to fail to be elected, to lose in an
election’
命名 mingming ‘to name, to christen’ 拍板 paiban ‘to give approval, to seal the deal’
续约 xuyue ‘to renew a contract’ 攀亲 panqin ‘to make a matrimonial alliance’
参选 canxuan ‘to run for election’ 驱车 quche ‘to drive a car, to travel by car’
出身 chushen ‘to come from, to be born into’ 就职 jiuzhi ‘to take office, to assume a position’
入籍 ruji ‘to become a citizen’ 连线 lianxian ‘to connect, to establish a connection’
受聘 shoupin ‘to be hired, to be employed’ 迁址 qianzhi ‘to relocate, to move’
把脉 bamai ‘to take the pulse, to diagnose’ 取经 qujing ‘to seek advice, to learn from
experience’
致力 zhili ‘to devote oneself to, to dedicate 缺阵 quezhen ‘to be absent, to miss out’
oneself to’
抗衡 kangheng ‘to counterbalance, to contend 入口 rukou ‘entrance, entry point’
with’
中意 zhongyi ‘to favor, to be fond of’ 撒手 sashou ‘to let go, to release one's grip’
操心 caoxin ‘to worry about, to be concerned 让利 rangli ‘to offer discounts, to make
about’ concessions’
约会 yuehui ‘to date, to make an 升温 shengwen ‘to rise in temperature, to warm up’
appointment’
留学 liuxue ‘to study abroad’ 示好 shihao ‘to show goodwill, to make friendly
gestures’
任职 renzhi ‘to take office, to hold a position’ 受益 shouyi ‘to benefit, to gain profit’
入股 rugu ‘to invest in shares, to become a 索赔 suopei ‘to claim compensation, to seek
shareholder’ redress’
感恩 gan’en ‘to be grateful, to feel thankful’ 贪污 tanwu ‘to embezzle, to commit corruption’
曝光 baoguang ‘to expose, to reveal’ 提速 tisu ‘to accelerate, to increase speed’
 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814 19

Appendix B (continued)

求助 qiuzhu ‘to seek help, to ask for 闻讯 wenxun ‘to hear news, to receive information’
assistance’
结缘 jieyuan ‘to form a bond’ 效力 xiaoli ‘to be effective, to exert effort’
更名 gengming ‘to change one's name’ 扬帆 yangfan ‘to set sail, to start a journey’
抱怨 baoyuan ‘to complain, to grumble’ 援手 yuanshou ‘to lend a helping hand, to offer
assistance’
拜师 baishi ‘to pay respects to a teacher, to 约会 yuehui ‘to make an appointment, to go on a
become a disciple’ date’
控股 konggu ‘to control shares’ 展翅 zhanchi ‘to spread one's wings, to stretch out’
解码 jiema ‘to decode, to decrypt’ 助威 zhiwei ‘to cheer, to show support’
过境 guojing ‘to pass through a border, to
transit through’

APPENDIX C. ABBREVIATION LIST

ZHE: aspectual marker 着 zhe

LE: aspectual marker 了 le
GUO: aspectual marker 过 guo
CL: classifier
DE: 的 de
3 SG: 3rd person singular

References

Babby, L.H., 1975. A transformational analysis of transitive-sja verbs in Russian. Lingua 35 (3–4), 297–332.
Baker, M.C., 1988. Incorporation: A theory of grammatical function changing. University of Chicago Press.
Bates, D., and Maechler, M., 2009. lme4: Linear mixed-effects models using S4 classes. R package version 0.999375-32.
HYPERLINK http://CRAN.R-project.org/package=lme4.
Bates, D., Mächler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67 (1), 1–48.
https://doi.org/10.18637/jss.v067.i01.
Beckner, C., Blythe, R., Bybee, J., Christiansen, M.H., Croft, W., Ellis, N.C., Holland, J., Ke, J., Larsen-Freeman, D., Schoenemann,
T., 2009. Language is a complex adaptive system: position paper. Language Learn 59, 1–26.
Blythe, R.A., Croft, W., 2012. S-curves and the mechanisms of propagation in language change. Language, 269–304.
Bresnan, J., Cueni, A., Nikitina, T., Baayen, R.H., 2007. Predicting the dative alternation. Cognitive Foundations of Interpretation,
69–94.
Cheng, R.L., 1990. Ci2hui4 Kuo4san4 Li3lun4 Zai4 Ju4fa3 Bian4hua4 Li3 de Ying4yong4 – Jian1tan2 Tai2wan1 Guan1hua4
You3zi4ju4 de Ju4fa3 Bian4hua4 (Lexical Diffusion Theory and its application in syntax: A discussion of the changes in you
constructions in Taiwan Mandarin). Language Teaching and Linguistic Studies 1, 66–73.
Diao, Y.B., 1998. Ye3 tan2 ‘Dongbinshi Dongci+Binyu’ (Research on VO+O construction). Language Planning 6, 39–41.
Gao, G.S., 1998. Dong4bin1shi4 Dong4ci2+Bin1yu3 de Da1pei4 Gui1lü4 (The Collocation Rules for VO+O construction).
Language Planning 6, 36–38.
Her, O.S., 1996. Variation of the VO construction in Chinese: a synchronic account. Linguistics 34 (4), 733–752.
Hopper, P.J., Thompson, S.A., 1980. Transitivity in grammar and discourse. Language, 251–299.
Hou, R., Huang, C.R., Ahrens, K., 2022. Regional varieties and diachronic changes in Chinese political discourse. Humanities and
Social Sciences Communications 9 (1), 1–18.
Huang, J.-C.-T., 1984. Phrase structure, lexical integrity, and Chinese compounds. Journal of Chinese Teachers Association 19 (2),
53–78.
Huang, C.R., 1985. On pseudo-possessive NPs in Mandarin Chinese. Cornell Working Papers in Linguistics 7, 129–148.
Huang, C.R., 1990. A unification-based LFG analysis of lexical discontinuity. Linguistics 28 (2), 263–308.
Huang, C.R., 2009. Tagged Chinese gigaword version 2.0, ldc2009t14. Linguistic Data Consortium.
Huang, C. R., Lin, J., Jiang, M., and Xu, H., 2014. Corpus-based study and identification of Mandarin Chinese light verb variations.
In Proceedings of the First Workshop on Applying NLP Tools to Similar Languages, Varieties and Dialects (pp. 1-10).
Huang, C.R., Shi, D. (Eds.), 2016. A Reference Grammar of Chinese. Cambridge University Press.
Iyeiri, Y., 2010. Verbs of Implicit Negation and their Complements in the History of English. John Benjamins Publishing.
20 M. Jiang, C.-R. Huang / Lingua 311 (2024) 103814

Jiang, M., and Huang, C.R., 2018. Transitivity variations in mandarin VO compounds—a comparable corpus-based approach. In:
Chinese Lexical Semantics: 18th Workshop, CLSW 2017, Leshan, China, May 18–20, 2017, Revised Selected Papers 18.
Springer International Publishing. Pp. 564–575.
Jiang, M., Huang, C.R., 2022. Han4yu3 dong4bin1 fu4he2 ci2 de ji2wu4 xing4 ji2qi2 yong4fa3 cha1yi4 – ji1yu2 yu3liao4ku4
qu1dong4 fang1fa3 de dui4bi3 yan2jiu1 (The transitivity of Mandarin Chinese VO compounds and variations on their
employment: A corpus-driven comparative study). Zhongguo Yuwen 1, 39–47.
Lee, Y.M.S., 2010. A linguistic approach to emotion detection and classification. The Hong Kong Polytechnic University, Doctoral
dissertation.
Li, A.Y.H., 2012. Order and Constituency in Mandarin Chinese, Vol. 19. Springer Science & Business Media.
Li, A.Y.H., 2016. Verbs and Verb Phrases. A Reference Grammar of Chinese. Cambridge University Press, Cambridge, pp. 81–115.
Li, C.N., Thompson, S.A., 1981. Mandarin Chinese: A functional reference grammar. University of California Press.
Lin, J., Shi, D., Jiang, M., Huang, C.R., 2019. Variations in world Chineses. In: The Routledge Handbook of Chinese Applied
Linguistics. Routledge, pp. 196–211.
Ling, D.X., 1999. Shi4lun4 Shuang1yin1jie2 Dong4bin1shi4 Dong4ci2 Dai4 Bin1yu3 de Ji1ben3 Gui1lü4 (An Analysis on Verb-
object Type + Object). Chinese Language Learning 5, 9–13.
Liu, Y.J., 1993. Dong4bin1 Shi4 Dong4ci2 Yu3 Suo3 Dai4 Bin1yu3 Zhi1jian1 de Yu3yi4 Guan1xi (The Semantic Correlation
between VO compounds and its Taken Complement). Chinese Language Learning 4, 48–53.
Luo, X.R., 1998. Dong4bin1 Shi4 Dong4ci2 + Bin1yu3 Gui1lü4 Tan4jiu1 (Analysis on VO+O). Language Planning 5, 27–30.
Masolo, C., Borgo, S., Gangemi, A., Guarino, N., Oltramari, A., Oltramari, R., Schneider and Horrocks, I., 2002. WonderWeb
deliverable D17. the WonderWeb library of foundational ontologies and the DOLCE ontology.
Mithun, M., 1984. The evolution of noun incorporation. Language 60 (4), 847–894.
Mithun, M., 1986. On the nature of noun incorporation. Language 62 (1), 32–37.
Nevalainen, T., 2006. Syntactic Structures. In Introduction to Early Modern English. Edinburgh University Press, 103-117.
Ogura, M., 1993. The development of periphrastic do in English: A case of lexical diffusion in syntax. Diachronica 10 (1), 51–85.
Pinheiro, J.C., Bates, D.M., 2000. Mixed-effects Models in S and S-Plus. Springer, New York.
Qian, C.Y., 2011. Xian4dai4 Han4yu3 ‘Dongbin Shi Fuhe Ci Dai Binyu’ Jiegou Fenxi (Structural Analysis on VO+O construction in
Modern Chinese) (Master’s thesis). Fudan University, Shanghai.
Qin, Z., Duanmu, S., 2017. A judgment study of word-length preferences in Chinese NN compounds. Lingua 198, 1–21.
Rao, C.R., 1984. Dong1bin1 Zu3he2 Dai4 Bin1yu3 (On VO+O). Zhongguo Yuwen 6, 413–418.
Richards, J., Platt, J., Weber, H., 1985. Longman Dictionary of Applied Linguistics. Longman, Harlow.
Robins, R.H., 1964. General Linguistics: An Introductory Survey. Longman, London.
Tenny, C., 1994. Aspectual Roles and the Syntax-Semantics Interface, Vol. 52. Kluwer, Dordrecht.
Tottie, G., 1991. Lexical diffusion in syntactic change: frequency as a determinant of linguistic conservatism in the. Historical
English Syntax 2, 439.
Traugott, E. C., and Trousdale, G., 2013. Constructionalization and constructional changes (Vol. 6). OUP Oxford.
Traugott, E.C., Trousdale, G., 2014. Contentful constructionalization. Journal of Historical Linguistics 4 (2), 256–283.
Trousdale, G., 2012. Grammaticalization, constructions and the grammaticalization of constructions. Grammaticalization and
language change: New reflections, 167-198.
Voigtlander, K., Bartholomew, D., 1972. Semology and transitivity in Eastern Otomi verbs. Lingua 29, 38–53.
Wang, W.-Y. (Ed.), 1977. The Lexicon in Phonological Change. Mouton, The Hague.
Wang, W.-Y., 1979. Language change a lexical perspective. Ann. Rev. Anthropol. 8 (1), 353–371.
Wang, H.D., 1997. Dong4bin1 Shi4 Dong4ci2 + Bin1yu3 Gui1lü4 He2zai4 (What is the pattern for VO+O construction). Language
Planning 8, 30–31.
Wang, William S-Y., 1969. Competing changes as a cause of residue. Language, 9-25.
Xu, H., Jiang, M., Lin, J., Huang, C.R., 2022. Light verb variations and varieties of Mandarin Chinese: Comparable corpus driven
approaches to grammatical variations. Corpus Linguist. Linguist. Theory 18 (1), 145–173.
Yang, H.M., 2001. ‘Vo+N’ yu3 Yu4yi5 Jie2gou4 de Jian1rong2 yu3 Chong1tu1 – Han4yu3 Dong4bin1 Zu3he2 Dai4 Bin1yu3
Jie2gou4 Zhong1 de Yu3yi4 Wen4ti2 (The compatibility and conflict of the meaning and structure on the ‘VO+N’ structure).
Chinese Language Learning 1, 28–34.
Yue-Hashimoto, A., 1993. The Lexicon In Syntactic Change: Lexical Diffusion In Chinese Syntax. Journal of Chinese Linguistics,
213–254.
Zhang, M., 2000. Syntactic change in Southeastern Mandarin: How does geographical distribution reveal a history of diffusion? In:
Memory of Professor Li Fang-Kuei: Essays of Linguistic Change and the Chinese Dialects. Academia Sinica and University of
Washington, Taipei and Seattle, pp. 197–242.
Zhu, J., Sheng, X.H., 2008. ‘Dong4bin1 Jie2gou4 Dai4 Bin1yu3’ Ge2shi4 Cheng1yin1 Tan4jiu1 (On Formation Causes of Structure
VN+O). Chinese Language Learning 3, 37–42.