Non-canonical Bei Passives: A Cognitive Linguistic Analysis

Yu Fan
Kyoto University
arther830622@gmail.com

Abstract: The aim of the present study is to examine the function of the new non-canonical bei-XX passive
construction in Mandarin, which seems not conforming to the general function of passives, “agent-
defocusing”. The novel construction bei-XX can be categorized into the “being forced” type and the “being
(falsely) reported/ said” type based on their constructional meanings, while they share similar syntactic
features: taking non-prototypical predicates like intransitives, the demoted agent being unrecoverable, and
having a non-Patient subject. In order to clarify the event construal of bei-XX, cognitive linguistic
conceptions like the action chain model are employed. As a result, two prominent characteristics of this
construction can be summarized as follows: (i) the most salient participant, which is realized as subject,
lacks some degree of “subjecthood” and therefore needs external energy, while (ii) the source of external
energy may be out of the scope of predication. It is pointed out that the difference in constructional
meanings of the two types of bei-XX resides in whether the event they denote corresponds to reality or not.
Lastly, the event construal of the two types of bei-XX, as well as the constructional schema for bei-XX, is
represented efficiently by the action chain model.

Keywords: bei; bei-XX construction; non-canonical passives; cognitive grammar; action chain
model; subjecthood

1. Introduction
The aim of this paper is to describe and analyze the new uses of Mandarin bei passives from the perspective
of cognitive linguistics. The voice alternation has been studied by functional and cognitive literature, and
it has been acknowledged that the general function of passive is “agent defocusing” in a transitive clause
(Shibatani 1985). This view is also supported by cognitive grammar which assumes that grammatical
relations of a clause correlate with relative salience among the elements of an event described. On the other
hand, the function of agent-defocusing does not apply to non-canonical passives such as Japanese indirect
passives, which are based on intransitive verbs and often lack agentive participants; they also display a
semantic characteristic that the referents of the subjects of the indirect passives are interpreted as negatively
affected by the occurrence of event described.
In recent Mandarin, one also finds non-canonical novel usages of bei passives involving verbs of

©︎2020 Yu Fan. All Rights Reserved.

76

low transitivity, such as intransitive verbs and so on, which can be semantically classified into “being
forced” type and “being (falsely) reported/ said” type. This study will motivate the functions of these two
types of bei passives by clarifying the event construal they denote employing the action chain model.
The paper is organized as follows. Following section 1, section 2 presents significant concepts of bei
passives in Mandarin, including the properties of its canonical and non-canonical variants, and the passive
marker 被 bei. Description of the aim of this study, the new bei-XX construction, and review regarding
its previous studies are presented in section 3. Subsequently, section 4 introduces Cognitive linguistic
approaches and gives an analysis of the new bei-XX construction. Finally, concluding remarks and
suggestions for further studies will be proposed in section 5.

2. Preliminaries: The function of passives and Bei Passives in Mandarin

2.1. Voice: active vs. passive
Voice is a grammatical category concerning the correspondences between constituents’ thematic roles with
their grammatical functions in a sentence (O’Grady 2001). Typically, when describing an event with a doer
(the agent) and a receiver (the patient) of an action, the participant who performs the action is prone to be
realized syntactically as subject, and the affected participant as direct object.

(1) John killed Bill. [active]

agent patient
SUBJECT DIRECT OBJECT

This results in the verbalized sentence in active voice, the unmarked or “basic” voice category of all
languages (Givón 1984, O’Grady 2001). However, we can reorganize the mappings of the sentence if we
choose to describe the event from a different perspective. For example, in English, if we want to shed light
on the affected participant, by downgrading the agent to oblique (prepositional by-phrase, as shown in (2))
and raising the theme to the subject position, we can obtain a corresponding passive sentence. (Of course,
an auxiliary and a verb in past participle form are needed for English passive.)

(2) Bill was killed by John. [passive]

patient by agent
SUBJECT OBLIQUE

Through the rearrangement of grammatical relations, voice provides a strategy for focusing on different
participants in an event (Berk 1999).
 Papers in Linguistics Science, No. 26 (2020) 77

2.2. Canonical passive in Mandarin: bei passive

Alternation of active and passive can also be seen in Mandarin Chinese. See (3) and (4) below. 1

(3) 小明 罵 了 他。 [active]
xiaomin ma le ta
Xiaomin scold ASP 3sg
‘Xiaomin scolded him.’
(4) a. 他 被 小明 罵 了。 [long passive]
ta bei xiaomin ma le
3sg BEI Xiaomin scold ASP

‘He is/was scolded by Xiaomin.’

b. 他 被 罵 了。 [short passive]
ta bei ma le
3sg BEI scold ASP

‘He is/was scolded.’

There are two possible constructions for canonical passive in Madarin as shown in (4). While the nominal
phrase bearing the patient role (他 ‘he’) appears in the subject position, the agent ( 小明 ‘Xiaomin
(name)’), which is the subject of the corresponding active sentence, either (i) appears in front of the
predicate and is introduced by the passive marker2 被 bei, or (ii) simply does not show up in the passive

1
The glossing of examples in this study consists of four parts: The first line preserves the original text.
Since the examples include data extracted from online news or websites from both China (People's
Republic of China) and Taiwan, both simplified Chinese characters and traditional Chinese characters can
be seen in the examples. The second line is the transcription of the pronunciation. Pinyin, the official
romanization in China, is used in the present thesis, while the tone diacritics are omitted. The third line is
the word-by-word translation, while words which performs certain grammatical functions will be glossed
by the abbreviations mostly following Li & Thompson (1981)’s convention. The last line is the free
translation of the utterance. Sometimes the second and the third line are omitted when the complete
glossing has already appeared or the discussion is mainly regarding to the meaning of the expression. The
glossing in this article is done by the author if there is no source of citation.
In addition, when the expression is unacceptable either syntactically or semantically by the native speaker,
it is marked with *; when the expression is odd but not necessarily unacceptable, it is marked with ?.
2
In addition to 被 bei, there are other passive markers such as 讓 rang, 叫 jiao, 給 gei, 挨 ai, 遭
(受) zao(sho), 為 wei...所 suo… in Mandarin. Since each of them has their own subtle lexical meanings,
each passive marker has their preferred collocations and differs in their respective constructional meanings.
For example, when 讓 rang ‘let’ which bears the meaning of permission is used in passive, it embarks
the nuance that the patient gives permission to the agent’s action (Li & Thompson 1981), and therefore
implies a further meaning of causation (Li & Thompson 1981, Huang & Shi 2016). e.g. 我讓他給騙了
78

sentence (Huang & Shi 2016:467). In the literature, the former construction like (4a) is called long passive,
while the latter construction like (4b) is called short passive (Li & Thompson 1981, Huang & Shi 2016).
From the fact that long passive and short passive are basically interchangeable, we can say that the
agent phrase in Chinese passive is basically optional (Chao 1968, Li & Thompson 1981, Huang & Shi
2016). Similar as English passives, when the agent is clear and recoverable from the previous context, or
on the contrary when the identity of agent is totally unknown or redundant, the corresponding nominal
phrase can be omitted. (Huang & Shi 2016) The difference is that in Mandarin only the nominal phrase is
dropped off, while the whole prepositional by-phrase disappears when ellipsis happens in English passives
(Li & Thompson 1981).
As for the predicate in Chinese passives, for most of the time, it is a two-argument transitive verb of
activity, accomplishment or achievement, and sometimes a three-argument ditransitive verb (Chao 1968,
Li & Thompson 1981, Huang & Shi 2016, Liu etc. 1996). Since passivization is “an operation of
‘downgrading’ the element that would otherwise have been the subject (O’Grady 2001)”, it is more likely
to suppose the predicate will be representing a transitive relation involving at least two participants,
canonically agent and patient. By “demoting” the doer of the action, the affected participant or other
elements can be brought up to the focus of the ongoing discourse. By doing so, the predicate of the passive
sentence becomes intransitive; therefore, passivization can also be deemed as a detransitivizing (valency-
decreasing) operation.

2.3. Properties of bei passive in Mandarin

2.3.1. Adversity
Apart from typical transitive verbs, verbs with lower transitivity can also be predicates of passive in
Mandarin. According to Li & Thompson (1981), experiencer verbs (verbs concerning one’s cognition and
perception and often requiring an experiencer as subject for their active sentence) can occur in bei passive,
but the passive sentences usually have to depict an adversative outcome.

(5) 張三 被 人 看見 了。
zhangsan bei ren kanjien le
Zhang San BEI people see CRS

‘Zhang San was seen by others.’

(Li & Thompson 1981:496)

wo rang ta gei pien le ‘I was deceived by him, I let him deceive me.’ This research focuses on the most
used (according to Xiao (2016)) passive marker 被 bei only.
 Papers in Linguistics Science, No. 26 (2020) 79

As for the semantic and pragmatic characteristics, “misfortune, adversity” seems to be the prototypical
constructional meaning for passive in Mandarin (Chao 1968, Li & Thompson 1981). Since Mandarin is a
topic-prominent language, when the patient is focused as the topic of the ongoing discourse, it is preferred
to raise the patient argument to the topic position while other elements remain unchanged and the verb
remains in active voice. Unless the patient is adversely affected, passive construction is rather restricted.

(6) ??這本 書 已經 被 出版 了。
zhe-ben shu yijing bei chuban le
this-CL book already BEI publish ASP/CRS

‘This book is already published.’

(6’) 這本 書 已經 出版 了。
zhe-ben shu yijing chuban le
this-CL book already publish ASP/CRS

‘This book is already published.’

(Li & Thompson 1981:498)

2.3.2. Disposal
Nonetheless, it is not the case that all verbs representing an unfortunate or an adverse situation can form
passive in Mandarin. Compare (7) and (7’) below.

(7) ?我被 他 氣 了。
wo bei ta qi le
I BEI he anger ASP
‘Lisi is irritated by him’
(7’) 我 被 他 氣 得 頭 都 昏 了。
wo bei ta qi de tou dou hun le
I BEI 3sg anger CSC head all dizzy CRS
‘I was angered by him to such an extent that my head got dizzy.’
(Li & Thompson 1981:502)

While 氣 qi ‘to anger, to irritate’ is a transitive verb describing the patient experiences an undesirable
situation (‘to be irritated’), (7) is still considered unacceptable owing to its lack of the sense of “disposal”.
According to Li & Thompson (1981), bei passive also bears the constructional meaning of “disposal”
similar to 把 ba (direct object-introducing preposition) construction, in which the patient is “dealt with,
80

tackled with” by the agent using some kind of means. In comparison to (7), 氣 qi in (7’) is followed by a
complex stative construction clause (頭都昏了 tou dou hun le ‘(my) head got dizzy’) describing the
manner or degree of the resultative action or state caused by the predicate. Since the meaning that the
patient is “dealt with” is implied, it legitimates the use of bei passive construction in (7’).

2.4. Other variants of bei passive

2.4.1. Passives formed by intransitive verb
Huang & Shi (2016:470) mentions a rare case that an intransitive verb become the head of the predicate in
bei passive. The verb 走 zou ‘go, walk’ is an intransitive verb and cannot take a direct object or appear in
the ba construction. However, since the verb forms a resultative compound to indicate a change of status,
like 走尽 zou jin ‘walk-exhaust’ in (8), it can appear in a bei passive. It is consistent to the constructional
meaning of “disposal” mentioned above, since the resultative change of state implies the patient has been
dealt with.

(8) 路 好像 被 人 走 尽 了。
lu haoxiang bei ren zou jin le
way seem BEI people go exhaust ASP/CRS

‘It seemed that the pathways had all been fully explored by others.’
(Huang & Shi 2016:470)

2.4.2. Indirect passive (Adversative passive)

There is another possibility that a bei passive can take an intransitive verb. See the example below:

(9) 被 他 这麽 一 坐，我 什麽 都 看不见 了。

bei ta zheme yi zuo, wo shenme duo kan-bu-jien le
BEI he this one sit, I what all cannot-see ASP
‘With he sitting like that, I cannot see anything.’
(Yang 1989:333)

(9) is an example similar to the indirect passive (or adversative passive) in Japanese3. Indirect passive in
Japanese is a well-known passive construction in which an experiencer becomes the subject, while there
is no corresponding active sentence with the experiencer being realized as the direct object. (O’Grady

3
However, indirect passive is seemingly more restricted in Mandarin than in Japanese. (Yang 1989:332)
 Papers in Linguistics Science, No. 26 (2020) 81

2001:122)
What indirect passive depicts is that when an event happens, a participant outside of the event
indirectly receives negative influence and ends up in an adverse circumstance. For example, owing to the
intransitive event 他这麽一坐 ta zheme yi zuo ‘he sit like that’, the subject of (9), 我 wo ‘I’, is
negatively affected: being blocked and unable to see anything well. From the fact that the affected
participant is outside of the intransitive event and is added up to form an indirect passive, there will be no
corresponding active sentence for (9) with the exact same number of arguments. Takami (1995) mentions
that indirect passives can only find a corresponding active sentence without the subject of the indirect
passive sentence, as shown in (10) and (10’).

(10) 僕は こどもに 泣かれた。

boku-wa kodomo-ni nak-are-ta
I-TOP kid-DAT cry-PASS-PAST
‘I was adversely affected by the kid’s crying.’
(10’)こどもが 泣いた。
kodomo-ga nai-ta
kid-NOM cry-PAST
‘The kid cried.’

2.5. The passive marker 被 bei

Before we move on to the new usage of bei passives in Mandarin, the quality and the role of 被 bei will
be briefly summarized in this section. According to Wang (1980), in Ancient Mandarin Chinese, 被 bei is
functioned as a verb with the meaning of “to cover, to receive, to suffer”, and is often used in adversative
situation. But in modern Mandarin Chinese, views on the grammatical category of 被 bei in canonical
passives are divided by scholars. Some insist that 被 bei remain as a verb as in Ancient Mandarin (Hong
1956, Hashimoto 1987), while most of the researchers seem to deem it as a preposition (Chao 1968, Li &
Thompson 1981, Zhu 1982), or a preposition in long passive (for it is taking the agent NP) and a particle
which functions as a passive marker in short passive (Lü 1980, Shi 1997, Huang & Shi 2016). There are
also some positions regarding 被 bei as fully grammatical marker without inherent meaning (McEnery et
al. 2006, Li 2007). However in this study, we will not go deeply into the discussion of the grammatical
category of 被 bei in canonical passives.

3. The new “bei-XX” construction

In recent years, a new usage of bei passive seems to allow predicates with low transitivity including one-
82

argument intransitive verbs, adjectival verbs (adjectives), and even nouns, and it will be referred as bei-
XX construction hereafter. See some of the examples below:

(11) 李明哲 3 月 19 日 進入 中國 後 「被失蹤」4 。

Limingzhe sanyue-shijiuri jinru zhongguo ho bei-shizong
Li Mingzhe March-19th enter China after BEI-disappear

‘After Li Mingzhe entered China on March 19th, he was “bei-disappear”.’

(extracted from https://newtalk.tw/news/view/2017-03-24/83438, last access on 2019/12/25)
(12) 城市 评选， 你 “被幸福” 了 吗？
chengshi pingxuan ni bei-xingfu le ma
city evaluate you BEI-well-being ASP Q

‘(Regarding) the evaluation and selection of (your) living city, are you “bei-well-being”?’
(extracted from http://news.cctv.com/special/meir/mulu/1228/index.shtml, last access on 2019/12/26)
(13) 男子 “被精神病” 诉讼 十年 终 胜诉
nanzi bei-jingshenbing susong shinien zhong shengsu
Man BEI-mental disorder lawsuit ten year finally win the court case
‘The man was “bei-mental disorder”. After ten years of lawsuits, (he) finally won the court case.’
(extracted from http://www.xinhuanet.com/2018-11/09/c_1123686231.htm, last access on 2019/12/28)

失蹤 shizong ‘to be missing, to disappear’ in (11) is an intransitive verb with only one argument (the actor),
幸福 xingfu ‘well-being’ in (12) is an adjective or an adjectival verb, and 精神病 jingshenbing ‘mental
disorder’ in (13) is a noun. All of these are usually not expected to be the predicate of a canonical bei
passive sentence. This kind of novel construction which allows atypical predicates appears and spreads
like wildfire on the Internet in the recent ten years, and its usage can even be seen on news and other media
((11), (12), (13) are all extracted from online news).
As previous literature reports (Li & Li 2015, Xiao 2016), this usage was originally triggered by the
death of an inmate named Li Guofu in China, 2008. Without any possible motivation to commit suicide,
Li is found dead in the prison and announced as “committed suicide” by the administration. Seeing the
official have no intention to disclose the actual cause to the death and therefore feeling upset, Chinese
Internet users coin the expression 被自杀 bei-zisha ‘bei-suicide, to be (falsely) said to commit suicide’

4
The quotation marks「」or “ ” are not added by the author but used in the original text. It shows that
the language user recognizes that it is a non-canonical expression, and it is intended to specify its
markedness. English translations of (11) to (22) are made by the author and are not in the original text,
since Lu (2013) provides only Japanese translation for the Chinese sentences.
 Papers in Linguistics Science, No. 26 (2020) 83

to express their unbelief and dissatisfaction. After its first appearance, 被自杀 bei-zisha is analogized and
the bei-XX (XX can be filled with any atypical predicates for bei passive mentioned above) form is copied
and proliferated, which makes 被 bei become a buzzword in the recent decade (Li & Li 2015:235, Xiao
2016:80).
In this section, literature concerning this newly emerged and apparently non-canonical construction
will be briefly summarized. The semantic/ functional characteristics and syntactic/ formal features of bei-
XX construction will be demonstrated in order to show its differences compared to canonical bei passive.
Remaining problems will also be pointed out and discussed.

3.1. Semantic/ Functional features

Lu (2013) collects instances of bei-XX construction from online news and websites in China, and
differentiates two categories based on their distinctive meanings: “being forced” type (“被強迫”類) and
“being (falsely) reported/said” type (“被說成”類).

3.1.1. “Being forced” type

First, let us look at the “being forced” type. As its name indicates, the “being forced” type characterizes
with the mere participant, which is located in the subject position, being forced or coerced to do the action
the predicate refers or to achieve the resultive state derived from other’s involvement. (14)-(17) belong to
this category.

(14) 钢铁大亨发妻 “被离婚”。

‘The wife of the steel entrepreneur is “bei-divorce” (‘forced to get divorced’). ’
(Lu 2013:24)
(15) 我在不知情的状态下，就莫名其妙地 “被捐款”。
‘Not knowing the whole situation, I was mysteriously “bei-donate” (‘forced to donate’).’
(Lu 2013:24)
(16) 近日某网站做起了 “被全勤” 现象调查，数据显示，超过三分之一的网友没有享受过带薪
年假，这种逼你拿全勤奖的公司不在少数。
‘Lately, some website made an investigation on the “bei-full-time-attendance” (‘being forced to be
full-time attendant at work’) phenomenon. The data shows that over one-third of people have not
enjoyed paid annual leave before. This kind of companies which force you to get the full-time
attendance prize are quite a lot.’
(Lu 2013:25)
84

(17) 两个月募捐 38 亿 神木县 “被慈善”。

‘3.8 billion donated in two months, Shenmu county was “bei-charity” (‘forced to do charity’).’
(Lu 2013:25)

Take (14) for example, by using the expression “被离婚” bei-lihun ‘bei-divorce’, it means that not
knowing the whole situation and without her agreement, the wife of the steel entrepreneur (钢铁大亨发
妻) is forced to end her marriage to her husband unwillingly.
Lu (2013) seems to regard this usage as interchangeable with passivized causative sentences,
assuming that 被 bei-XX is alternative of 被迫 XX bei po XX ‘be forced to XX’ without verbalizing 迫
po ‘to force’, so that “被离婚” bei-lihun ‘bei-divorce’ can be deemed equivalent to 被迫离婚 bei po lihun
‘be forced to divorce’. Yet comparing (14) to (18) and (18’), bei-XX and passivized causatives (被迫 XX
bei po XX) should be considered different constructions owing to their subtle semantic dissimilarity:

(18) ?得知丈夫外遇後，鋼鐵大亨髮妻 “被離婚”。

‘After knowing her husband’s affairs, the wife of the steel entrepreneur is “bei-divorce”.’
(18’) 得知丈夫外遇後，鋼鐵大亨髮妻被迫離婚。
‘After knowing her husband’s affairs, the wife of the steel entrepreneur is forced to divorce.’

If it is under the circumstance that the wife of the steel entrepreneur is aware of any knowledge on the
inside that leads to their divorce, it is seemingly more difficult to use the bei-XX construction. Therefore,
while the nuance of “not knowing the whole situation” is not necessary for passivized causatives, it seems
to be a prominent semantic feature for bei-XX.

3.1.2. “Being (falsely) reported/said” type

The other type of the bei-XX construction is named “被說成”類 (Lu 2013:25), the “being (falsely)
reported/ said” type. This type of bei-XXs share a distinctive meaning that someone is reported or said that
he/she has done an action or been in a certain state no matter if it really happened or not, and usually what
happens in reality is the opposite.

(19) 职工平均工资每统计一次，我的工资就 “被增长” 一次。

‘Every time when the average payment of employees is calculated, my pay is “bei-rise” (‘said to
be raised’).’
(Lu 2013:25)
 Papers in Linguistics Science, No. 26 (2020) 85

(20) 神奇的就业率让学校为我代签就业协议书，不明真相中我 “被就业”。

‘Mysterious employment rate made the school sign the employment agreement for me. While not
knowing the truth, I was “bei-get a job” (‘said to have got a job’).’
(Lu 2013:25)
(21) 在有关部门的统一部属下，当地民众一夜之间 “被小康”。
‘Under the deployment of related departments, local people were “bei-comparatively-good”
(‘announced as comparatively good’) in one night.’
(Lu 2013:26)
(22) 笔者想知道的是，这其中有多少是 “被网瘾” 的?
‘The writer wants to know that, (among those teenagers) how many of them were “bei-Internet-
addict” (‘considered to be suffering Internet addiction’).’
(Lu 2013:26)

Although the regular pay of the employee in (19) is not raised, for some reasons (to meet the standard of
welfare regulations for example) the statistics are manipulated and the average payment appears to be
raised. The graduate in (20) gets a job without knowing it himself and finds out that his school signs the
employment agreement for him. Local people in (21) are reported as comparative wealthy while some of
them are in fact in poverty; ordinary teenagers are said to be Internet addicted but they are actually not in
(22). From the above instances, users of the bei-XX construction intentionally choose the predicate that
does not reflect reality, in order to indirectly express that the mere participant of the event receives negative
influence (damage and harm) when the wrong information is being reported and spread. (11)-(13) above
also belong to this category.
Reporting an untrue situation of someone and making him stand as a victim in an adversative
situation as the constructional meaning, the second usage of 被 bei-XX is claimed to be interchangeable
to 被說成 XX bei shuo cheng XX ‘to be said XX’ by Lu (2013), and XX is what is being falsely rumored
about. Therefore, 被精神病 bei-jingshenbing ‘bei-mental-disorder’ in (13) for instance is 被說成精神
病 bei shuo cheng jingshenbing ‘said to become having mental disorder’ without realizing 說成 shuo
cheng ‘said to be/ become’. Yet it is not explained why 說成 shuo cheng is truncated, or why we can
recover 說成 shuo cheng instead of other predicates. Even if the expression can be interpreted as “said to
be XX”, there can be seen trivial differences in meaning when the same construction is placed in different
sentences.5

5
Lu (2013) admits that there are expressions that cannot be clearly categorized in the two types of
meaning. For instance, 被自杀 bei-zisha ‘to be “suicided”’, which is categorized as the “being (falsely)
86

3.1.3. Relationship of “being forced” and “being (flasely) reported/ said” types
Furthermore, note that the same bei-XX may be ambiguous between the first and the second type proposed
by Lu (2013) and can only be distinguished by context. For example, while “被离婚” bei-lihun ‘bei-
divorce’ in (11) is categorized of the “being forced” type, “被离婚” in (23) is better interpreted as of the
“being (falsely) reported/ said” type (since Beckham does not get divorced in reality):

(23) 贝克汉姆「被离婚」英媒体见怪不怪将消息屏蔽。
‘Beckham is “bei-divorce”. British media block the (fake) news without shock or surprise.’
(Chen 2017:65)

3.1.4. Implicit causer

In addition to the differentiation of the “being forced” type and the “being (falsely) reported/ said” type
proposed by Lu (2013), a prominent property of the bei-XX construction is the existence of an implicit
causer. Huang & Shi (2016:482) concludes that the overall meaning of this novel construction is “the
reporting of false information about the subject NP, including the padding out of official figures and other
types of cover-ups, such as the cause of unexplained deaths and disappearances”. The source of the untrue
rumor is usually official, the media or other authorities, and therefore the cause of the adversative situation
is intentionally wiped out in the expression and cannot be recovered. (Syntactic features of bei-XX will be
further discussed in 3.2.) In fact, Xiao (2016:90) points out that bei-XX as an innovative is originally to
“avoid notice by online censors” in China, since bei-XX “conveys a subversive message that reveals social
absurdity and speaks for the disadvantaged and deprived.”

3.1.5. Expressive function

Last but not the least, Chen (2017) mentions that bei-XX also demonstrates extension in function compared

reported/ said” type, invites different interpretations in the following two sentences:
(i) 白岩松 “被自杀”多次。
‘Bai Yensong is “suicided” several times.’
(ii) ...李国福在监狱医院 “被自杀”。
‘Li Guofu was “suicided” in the prison hospital.’
(Lu 2013:27)
白岩松 Bai Yensong in (i) is someone who does not even die but just becomes the victim of an untrue
rumor that he has killed himself, which is similar to (23). While 李国福 Li Guofu in (ii) did die in the
prison hospital, the reason of his death is unclear yet it was announced as suicide, which is not believed by
the user of the expression.
 Papers in Linguistics Science, No. 26 (2020) 87

to canonical passive. While the meaning of canonical passive is “someone or a thing is caused to experience
some state of affairs” no matter if it is positive, neutral or negative, the extended bei-XX construction is
restricted to negative situations and adds the nuances that someone is rumored on news or is forced without
his/her willingness. In other words, comparing to prototypical passives which are representational in
function, bei-XX is more of expressive function, to express feelings like anxiety, sarcasm, mockery,
uncertainty, or unbelief.

3.2. Syntactic/ Formal features

The syntactic structure of the new bei-XX construction can be formalized as follows:

(24) New bei-XX: NP 被 bei ∅ XX(Vi, VN, AP, NP)

Short passive: NPPatient 被 bei (NPAgent) Vt
(Lu 2013, Xiao 2016)

Although the structure of bei-XX construction looks similar to prototypical short passive, three syntactic
features differ the new bei-XX from canonical passives. First, no NP can be recovered between the passive
marker 被 bei and the following predicate XX in this new construction, and there is also no corresponding
active sentence for bei-XX6, as illustrated in (25) and (25’). In contrast, as for short passive, an agent NP
can still be added back to make a long passive, and corresponding active sentence can also be obtained by
the rearrangement of the grammatical relations of the arguments, as shown in (3) and (4). Indirect passive
discussed in 2.4.2 also does not have a corresponding active sentence, yet an NP is realized after 被 bei in

6
Since the cause of the subject NP’s adversative situation is intentionally wiped out in bei-XX, it is not
expected to be realized in the sentence as mentioned in 3.1.4. However, there are a few instances that a
causer NP seems to appear after the passive marker 被 bei.
(iii) 宫鲁鸣又一次被媒体 “解职” 了。
‘Gong Luming is “dismissed” by the media again.’
(Chen 2017:64)
However, the sentence above is not the target bei-XX in this study, since it has the syntactic structure of a
long passive. Not only can its corresponding active sentence (媒体又一次 “解职”了宫鲁鸣) be made
(comparing to (25’)), the subject 宫鲁鸣 Gong Luming bears the thematic role of patient related to the
predicate 解职 ‘dismiss’, which is a syntactic feature of canonical passive. What is confusing is the
sentence above shares a similar meaning with the new bei-XX: being falsely reported by the media. Chen
(2017) explains that it is achieved through pragmatic implication: the media is not the boss of Gong
Luming and cannot disemploy him, so based on the optimal relevance, the media is just spreading rumors.
The usage of 解职 ‘dismiss’ is also accompanied by quotation marks to specify it is a marked expression.
88

indirect passive.

(25) ...没有自杀动机的李国福在监狱医院 “被自杀” 。

‘Li Guofu, who has no motivation to commit suicide, was “bei-suicide” in the prison hospital.’
(Lu 2013:22)
(25’)
passive sentence:
*没有自杀动机的李国福在监狱医院 “被{书记张治安/警察}自杀” 。
‘Li Guofu, who has no motivation to commit suicide, was “bei-suicide” {by secretary Chang
Zhian/ the police} in the prison hospital.’
(Lu 2013:24)
active sentence:
?{书记张治安/警察}在监狱医院“自杀”了没有自杀动机的李国福 。
‘{Secretary Chang Zhian/ the police} “suicided” Li Guofu, who has no motivation to commit
suicide, in the prison hospital.’

Second, possible predicates for the new bei-XX construction include intransitive verbs (Vi), VN
construction (which structurally cannot take a direct object), adjectival verb (adjective) phrases (AP), and
noun phrases (NP), which are mostly those do not appear in canonical passives because of their low
transitivity. They are mostly “actions taken by people on their own initiative or at their own will […] or
situations arrived at spontaneously” (Xiao 2016:90), and have only one argument. In comparison,
predicates of canonical passives are often of higher transitivity, presupposing they have at least two
arguments (agent and patient). Since transitivity is manifested when a patient receives force or influence
from the agent and experiences changes, and passivization aims at demoting the agent in order to switch
to other participants’ (patient for most of the time) perspective as mentioned in 2.1, so it is plausible that
passivization for predicate with only one argument in nature will be rare and non-canonical. Compared to
indirect passives (2.4.) which add an additional affected participant, the bei-XX implies the existence of a
causer (however implicit since it cannot be recovered) leading to an adversative situation. Therefore, Xiao
(2016:90) concludes that “with bei preceding them, the strings become ungrammatical syntactically and
absurd semantically,” as discussed in 3.1.4.
Last but the most distinctive syntactic feature of bei-XX is that the subject of bei-XX construction is
an affected participant of the adversative situation but not the patient of the predicate XX. On the contrary,
the subject can be the doer of the action which the predicate XX referred to. Take 被捐款 bei-juankuan
‘bei-donate’ in (15) for example, Lu (2013) categorizes it as the “being forced” type, which is
 Papers in Linguistics Science, No. 26 (2020) 89

interchangeable to passive of causative construction 被迫捐款 bei po juankuan ‘be forced to donate’.
The subject 我 wo ‘I’ is the doer of the action 捐款 juankuan ‘to donate’, and the patient of the
unverbalized co-verb 迫 po ‘to force’ at the same time. Therefore, 我 wo ‘I’ in (15) is the one who
“donates” money, even though the “donation” is done without the acknowledgement and against the
willingness of the subject NP. Though it is questionable whether the bei-XX construction and passive of
causative construction can be regarded as equivalent (as discussed in 3.1.1), it is clear that subject in
canonical passives is definitely not the doer of the action but the receiver, while it is the participant who
performs the action that becomes the subject of bei-XX construction.7

3.3. 被 bei in the new “bei-XX” construction

Before we turn to the next section for analysis of bei-XX in the thesis, Xiao (2016)’s analysis on the
grammatical category of the passive marker 被 bei will be briefly discussed. Xiao (2016)’s approach is
based on the theory of grammaticalization proposed by Hopper & Traugott (2003). Grammaticalization is
“the process whereby lexical items and constructions come in certain linguistic contexts to serve
grammatical functions, and once grammaticalized, continue to develop new grammatical functions.”
(Hopper & Traugott 2003) A unidirectionality can be seen in grammaticalization, from a content item, to
a grammatical word, clitic, and further inflectional affix. The result of Xiao (2016)’s analysis follows this
unidirectionality: It is claimed that 被 bei has become a pseudo-/ semi-/ quasi- prefix, which has fixed
position but no longer carry NP (a feature of a preposition), experiences semantic bleaching (it lost the
meaning of “passive experience caused by a concrete and recoverable source”) yet not completely loses
the original semantic meanings (representing an adverse situation), and even gains some new meanings
(“being compelled, forced, faked, coerced by unknown, uncontrollable force” and a feeling of being
helpless or grievously wronged by an unknown, unfathomable source). 被 bei acts like prefixes which
make intransitive verbs, yet is only to polysyllable words.

4. Theoretical framework of Cognitive Linguistics and analysis

This section presents a cognitive linguistic analysis of the new bei-XX passives described in the previous
section. The action chain model will be employed to represent the event construal denoted by certain
grammatical constructions. After showing how transitives, intransitives, causatives and passives can be

7
As for the “being (falsely) reported/ said” type, Lu (2013) claims that the subject NP is not the doer of
the action referred by the predicate XX since the described event does not happen in reality. For instance,
李国福 Li Guofu in (25) is not the doer of the action 自杀 zisha ‘to commit suicide’ but merely the
affected participant of “rumor, say or report.” However, it can also be analyzed as it is a rumor or false
information about Li Guofu that he commits suicide, and Li Guofu is the doer in the rumored proposition.
90

captured by this model, this chapter proposes the event schemas for the two types of bei-XX passives, i.e.,
“being forced” type and “being (falsely) reported/ said” type, and characterizes these novel usages of
Mandarin passives in terms of the function of passives.

4.1. The “billiard-ball model”, action chain, and prototypical transitive relation
A central claim of cognitive linguistics is that all grammatical constructs in linguistic expressions are
symbolic units with a phonological pole and a semantic pole. Consequently, even elements like inflectional
morphemes which was considered with mere grammatical functions, or rather abstract units such as
grammatical constructions, have some sort of conceptual input. (Langacker 1991, 2008) Before we discuss
how cognitive linguistic approaches deal with linguistic phenomenon regarding transitivity and voice, a
conceptual model which plays as the basis of semantic characterization for grammatical constructs in
linguistic coding will be first mentioned: the “billiard-ball model” and action chain.
According to Langacker (1991), we conceive of the world as being composed of numerous “discrete
objects”. With energy supplied from the inside or received from the outside, these objects move from the
location they exist, and transmit energy by physical contact to another object, and perform interaction with
each other. The interactions are defined as several events, and the involving objects are considered
participants. This idealized force-driven model is called the “billiard-ball model” and is claimed
substantial to our thought processes, which makes it one of the conceptual archetypes within cognitive
linguistic framework.
However, as we cannot pay attention to all the interacting objects at the same time by the limit of
perception, linguistic coding for a situation is also selective to certain objects in focus. From the network
of interacting objects as schematized as (a) in Figure 1, an asymmetric energy-transferring relation is
especially of our concern: one participant has physical contact and transfers its energy (illustrated by
double arrow in Figure 1) to another participant, and then continuously to the third until the transmission
stops. The unidirectional energy-transmission is called “action chain” and can be sketched as (b). Yet a
finite sentence cannot cover all parts of the action chain. Therefore, only the participants in the scope of
predication specified by the finite sentence are selected (as shown in (c)), and further participants will be
profiled according to their cognitive prominence (as shown in (d)). This is the conceptual structure invoked
by a finite clause.
 Papers in Linguistics Science, No. 26 (2020) 91

Figure 1. The “billiard-ball model” and action chain (Langacker 1991:215)

As for the participants, there are two main archetypes when constructing an event: an agent and a patient,
which are consistent to the previous studies concerning thematic roles. A prototypical agent is the source
of energy and is the participant volitionally carrying out action, resulting in having physical contact with
the next participant, and transmitting the energy. On the contrary, an archetypal patient is the opposite
participant receiving the energy and experiencing changes. As a result, a prototypical transitive event can
be defined as the profiled action chain starting from a prototypical agent transferring energy to a
prototypical patient which eventually have a spatial change or change of state. Based on the above ideas,
Taniguchi (2005:34) generalizes the prototypical transitive relation (P-transitive relation) as below:

Figure 2. P-transitive relation (Taniguchi 2005:34)8

8
Taniguchi (2005) combines Croft’s causative construction into Langacker’s action chain to make a
distinction for segments on a profiled action chain representing the P-transitive relation. The transmission
of energy from the agent to the patient (illustrated as a double arrow) corresponds to a CAUSE segment,
while the locational or stative change which the patient experiences corresponds to the CHANGE segment,
illustrated by a single arrow. The last circle in a square on the end of the action chain is schematized for
the end location/ state of the patient object. The agent on the head of the action chain is shaded to
distinguish itself from other participants.
92

But how could the grammatical relations in a finite clause, that is, the choice of subject, (direct) object and
oblique be explained by using action chain? Since not always is the agent realized as subject, it is said that
it is difficult to solely rely on thematic roles to account for the grammatical structure of a sentence.
Langacker (1991) claims that in fact as long as the scope of predication and the profiled portions of the
action chain are decided, the grammatical relations of a finite clause are predictable. Consider an event
that Floyd use the hammer to break the glass on the window. There are three prototypical thematic roles in
this event: an agent (Floyd), an instrument (the hammer) and a patient (the glass). Energy is transmitted
from the agent, through the instrument, and affecting the patient, as Figure 3(a) illustrates. When the scope
of predication is determined, different parts of the action chain can be profiled, and possible linguistic
codings are instantiated in (26) and (27).

(26) a. Floyd broke the glass (with the hammer).

b. The hammer (easily) broke the glass.
c. The glass (easily) broke.
(27) a. Floyd hit the glass (with the hammer).
b. The hammer (easily) hit the glass.
c. Floyd hit the hammer against the glass.
(Langacker 1991:216)

Figure 3. Action chains corresponding to (26) and (27) (Langacker 1991:217)9

9
The squiggly arrows in the rightmost participant of the action chain in Figure 3 indicate changes which
the patient experiences after receiving the transferred energy.
 Papers in Linguistics Science, No. 26 (2020) 93

Action chain (b) in Figure 3 corresponds to (26a) and (27a), where the agent is realized as the subject and
the patient as direct object. Action chain (c) corresponds to (26b) and (27b), where the instrument is
realized as the subject. Action chain (d) can be represented by (26c), in which the patient is the only
participant being profiled and therefore becomes the subject. On the contrary, in action chain (e) only the
patient is not profiled, and the corresponding (27c) has an agent-subject and an instrument-object.
It seems random that any participant is possible to be the subject, yet the selection of subject and
direct object is actually determined according to the profiled portion of the action chain. The first
participant of the profiled portion, that is, “the head” of the action chain is realized as the subject of the
clause, while the last participant, “the tail” of the action chain as the direct object. This tendency results
from the accommodation (or harmonization) of unidirectionality of the “natural path” in an action chain.
(Langacker 1999: 363, Taniguchi 2005:37-38) From the perspective of energy transmission, the agent is
often the starting point and the initial constituent of the action chain and deliver energy to its downstream
participants. From the perspective of cognitive prominence, the agent is the trajector which the speaker
will first access to. As a result, the agent often appears in the initial position of the finite clause, which
makes it the subject. As for the patient, it is the tail located in the downstream of the action chain, which
makes it the second salient participant following the agent. Therefore, the patient is often realized as the
direct object. If an agent and a patient are profiled in the action chain for an unmarked prototypical
transitive event, although other participant like instrument is also in the downstream of the action chain
after agent, it cannot be realized as direct object but oblique since its cognitive prominence is relatively
lower to the patient. “With the hammer” in (26a) is such a case. Nonetheless, if the patient is not profiled
in the action chain ((e) in Figure 3), the instrument as the tail is also likely to be the direct object, as shown
in (27c).
In conclusion, the schema of the subject can be characterized the most salient object located in the
head of the natural path, while the schema of the direct object is the secondarily salient object and is located
in the downstream of the natural path following the agent.

4.2. Cognitive approaches to voice

4.2.1. Direct passive10 and indirect passive
However, passives seem to violate the schema summarized above. Compare the active and passive
sentence in (28) and their different corresponding action chains in Figure 4.

10
In comparison to the indirect passive mentioned in 2.4.2, direct passive refers to passives having a
corresponding active sentence with the same number of arguments, especially to the canonical passive in
which the subject is the patient and is realized as the direct object in its corresponding active sentence, as
shown in 2.1.
94

(28) a. John broke the glass.

b. The glass was broken (by John).
(Taniguchi 2005:39)

Figure 4. Action chain of active and corresponding passive (Taniguchi 2005:40)11

While (28a) and (28b) seem to depict the objectively same event, in cognitive linguistic framework they
show the difference in choosing which the most prominent participant (that is, the trajector) of the event
is. (Taniguchi 2005:40) As we can see in Figure 4(b), the participant which is chosen to be the trajector is
not the head of the action chain (the agent) but the tail (the patient). The apparent violation of “natural
path” makes passive a marked expression, and it must be motivated by some kinds of functions. As
mentioned in 2.1, passivization does raise the direct object to the subject position in the corresponding
passive sentence, yet according to Ramchand (2013), the most important function of passivization is “the
demotion of the rather winning argument”. Shibatani (1985) also characterizes passivization as “agent
defocusing”, which the most salient agent is defocused and other participants (often the patient) are
therefore being focused instead.
It is worth mentioning that passivized sentences are intransitive since there is no other participant
following the patient in the downstream of the action chain. The agent will not be profiled as landmark
and become the direct object in the passive sentence, for the direction of the energy transmission cannot
be reversed and the agent is still in the upstream even if it is not profiled. Thus, as a marked grammatical
construction, passives still support the “natural path” in the action chain as we discussed above.
But how about indirect passive? Can passive construction like indirect passive which has no

11
The diagrammatic representation in Taniguchi (2005) differs from those in Langacker (1991) in that
only the most salient participant, that is, the trajector is represented in bold lines. Langacker tends to use
bold lines for all the profiled portions within the scope of predication, which can be seen in Figure 3. The
dotted line connecting the trajectors stands for correspondence.
 Papers in Linguistics Science, No. 26 (2020) 95

corresponding active sentence be explained by using active chain? As mentioned in 2.4.2, indirect passive
can only find corresponding active sentence by removing the subject, the adversely affected participant
(the experiencer).

(29) 僕は こどもに 泣かれた。

boku-wa kodomo-ni nak-are-ta
I-TOP kid-DAT cry-PASS-PAST
‘I was adversely affected by the kid’s crying.’
(= (10))
(29’)こどもが 泣いた。
kodomo-ga nai-ta
kid-NOM cry-PAST
‘The kid cried.’
(= (10’))

The active sentence without the experiencer can be conceived as an autonomous unitary event alone like
(29’) (minimal autonomous event conception will be discussed in 4.2.2). Taniguchi (2005:307) concludes
that the autonomous single event becomes the cause of the mental change (being adversely affected) of the
subject of the indirect passive sentence. She proposes the “extensive transitive relation (E-transitive
relation)” as the conceptual basis for indirect passive in Japanese. What differs E-transitive with P-
transitive is that while the CAUSE segment in P-transitive is composed of an agent, the CAUSE segment
in E-transitive includes an event. To put it in another way, an event plays the role of what an agent does in
P-transitive, to cause other participant’s changes. Since this event, the CAUSE-event, is causing other
participant’s changes, it can be construed holistically and is required to be an autonomous dynamic event.
See how E-transitive is used to explain the event which an indirect passive depicts in Figure 5.

Figure 5. Illustration of indirect Passive in Japanese (29) using E-transitive (Taniguchi 2005:308)
96

If we take (29) for example, the autonomous CAUSE-event こどもが泣く kodomo ga naku ‘The kid
cries’ is the cause to the change of mental state (to feel adversely affected) of the subject 僕 boku ‘I’.
Though indirect passive is based on E-transitive relation, it also profiles the tail (the experiencer) of the
action chain, which is an inherited feature from the canonical direct passive. Taniguchi (2005:308) suggests
that since E-transitive relation is extended from P-transitive relation, indirect passive can be deemed as
extensive construction of direct passive while the former is based on E-transitive relation and the latter is
based on P-transitive relation.

4.2.2. Complex event and causatives

In the 4.2.1 we have discussed the autonomous event in E-transitive relation. So in cognitive linguistic
framework what is the minimal, autonomous event conception? In a profiled process, an entity might
“either moves, has a mental experience, or undergoes internal change of state” (Langacker 1991:244),
which makes it either a mover, an experiencer or a patient. Langacker generalizes the above three
possibilities and uses a schematic conception, a theme, to conclude them. A thematic participant therefore
can be characterized as experienced some kind of change, no matter if it is internal or external. With the
minimum of conceptual content (and a single participant, the theme), the minimal, irreducible unit in a
profiled process is a thematic relation. A thematic relation can be further construed in different ways with
respect to the input of energy: in an absolute thematic relation, the input of energy is abstracted away; in a
self-induced thematic relation (S-thematic relation), energy is from the internal source of the mere
participant; while in an externally-driven thematic relation (E-thematic relation), the energy is supplied
from the outside. Schematized action chain for thematic relation in general and both S-thematic relation
and E-thematic relation is illustrated below.

Figure 6. Thematic relation in Taniguchi (2005:120)’s framework12

Figure 7. S-thematic relation and E-thematic relation (Taniguchi 2005:123)13

12
The dotted lines in Figure 6 represent that there are events in which the goal of movement will not be
clear if it is not specified by a prepositional phrase (an oblique). For example, John walked (to the station).
(Taniguchi 2005:121)
13
The double arrow in Figure 7(a) indicates that the energy source of S-thematic relation is within the
 Papers in Linguistics Science, No. 26 (2020) 97

A thematic relation is defined as the processual event which constitutes a single participant’s change of
location or state. In Taniguchi (2005)’s framework, a thematic relation (Figure 6) corresponds to the
CHANGE-LOC(STATE) segment compared to a P-transitive relation (CAUSE-CHANGE-LOC(STATE))
in Figure 2, which shows that a thematic relation is included in a P-transitive relation, or we can say P-
transitive relation is a thematic relation adding the source of energy which drives the change, the CAUSE
segment. In fact, the CAUSE segment relies on a thematic relation, while a thematic relation is
conceptually autonomous, which it does not rely on the CAUSE segment. (Langacker 1991:245)
As the nucleus of an event conception, a thematic relation can be construed absolutely without the input
of energy, or it can specify the source of energy and elaborate the event, making a complex event. By
adding additional layers of energy input (the CAUSE segment), the action chain can be expanded in any
desired length, as shown in Figure 8 below. The whole action chain can not only be viewed holistically as
a unitary complex event, but also be chunked into a series of several events.

Figure 8. Complexation of the event (Langacker 1991:255)

mere participant and not supplied from the outside. The dotted lines in Figure 7(b) represent the portions
of the action chain which are within the scope of predication yet is not profiled.
98

The successive incrementation of energy input can be utilized in the conception of causativization. Let us
first discuss causativization related to intransitive verbs. Intransitive verbs are monovalent, which means
they only have one participant to be their argument. In cognitive linguistic framework, the event they
represent are of thematic relations. Therefore, the causativization of a thematic relation can be deemed as
adding an additional CAUSE segment (an additional layer of energy input) to a thematic relation.
(Taniguchi 2005:270)
There will be two ways to code the causative event. One is called lexical causatives. After the
addition of a CAUSE segment to a thematic relation (CHANGE-LOC(STATE)), it will make the extended
action chain have similar conceptual structure of a P-transitive relation (CAUSE-CHANGE-LOC(STATE),
as shown in Figure 2). Therefore, the whole action chain can be holistically construed as a P-transitive
relation, and its linguistic coding can be represented by a finite clause with a transitive verb. For instance,
regarding the event of Bill died, if we specify that the energy input leading to the participant’s (Bill) change
of state (to become dead) is from another participant John, then we can also construe the whole event as
John killed Bill. Taniguchi (2005:270) mentions that lexical causative is reasonable for the causativization
of E-thematic relation since in nature E-thematic relation implies supplication of energy from the outside,
and it can also be justified from the fact that unaccusative intransitive verbs, which represents a E-thematic
relation14, often have a pairing transitive verb.
While the extended action chain in which a CAUSE segment is added onto a thematic relation can
be construed as a unitary event, it can also be chunked into several events to form a complex event. In
comparison to E-thematic relation, a S-thematic relation does not require external energy and hence is
conceptually independent. As a consequence, even adding a CAUSE segment onto a S-thematic relation,
the CHANGE-LOC(STATE) segment is chunked as an autonomous core sub-event in the whole extended
action chain, which makes the whole event a complex one. The complex event can be realized in a way of
using causative verbs (make, let, have, etc.) to form causative construction in English, which is called
periphrastic causatives. (Taniguchi 2005:270) For instance, an event demonstrating a S-thematic relation
John sang can be causativized as She made John sing.
Note that not only causative of S-thematic relation can be coded in the periphrastic causative fashion,
causative of E-thematic relation can also be coded with causative verbs in English, such as John caused
Bill to die. According to Kemmer and Verhagen (1994), periphrastic causatives are used in English when
the causation involves “some perceived mediacy”. On the contrary, “indirectness” seems to be not what
motivates periphrastic causatives in Japanese, as reported by Taniguchi (2005:272). Taniguchi (2005:274)

14
According to Taniguchi (2005:123), the differentiation of S-thematic relation and E-thematic relation
corresponds to the difference of unergative verbs and unaccusative verbs.
 Papers in Linguistics Science, No. 26 (2020) 99

proposes that as long as the thematic relation which the intransitive verb denotes can be construed as
autonomous, periphrastic causatives can be applied. And the key to the autonomy of the thematic relation
is the “subjecthood” of the subject of the intransitive verb. If the mere participant of the thematic relation
demonstrates “volition” to achieve the change by itself without external causer, or has “controlability” over
the change (being able to control the change of state), as the autonomous reading is possible, even
unaccusative verbs which denote E-thematic relation may have acceptable periphrastic causatives.

Figure 9. The Schema of Periphrastic Causatives in Japanese (Taniguchi 2005:274)

The “subjecthood” of the head of the original action chain can also be seen (or it is more apparent)
in causativies derived from transitive verbs. While the event represented by a transitive verb is already a
P-transitive, we can still add another participant to specify the source of energy. The new participant, the
primary agent or we can call it the causer, is added in the upstream of the original profiled action chain
and therefore is realized as the new subject. While the original agent becomes “the secondary agent”, and
is often marked by dative or prepositional phrase. Langacker (1991:256) calls the head of the original
action chain the “causative pivot”, of which energy is given from the primary agent in the upstream, yet
it is not completely thematic. The causative pivot still has the “responsibility” to certain degree of
initiating and providing energy to drive the thematic relation in the downstream, which is the characteristics
of “subjecthood”.

Figure 10. Action chain of causatives derived from transitive verbs

100

Yet no matter how a causative is coded linguistically, they still follow the “natural path” mentioned in 4.1:
The participant located in the head of the profiled action chain will be verbalized as subject, while the tail
participant will be verbalized as the direct object.

4.3. The new bei-XX construction

After reviewing the representative studies regarding voice in cognitive linguistic framework, the target,
the new bei-XX construction and related constructions will be discussed in this section. By inspecting the
two types of constructional meanings of bei-XX mentioned in Chapter 3 and applying the cognitive
linguistic conceptions in this chapter, a coherent explanation for bei-XX’s conceptual structure is expected
to be provided. The illustrations in this section will be following the framework and conventions of
Taniguchi (2005).

4.3.1. “Being forced” type

According to Lu (2013), the semantic feature of “being forced” type is characterized by the subject (the
mere participant) being forced or coerced to do or become XX, and the bei-XX construction of “being
forced” type can be deemed equivalent to passivized causatives 被迫 XX bei po XX ‘be forced to XX’.
For example, “被离婚” bei-lihun ‘bei-divorce’ in (14) is deemed as alternative of 被迫离婚 bei po lihun
‘be forced to divorce’ by Lu (2013).
Let us examine the two seemingly interchangeable expressions by applying the conception of action
chain. The processual predication 离婚 lihun ‘to get divorced’ in (30a) invokes a thematic relation that
the mere participant experiences a change of his/her marriage state. (強)迫 ... 离婚 (qiang)po ... lihun
‘to force ... to divorce’ in (30b) is the causativized construction of 离婚 lihun with the co-verb (強)迫
(qiang)po ‘to force’. While the causer, the external source of energy, is realized as subject in the sentence-
initial position before (強)迫 (qiang)po, the causative pivot 发妻 faqi ‘wife’ is located between co-verb
(強)迫 (qiang)po and the intransitive predicate 离婚 lihun. 发妻 faqi plays exactly the role of what a
pivot does as mentioned in 4.2.2 since she receives energy from the causer and initiates the thematic
relation (离婚 lihun). 被迫离婚 bei po lihun ‘be forced to divorce’ in (30c) is the passive of the causative
construction (強)迫 ... 离婚 (qiang)po ... lihun, in which the causer, head of the profiled action chain, is
defocused and the pivot is selected as the trajector.

(30) a. 钢铁大亨发妻离婚了。 [intransitive]

‘The wife of the steel entrepreneur is divorced.’
b. 钢铁大亨強迫发妻离婚。 [causative]
‘The steel entrepreneur forced his wife to get divorced.’
 Papers in Linguistics Science, No. 26 (2020) 101

c. 钢铁大亨发妻被迫离婚。 [passivized causative]

‘The wife of the steel entrepreneur is forced to get divorced.’
d. 钢铁大亨发妻 “被离婚”。(Lu 2013:24) [bei-XX (“being forced” type)]
‘The wife of the steel entrepreneur is “bei-divorce” (‘forced to get divorced’).

Figure 11. Schema of Passivized Causative 被迫 ... bei po … ‘be forced to ...’15

Note that the causer is still within the scope of predication but is just defocused to be less cognitively
prominent. As a result, the causer can still be realized as oblique16. In addition, as mentioned in 4.2.2, the
pivot in the passivized causative (发妻 faqi ‘wife’ in (30c)) demonstrates some level of “subjecthood”:
although she is being forced, the wife of the steel entrepreneur still initiates the thematic event 离婚 lihun
‘to get divorced’ based on her own volition (since there are situations like 雖然被迫離婚，但她拒絕了
‘Though being forced to get divorced, she refused.’). The block surrounding the CHANGE-LOC(STATE)
segment represents that it is depicting an autonomous event, for the whole event is realized by periphrastic
causative using co-verb 迫 po ‘to force’ instead of lexical causative.

Figure 12. Schema of “being forced” type of bei-XX

Now let us think about the “being forced” type of the new bei-XX construction. The event depicted

15
It seems that the passivized causative construction 被迫 XX bei po XX ‘be forced to XX’ requires a
S-thematic relation as the core event, for an E-thematic relation seems difficult to become the predicate
XX, e.g. *她被迫死了 ta bei po si le ‘She is forced to die’. Further study is needed for the core event of
a complex event like passivized causatives in Mandarin.
16
For example, 发妻被钢铁大亨強迫离婚 ‘The wife of the steel entrepreneur is forced to get divorced
by him.’
102

in (30d) is portraying that the thematic event 离婚 lihun ‘to get divorced’ happens without the subject’s
willingness and acknowledgement (See the comparison of (18) and (18’)). Furthermore, the cause of the
event is unclear: it is not sure who is the causer of the thematic event. Based on the semantic and syntactic
features described in Chapter 3, the schematization of “being forced” type of bei-XX is proposed as in
Figure 12.
Two differences can be observed compared to passivized causatives (Figure 11). First, the block
surrounding the CHANGE-LOC(STATE) segment remains in bei-XX, since a thematic event which is
conceptually autonomous is still invoked. However, the pivot seems to lose some of its “subjecthood”, as
it cannot control over the thematic event and the change of state is not based on its own volition, which is
illustrated as the double arrow in dashed line. Second, the causer is not profiled in Figure 11 because it is
defocused due to passivization, yet the causer is not even included within the scope of predication of the
new bei-XX, which is diagrammed as the dotted circle in Figure 12. From the fact that the causer cannot
be recovered and remains implicit, the causer is thought to be out of the scope of predication of bei-XX.

4.3.2. “Being (falsely) reported/ said” type

As for the “being (falsely) reported/ said” type, let us compare (30d) and (31).

(31) 贝克汉姆「被离婚」英媒体见怪不怪将消息屏蔽。
‘Beckham is “bei-divorce”. British media block the (fake) news without shock or surprise.’
(= (23))

The difference between (31) and (30d) is whether the thematic event, 离婚 lihun ‘to get divorced’ in this
case, does happen in reality or not. In reality, Beckham is not divorced but is rumored to be divorced, and
by being falsely rumored Beckham receives adverse influence from it. It can be said that the reading of
“being falsely reported/ said” of bei-XX establishes in the context that what is described does not
correspond to reality. If Beckham were divorced in reality, then the interpretation of (31) would be of
“being forced” type.

Figure 13. Schema of “being (falsely) reported/ said” type of bei-XX

 Papers in Linguistics Science, No. 26 (2020) 103

The schema of “being (falsely) reported/ said” type is proposed as diagrammed in Figure 13, which differs
from Figure 12 with regard to the occurrence of change and its resultative state being illustrated with
dashed lines. The block representing the autonomy of the thematic event is also illustrated in dashed lines.
In reality, a change of the profiled participant might have happened, yet is not portrayed as XX denotes.
For instance, in the case of 李国福 “被自杀” ‘Li Guofu is “bei-suicide” (falsely said to commit suicide)’
in (25), Li Guofu did die in the prison hospital, but the user of expression “被自杀” believes he did not
die of suicide but of other causes. So the profiled participant does experience some change owing to
external energy source, yet it is not what is described by the predicate XX.

4.3.3. Schematization of the new bei-XX construction

Concluding what has been discussed in 4.3.1 and 4.3.2, the schematization of the new bei-XX construction
is diagrammed as Figure 14 below.

Figure 14. Schema of bei-XX

Figure 14 abstracts the similarity of Figure 12 and Figure 13 in: (i) the thematic relation (CHANGE-
LOC(STATE) segment) being conceptually autonomous, (ii) the external source of energy being not
selected in the scope of predication, and (iii) the participant which experiences change being profiled as
the trajector. Each of them corresponds to the syntactic features of bei-XX summarized in 3.3.
It can be observed that the predicate XX may be an intransitive verb, VN construction, adjectival
verbs or even a nominal phrase, and what they have in common is that they all require only one argument,
as a thematic relation only include one participant. The event XX mostly represents volitional action and
spontaneous situations as mentioned in 3.2, which corresponds to the thematic event being autonomous in
Figure 14. Yet the mere participant which initiates the change lacks a certain degree of “subjecthood”, for
it cannot control the change and has no volition and willingness to initiate it. In fact, it is presumed to have
an energy source outside of the thematic relation that initiates the change. The source of energy, or the
causer of the event, however, is not included in the scope of predication, and therefore the causer remains
implicit and cannot be recovered. The external energy source must be strong enough to initiate the thematic
relation with it head lack of “subjecthood”, so the implicit causer is often authority, officials or social
media. Last but not the least, the participant who experiences change is selected to be the trajector since it
104

is the mere participant within the profile action chain. It is realized as subject of the bei-XX construction.
What distinguishes “being forced” type and “being (falsely) reported/ said” type is whether the event
represented by the thematic relation is consistent with what really happens or not. “Being (falsely)
reported/ said” type depicts an event which does not happen or happens in a different manner.

5. Conclusion
So far this paper has considered a novel usage in Mandarin passive, bei-XX construction. To define bei-
XX and to distinguish it from other uses, functions of passives and properties of bei passives have first
been examined. To perform the function of demoting the agent, canonical bei passives have the structure
with a patient as subject, followed by passive marker 被 bei, the optional agent and the transitive predicate.
Two qualities of bei passives can be observed: Adversity and disposal of the patient function as necessary
constructional meaning. While transitive verbs are expected to be the predicate of bei passives, the new
bei-XX construction seems to allow intransitives and others of atypical grammatical categories to be the
predicates. Based on previous studies, this thesis has categorized bei-XXs into two types according to their
constructional meaning. “Being forced” type has the meaning of being forced to do an action or reach a
certain state against their willingness, while “being (falsely) reported/ said” type depicts a situation which
is incompatible with the reality. The two types share three syntactic features distinct from canonical
passives: (i) No NP can be recovered following bei as the demoted agent. (ii) bei-XX requires predicates
other than transitives. (iii) The subject of bei-XX does not play the role of receiving action but acts like
the doer of an action.
Even though Xiao (2016) may come to a conclusion that the passive marker 被 bei in bei-XX has
grammaticalized to a pseudo-prefix and makes the whole construction of intransitive, it has yet to clarify
how we construe the event represented by bei-XX, and how it is reflected on bei-XX’s structural features.
Hence, conceptions in cognitive linguistic framework are introduced in this study. By employing the action
chain model, it has been found that the event construal of the two types share similarities in that: (i) the
most salient participant (which is realized as subject) loses some “subjecthood” like lack of volition and
controlability over the thematic relation the predicate denotes, and at the same time receiving external
energy to drive the whole event, (ii) the source of external energy may be out of the scope of predication
for it cannot be recovered in the corresponding bei-XX construction. The difference of “being forced” type
and “being (falsely) reported/ said” type resides in whether the event they denote corresponds to reality or
not.
Although it is presumed that the thematic relation XX denotes in bei-XX is conceptually autonomous
and therefore it is expected that XX will be predicates denoting S-thematic relations, only a small number
of predicates have been examined in this study. Corpora-driven approaches are expected in the future
 Papers in Linguistics Science, No. 26 (2020) 105

research, to see if the distribution of the predicates justifies this study. Furthermore, the distinction of
“being forced” type and “being (falsely) reported/ said” type is attributed to a pragmatic factor: reference
to the real world or to a certain context. Therefore, the interpretation of the two types of interpretation can
also be examined under frameworks concerning context.
106

References
Berk, L. M. 1999. English Syntax: From Word to Discourse. Oxford: Oxford University Press.
Chao, Y. R. 1968. A Grammar of Spoken Chinese. Berkeley and Los Angeles: University of California
Press.
Chen, Xinren. 2017. Extensions of the Chinese Passive Construction: A Memetic Account. East Asian
Pragmatics 2.1: 59-74.
Givón, T. 1984. Syntax: A Functional-Typological Introduction, Vol. I. Amsterdam/Philadelphia: John
Benjamins.
Hashimoto, Mantaro (桥本万太郎). 1987.「汉语被动式的历史・区域发展」. 『中国语文』 第 1 期:
36-49.
Hong, Shin-heng (洪心衡). 1956. 『汉语语法问题研究』. 新知识出版社.
Hopper, P. J., & Traugott, E. C. 2003. Grammaticalization (2nd ed.). Cambridge. MA: Cambridge
University Press.
Huang, Chu-ren, and Shi Dingxu (ed.). 2016. A Reference Grammar of Chinese. Cambridge: Cambridge
University Press.
Langacker, Ronald W. 1991. Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin and
New York: Mouton de Gruyter.
Langacker, Ronald W. 2008. Cognitive Grammar: A Basic Introduction. Oxford University Press.
Li, Chao. 2007. Evolution of the bei constructions in Chinese. Journal of Chinese Linguistics 35(1): 98-
127
Li, Charles N., and Sandra A. Thompson. 1981. Mandarin Chinese: A Functional Reference Grammar.
Berkeley: University of California Press.
Li, Yuming, and Li, Wei. 2015. The Language Situation in China. Walter de Gruyter.
Liu, Y. H, Pan, W. Y., & Gu, W. (劉月華, 潘文娛, 故韡). 1996.『實用現代漢語語法』. 師大書苑.
Lu, Hao-yu (路浩宇). 2013.「中国語の自動詞述語受身表現について：インターネットで用いられ
る“被就业”のタイプを例として」名古屋大学教養教育院紀要論文.
Lü, Shu-xiang (呂叔湘) (主編). 1980.『现代汉语八百词』. 商务印书馆.
McEnery, T., Xiao, R., & Qian, Y. 2006. Passive constructions in English and Chinese: A corpus-based
contrastive study. Language in Contrast 6(1): 109-149
O’Grady, W. 2001. The Syntax Files. Honolulu: University of Hawaii at Manoa.
Ramchand, Gillian. 2013. Argument structure and argument structure alternation. In Marcel den Dikken
(ed.), The Cambridge Handbook of Generative Syntax, 265-321. Cambridge University Press.
Shi, Dingxu. 1997. Issues on Chinese passive. Journal of Chinese Linguistics 2591: 41-70.
Shibatani, Masayoshi. 1985. Passives and Related Constructions: A Prototype Analysis. Language 61: 821-
 Papers in Linguistics Science, No. 26 (2020) 107

848.
Takami, Ken-ichi (高見健一). 1995.『機能的構文論による日英語比較ー受身文・後置文の分析』東
京: くろしお出版.
Taniguchi, Kazumi (谷口一美). 2005.『事態概念の記号化に関する認知言語学的研究』ひつじ書房.
Wang, Li (王力). 1980. 『汉语史稿』(中). 中华书局.
Xiao, Yun. 2016. The Rise of Catchword “被 bei-XX”: Grammaticalization and reanalysis. Chinese as a
Second Language 51(1): 79-97.
Yang, Kai-rong (楊凱栄). 1989.「文法の対照研究一中国語と日本語一」, 山口佳紀(編)『講座日本
語と日本語教育』第 5 巻：日本語の文法・文体(下). 明治書院.
Zhu, De-xi (朱德熙). 1982. 『语法讲义』. 商务印书馆.
108

「被」を用いた非典型的受動構文－認知言語学による分析－

樊 毓（ハンイク）

本研究は中国語の新規の非典型的受動構文、「被 bei-XX 構文」の機能の分析を目的とする。当

該構文は「動作主の脱焦点化（“agent-defocusing”）
」という言語一般的な機能を持つ受動文から
逸脱しているように見える。新規の被 bei-XX 構文には、構文的意味によって「強いられる」タ
イプ（“being forced” type）と「言われる」タイプ（“being (falsely) reported/ said” type）に分けら
れるが、
「非典型的述語」
、「降格された動作主が復元不可能」
、「主語が被動作主でない」という
統語的特徴が共通して見られる。本研究では、被 bei-XX 構文によって表される事態の捉え方を
解明するために、行為連鎖モデルなどの認知言語学的概念を用い、 (i) 被 bei-XX 構文が表す事
態の中で、最も際立つ参与者の「主語性（“subjecthood”）
」の欠如、(ii) 述定スコープの中のエネ
ルギー源の不在、を当構文の特徴としてまとめる。また、
「強いられる」タイプと「言われる」
タイプ両タイプの構文的意味の違いが現実世界への参照に関わることも指摘する。最後に、被
bei-XX 構文のスキーマを提案する。