Jump to ContentJump to Main Navigation
Show Summary Details
More options …


An Interdisciplinary Journal of the Language Sciences

Editor-in-Chief: Gast, Volker

IMPACT FACTOR 2017: 0.644
5-year IMPACT FACTOR: 0.878

CiteScore 2017: 0.79

SCImago Journal Rank (SJR) 2017: 0.418
Source Normalized Impact per Paper (SNIP) 2017: 1.386

See all formats and pricing
More options …
Volume 56, Issue 1


Pathways to adversity and speaker affectedness: On the emergence of unaccusative ‘give’ constructions in Chinese

Weirong Chen
  • Corresponding author
  • Institute of Linguistics, Chinese Academy of Social Sciences, No. 5, Jianguomennei Street, Beijing, China
  • University of International Business and Economics, Beijing, China
  • Email
  • Other articles by this author:
  • De Gruyter OnlineGoogle Scholar
/ Foong Ha Yap
  • Department of English, Hong Kong Polytechnic University, Hong Kong, China
  • School of Humanities and Social Science, Chinese University of Hong Kong, Shenzhen, China
  • Email
  • Other articles by this author:
  • De Gruyter OnlineGoogle Scholar
Published Online: 2018-01-05 | DOI: https://doi.org/10.1515/ling-2017-0038


In this paper, we examine the characteristics of unaccusative ‘give’ constructions in Chinese, and additionally identify the pathways for their emergence in some Chinese dialects, in particular Southern Min and Mandarin varieties. 1 We distinguish between Type 1 and Type 2 unaccusative ‘give’ constructions, the former involving reversible ‘escape’-type intransitive predicates, and the latter irreversible ‘die’-type intransitive predicates. Type 1 constructions are attested in many Chinese varieties, such as Mandarin, Min, Wu, Hui, Hakka and Cantonese, whereas Type 2 constructions are more rare and thus far are mainly attested in Southern Min and some Mandarin varieties. Two major pathways in the development of unaccusative ‘give’ constructions are identified in this paper, namely, the causative pathway and the passive-mediated pathway. Our analysis also traces how the unaccusative ‘give’ construction develops into a marker of adversity and speaker affectedness. The findings of this study have implications for understanding the relationship between changes in valence (i.e., the number of core arguments that are profiled in a given construction) and speaker’s subjective stance.

Keywords: ‘give’ constructions; unintentional causative; passive; unaccusative; speaker affectedness marker

1 Introduction

Versatile constructions are known to be a frequent conduit for the emergence of a wide range of new grammatical and pragmatic functions. Among the most extensively studied constructions are those involving motion and transfer verbs such as ‘give’, ‘take’, and ‘get’. In this paper, we will focus on ‘give’ constructions in Chinese, such as the Mandarin gĕi construction in (1a) below, with special attention to their extended uses as causative, passive and unaccusative constructions, as illustrated in the Mandarin examples in (1b)–(1d) respectively. 2


Lexical transfer verb ‘give’

2SGGIVE3SG 3CLchance

‘Give him a chance.’


Permissive causative ‘give’ construction


‘Let him go home.’


Passive ‘give’ construction


‘The fish got eaten by the cat.’


Unaccusative ‘give’ construction


‘The thief, alas, got away.’

Previous studies have identified a robust development from causative to passive ‘give’ constructions among Chinese dialects (see Yue-Hashimoto 1976; Hashimoto 1986; Hashimoto 1988; Cheng et al. 1999; Jiang 2002; Zhang 2000; Chen 2009; Chin 2011), arguably mediated by reflexive ‘give’ constructions (see Yap and Iwasaki 2003; Yap and Iwasaki 2007). This development from a causative to passive interpretation is related to the extended uses of ‘give’ constructions, from 3-place (3P) to 2-place (2P) predicate constructions. That is, we see a semantic extension across verb types with decreasing valence. This is illustrated in the extension of Mandarin gĕi from a ditransitive transfer verb meaning ‘give’, as in ‘(you) give him a chance’ in (1a), to a permissive causative ‘give’ verb, as in ‘let (<‘give’) him (to) go home’ in (1b), where the patient/theme argument is replaced by a complement clause ‘(to) go home’. This subsequently leads to a 2-place predicate passive construction, as in ‘the fish got eaten by the cat’ (lit. ‘the fish (inadvertently) let the cat eat (it)’ in (1c).

Note that reflexive contexts are scenarios where the argument in topic or subject position is co-referential with an elided patient argument in the complement clause, as illustrated by ‘fish’ in (1c) above. These reflexive contexts can facilitate the reanalysis of ‘give’ constructions from 3-place-predicate causatives to 2-place-predicate passives. The former is comparable to the English (inadvertent) let-causative and the latter to the English get-passive.

While the causative-to-passive development in Chinese has received increasing attention and is now fairly well understood, the grammaticalization pathway(s) that give rise to unaccusative ‘give’ constructions, which involve 1-place (i.e., intransitive) predicates such as ‘the thief, alas, got away’ as in (1d), have yet to be fully described. An intriguing question for the present study is whether unaccusative ‘give’ constructions in Chinese are derived directly from causative ‘give’ constructions, or are mediated by passive ‘give’ constructions.

The unaccusative ‘give’ construction is found not only in Mandarin, as shown in (1d) above, but is also attested in Southern Min varieties such as Jieyang Chaozhou, Hui’an and Taiwan Southern Min (see Matthews et al. 2005; Matthews and Yip 2008; Chen 2011; Lin 2011). 4 Structurally, the ‘give’ morpheme occurs in an intransitive construction, with a [Patient/Theme NP+‘give’+3SG+unaccusative predicate] configuration. In Mandarin varieties, the third person singular pronoun (3SG) that follows the ‘give’ morpheme is usually dispreferred and often omitted, as shown by the optional presence of third person singular pronoun in (1d) above. In Southern Min varieties, on the other hand, the third person singular pronoun (3SG) is typically obligatory, as seen in the Jieyang Chaozhou and Hui’an examples in (2) and (3) below. In these Southern Min varieties, the predicate (si k’ɯ in Jieyang Chaozhou and si khɯ in Hui’an, both meaning ‘die’) is unaccusative, and the third person pronoun i that follows the ‘give’ verb is obligatory. This third person singular pronoun is argued to be highly grammaticalized and pleonastic, i.e., it is not constrained by person and number agreement and can co-occur with non-third person and plural referents in topic position. 5 As will be shown in this paper (see Section 2.2.2), the pleonasticity of this third person singular pronoun i provides valuable clues to the evolving semantic and morphosyntactic status of the ‘give (it)’ construction in Chinese (see also similar observations in Matthews et al. 2005; Lin 2011).


Jieyang Chaozhou dialect


‘Unfortunately, the flower has died.’


Hui’an dialect


‘Unfortunately, the flower has died.’

In this paper, we will trace the grammaticalization pathways that give rise to unaccusative ‘give’ constructions in Chinese, with a special focus on Southern Min and Mandarin varieties. The specific questions we seek to address are as follows:

  1. What is the function of the unaccusative ‘give’ construction? More specifically, what is its pragmatic function? Does it yield adversative and speaker-affectedness readings, and if so, how does it do so?

  2. How did the unaccusative ‘give’ construction emerge? Is it derived from causative constructions or passive ones? Is there evidence of intermediate ‘bridging’ constructions along the pathway to unaccusative ‘give’? If so, what are/were these intermediate constructions?

Data for our analysis of the unaccusative ‘give’ constructions in Chinese are mainly based on a questionnaire focusing on native speaker judgment on the grammaticality of various types of unaccusative ‘give’ constructions in different Chinese dialects (e.g., Southern Min, Hui, Hakka, Cantonese, Wu, and Mandarin; see Tables 1 and 2 in Section We adopt a grammaticalization analysis that identifies the various functions of ‘give’ constructions in these Chinese varieties, focusing in particular on the semantic extensions of causative, passive, and unaccusative uses. We also adopt a crosslinguistic analysis to determine the possible direction of extensions, more specifically to determine the relationship between the causative, passive and unaccusative ‘give’ constructions.

Table 1:

Types of unaccusative ‘give’ predicates in different Mandarin varieties.

Table 2:

Types of unaccusative ‘give’ predicates in various Chinese dialects.

The rest of this paper is organized as follows. In Section 2, we first identify the various functions of ‘give’ constructions in Chinese and then highlight the defining characteristics of the unaccusative ‘give’ construction. In Section 3 we propose two pathways for the emergence of unaccusative ‘give’ constructions in Chinese, namely, the causative pathway and the passive-mediated pathway. We also examine the relationship between adversative events and the extension of ‘give’ constructions to unaccusative predicates, and the concomitant reinterpretation of the unaccusative ‘give’ construction as a marker of speaker affectedness. Section 4 concludes.

2 Versatile uses of ‘give’ constructions in Chinese

2.1 Commonly attested functions of ‘give’ constructions in Chinese dialects

Crosslinguistically, ‘give’ constructions serve a wide range of functions including lexical, dative, benefactive, causative and passive (see Lord 1993; Nedjalkov 1993; Knott 1995; Yap and Iwasaki 1998; Yap and Iwasaki 2003; Yap and Iwasaki 2007; Zhang 2000; Lord et al. 2002; Matthews et al. 2005; Chen 2009; Chen 2011; Chin 2011; Lin 2011). Examples of ‘give’ functions from Mandarin are given in (4). However, it should be noted that the ‘give’ constructions in most Chinese dialects do not have the full range of functions illustrated in (4). For example, the benefactive ‘give’ function in (4c) is not attested in the Hui’an dialect.


Lexical ‘give’


‘Give him a book.’


Dative ‘give’


‘Leave the key to him.’


Benefactive ‘give’


‘I translated (it) for him.’


Causative ‘give’


‘Let me take a look at (it).’


Passive ‘give’


‘The door was blown open by the wind.’

Matthews et al. (2005) further identified a typologically rare ‘give’ construction in the Jieyang Chaozhou variety of Southern Min, which they classify as an unaccusative construction. As seen in (5), ta is a stative verb meaning ‘be dry’, and in combination with the resultative verb complement k’ɯ (derived from the verb ‘go’), the verbal predicate ta k’ɯ ‘has gone dry’ yields an unaccusative reading (involving a change of state and an affected patient), while k’eʔ no longer functions as a ‘give’ verb despite residually retaining the third person pronoun i in object-like fashion.


‘The ink of the pen has gone dry.’

Unaccusative ‘give’ constructions have also been identified in the Hui’an dialect, another variety of Southern Min (Chen 2011), and in Taiwan Southern Min (Lin 2011). In the next section we will examine the nature of unaccusativity and its particular realization via the ‘give’ construction in Chinese, more specifically in the Mandarin and Southern Min varieties.

2.2 Unaccusative ‘give’ constructions in Chinese

2.2.1 Defining unaccusativity

According to Perlmutter’s (1978) Unaccusative Hypothesis (see also Burzio 1986), intransitive verbs fall into two categories: unergative and unaccusative verbs. These two categories differ in that the former has an external argument but no internal argument, while the latter has an internal argument but no external argument. In other words, the subject of an unergative verb has the role of agent (implying volitional control over the action), while the sole argument of an unaccusative verb has the role of theme or patient (lacking volitional control) (see also Matthews et al. 2005: 275). For example, the verb tsau3 ‘run’ in the Hui’an dialect is an unergative verb, since it involves an external argument, i.e., an agent, who has volitional control over his/her action of running, whereas the verb si3 ‘die’ is an unaccusative verb, since it involves an internal argument, i.e., a patient/theme argument (such as hue1 ‘flower’ in example (3) noted earlier) that lacks volitional control over its dying. In Chinese, however, perfective aspect markers (e.g., Mandarin le) and verbal complements (e.g., Hui’an khɯ0 and Jieyang k’ɯ) can be added to an unergative verb to yield an unaccusative predicate (see Section 2.2.2 for more discussion).

2.2.2 Distribution and characteristics of Chinese unaccusative ‘give’ constructions Unaccusative ‘give’ constructions in non-Mandarin Chinese varieties

In addition to the three varieties of Southern Min mentioned earlier (i.e., Jieyang Chaozhou, Hui’an and Taiwan Southern Min varieties), the unaccusative ‘give’ construction is also attested in other Southern Min varieties (e.g., Quanzhou and Xiamen in Fujian; Shantou and Chenghai in Guangdong), as well as Eastern Min varieties (e.g., Fuzhou), and also in other non-Mandarin Chinese dialects such as Wu (e.g., Suzhou, Shanghai, Gaochun, Fuyang, and Wenzhou), Hui (e.g., Shangzhuang Jixi), Hakka (e.g., Meixian and Fengshun), and Cantonese (e.g., Hong Kong Cantonese) (Lin 1996; Li and Chang 1997; Li and Chang 2000). 7 In these varieties, the unaccusative ‘give’ construction has traditionally been regarded either as a passive use 8 (e.g., Li 1997: 123–125; Shi 1997: 22–23), a causative use (Pan 1997:63–64), or a special construction (e.g., Liu 1997:6). 9

Previous studies such as Matthews et al. (2005) and Lin (2011) have shown that unaccusative ‘give’ constructions in the Chaozhou variety of Southern Min and in Taiwan Southern Min share the following three features:

  1. The unaccusative ‘give’ construction denotes adversity.

  2. The unaccusative ‘give’ construction requires a non-referential/expletive third person pronoun i following the ‘give’ morpheme.

  3. The predicate of an unaccusative ‘give’ construction denotes a change of state, and a resultative verbal complement is often used.

These features are also shared by unaccusative ‘give’ constructions in many other Southern Min varieties. Let us take the Hui’an dialect as an example, as shown in (6) below. Both the unaccusative ‘give’ constructions in (6a) and (6b) denote adversity and unexpectedness, i.e., the fact that ‘the thief got away’ and ‘the flower has died’ are perceived as unfortunate and unexpected events by the speaker. Crucially for the present study, the speaker is emotionally affected (e.g., upset) by what has happened.


‘Unfortunately, the thief got away.’


‘Unfortunately, the flower has died.’

Given the adversity and unexpectedness reading of the unaccusative ‘give’ constructions in (6a) and (6b) above, an adverb such as saʔ7 ‘unexpectedly’ can precede the predicate, and an interjection such as ua khɔŋ ‘alas’ expressing dissatisfaction and regret can be placed before the topic NP, e.g., before tshat8-4-a3 ‘thief’ and hue1 ‘flower’ in (7a) and (7b) respectively.

□□ 10,賊仔
ua khɔŋ,tshat8-4-a3saʔ7-8khɔ5-4i1tsau3khɯ0

‘Alas, the thief got away.’

ua khɔŋ,hue1saʔ7-8khɔ5-4i1si3khɯ0

‘Alas, the flower has died.’

The intransitive predicates tsau3 khɯ0 ‘run away, escape’ and si3 khɯ0 ‘die’ – both of which follow the khɔ5-4 i1 (lit. ‘give’+3SG) construction – are unaccusative and denote non-volitional changes of state. In (6a) and (7a), the verb tsau3 ‘run’, originally an unergative verb, expresses unaccusative use with the resultative verbal complement khɯ0. 11 In (6b) and (7b), the verb si3 ‘die’ (with or without the verbal complement khɯ0) likewise yields an unaccusative predicate. An important difference between unaccusative predicates tsau3 khɯ0 and si3 khɯ0 is that the former is reversible, whereas the latter is irreversible: the thief may be caught again after running away, whereas the flower normally cannot return to life after death. Thus, in this paper, we distinguish the former type of unaccusative predicate as Type 1 unaccusative (or ‘escape’-type, reversible type) and the latter as Type 2 unaccusative (or ‘die’-type, irreversible type). By way of illustration, then, (6a) and (7a) are Type 1 (or ‘escape’-type) unaccusative ‘give’ constructions, while (6b) and (7b) are Type 2 (or ‘die’-type) unaccusative ‘give’ constructions, the latter consistent with what is often referred to in the literature as “pure unaccusatives” (Lin 2011).

There is yet another difference between the (a) and (b) examples above: the topic tshat8-4-a3 ‘the thief’ in (6a) and (7a) is a beneficiary in the sense that he is not apprehended but instead is free, whereas the topic hue1 ‘flower’ in (6b) and (7b) is a malefactee/affected patient in the sense that it is now lifeless. In other words, Type 1 (‘escape’-type) unaccusatives tend to involve beneficiary topic NPs, while Type 2 (‘die’-type) unaccusatives tend to involve malefactee/affected patient topic NPs.

This beneficiary vs. malefactee distinction for the topic NP is not always easily determined on the basis of an unergative vs. unaccusative verb distinction. This is because unergative verbs can combine with verbal complements to yield predicates with unaccusative interpretation. In (8) below, for example, the verb 1 ‘fly’ is originally an unergative verb, but it expresses unaccusative meaning with the verbal complement khɯ0. In other words, an affected patient usage for tsua3 ‘paper’ is possible because 1 khɯ0 ‘fly away’ is an unaccusative predicate, often with an unexpected as well as adversative reading.


‘Unfortunately, the piece of paper flew away.’

In these Type 1 and Type 2 unaccusative constructions, there is at least one covert affectee that is not co-referential with the topic NP. This affected entity can be allophoric referents (i.e., ‘someone other than the speaker’, hence second or third person referents in the discourse, e.g., the possessor of the paper, when the speaker is talking about someone else’s experience). Crucially however, this covert affected entity necessarily includes the speaker himself/herself (regardless of whether the speaker is talking about his/her own experience, e.g., where the speaker himself/herself is the possessor of the paper). In either case, the speaker is emotionally affected by the event and regards the event as unfortunate and unexpected. This means that the expression khɔ5-4 i1 (along with the entire unaccusative construction) is used to express the speaker’s subjective stance.

Given that unaccusative constructions are generally associated with adversative outcomes (at least for the speaker), the speaker often either experiences and expresses some negative feelings (e.g., frustration, disappointment, annoyance), or evaluates the situation negatively (e.g., remarking that the outcome is unfortunate, sometimes accompanied by emotional interjections or adverbials such as ua khɔŋ ‘alas’ as seen in (7a) and (7b) earlier). In the (a) example, the speaker may be upset upon hearing that the thief has got away; in the (b) example, the speaker may be saddened upon finding that a favorite plant (‘the flower’) has withered and died. In (8), the speaker may be frustrated or alarmed that an important document or essential stationery is about to disappear. In this regard, unaccusative ‘give’ constructions have come to explicitly encode, not just the adversative situation, but also the speaker’s subjective evaluation of the adversative situation. In other words, the khɔ5-4 i1 construction in Hui’an has developed into a subjectivity marker, more specifically, a marker of speaker affectedness. Recall that Lin (2011) has used the term ‘adversative marker’; in this paper, we will further highlight that the expression and evaluation of adversity in khɔ5-4 i1 constructions in Hui’an is viewed from the perspective of the speaker. In this respect, the khɔ5-4 i1 construction in Hui’an (as well as other unaccusative ‘give’ constructions in other Chinese varieties) constitute a subjectification mechanism that expresses the speaker’s subjective stance (for discussions of subjectification and subjectivity, see Traugott 1995; Traugott 2003; Traugott 2010).

Among the Type 2 constructions, the unaccusative predicate can also be formed using a stative verb plus the verbal complement khɯ0, as in (9). 12 This Type 2 unaccusative ‘give’ construction derived from a stative verb focuses on an adverse and undesirable resultant state (or outcome). In the case of the “pure” unaccusative constructions in (6b) and (7b), the adverse outcome is a dead flower; in the ‘stative verb’ unaccusative construction in (9), the adverse outcome is “blackened” vegetables (i.e., vegetables that have gone bad). Both situations are irreversible and undesirable to the speaker.


‘Unfortunately, the vegetables have turned black.’

Note that in both the Type 1 and Type 2 unaccusative ‘give’ constructions in the Hui’an examples above, the third person singular pronoun i1 following the ‘give’ morpheme khɔ5 is obligatory. 13 Its obligatory presence is highlighted by the need to express tone sandhi change in contexts such as (10) and (11) below, where the ‘give’ morpheme is realized as khɔ5-4 when the pronoun i1 is present, as in (10a) and (11a), but is realized instead as khɔ5-1 when the pronoun is absent, as in (10b) and (11b). In other words, when followed by pronoun i1, the ‘give’ morpheme khɔ5 undergoes regular tone sandhi where it yields khɔ5-4, but in the absence of this pronoun, the ‘give’ morpheme undergoes irregular tone sandhi and yields instead khɔ5-1. The regular tone sandhi comprises the citation tone and the sandhi tone, i.e., tone 5 and tone 4 respectively in khɔ5-4, following the general tone sandhi rules for the Hui’an dialect (refer to Chen 2011). On the other hand, the irregular tone sandhi seen in khɔ5-1 in (10b) and (11b) reflects the incorporation of tone 1 from the elided pronoun i1. Note that the pronoun cannot be elided without this tone sandhi change, as indicated by the ungrammaticality of (10c) and (11c). Also worth noting is that the unaccusative ‘give’ construction with the tonally incorporated pronoun, which yields the irregular tone sandhi khɔ5-1 as seen in (10b) and (11b), is much more common than the construction with the overt pronoun i1 which retains the regular tone sandhi, as in (10a) and (11a). This asymmetry in usage frequency suggests that the third person pronoun i1 is increasingly used in an unstressed form, and khɔ5-1 (<khɔ5-4 i1) is the more grammaticalized form.


賊仔 與 伊 走 去(=6a)


‘Unfortunately, the thief got away.’


‘Unfortunately, the thief got away.’

*tshat8-4-a3khɔ 5-4tsau3khɯ0

‘Unfortunately, the thief got away.’


花與伊死 去 (=6b)


‘Unfortunately, the flower has died.’


‘Unfortunately, the flower has died.’


‘Unfortunately, the flower has died.’

Matthews et al. (2005) argue that the third person singular pronoun i in the Jieyang Chaozhou k’eʔ i (‘give’+3SG) unaccusative construction is expletive and non-referential, since “it remains invariable regardless of the person/number features of the subject” (p. 282). Lin (2011) provides the same observation for the pronoun i1 in the hoo7 i1 (‘give’+3SG) unaccusative construction in Taiwan Southern Min. Likewise, the third person singular pronoun i1 in the khɔ5-4 i1 unaccusative construction in Hui’an is also expletive and non-referential. More specifically, similar to what has been observed in the Jieyang Chaozhou and Taiwan Southern Min dialects, the unaccusative ‘give’ construction in the Hui’an dialect allows a plural subject, as in (12), as well as a second person subject, as in (13). This shows that the pronoun i1 in the khɔ5-4 i1 unaccusative construction in Hui’an is no longer used to signal person and number agreement, but is instead a highly grammaticalized, expletive, and non-referential form. This contributes to the development of khɔ5-4 i1 or khɔ5-1 (with i1 incorporated) as a fixed expression denoting adversity and speaker affectedness. 14


‘Unfortunately, those people died.’


‘Are you stupid?’

Note that we cannot rule out the possibility that the pronoun i1 in some unaccusative ‘give’ constructions may be interpreted in multiple ways. Take (6a) above as an example. On the one hand, khɔ5-4 i1 as a whole may serve as a marker of speaker affectedness, with i1 being non-referential. On the other hand, i1 in khɔ5-4 i1 can still be co-referential with the topic NP tshat8-4-a3 ‘the thief’. Bridging contexts such as this suggest a possible development from referential pronoun to non-referential pronoun. 15 In other words, on the one hand, the pronoun i1 in unaccusative ‘give’ constructions in some Southern Min varieties has become highly grammaticalized, and can be used as an expletive and non-referential form, but its use as a referential pronoun is still available, and thus its co-indexicality (or co-referentiality) with the topic is still retrievable in some contexts. Note also that the ‘co-referential’ interpretation has also been suggested in previous works on Southern Min such as Li (1997: 125), as well as previous works on other non-Mandarin Chinese dialects such as Liu (1997: 6), Shi (1997: 21) and Pan (1997: 64).

Thus far, we have been analyzing examples from Southern Min. We will now turn to other non-Mandarin Chinese dialects with unaccusative ‘give’ constructions, e.g., Eastern Min, Wu, Hui, Hakka, and Cantonese. In these dialects, only the Type 1 ‘escape’-type unaccusative ‘give’ construction similar to (6a) is attested, while the Type 2 ‘die’-type unaccusative ‘give’ construction similar to (6b) is not acceptable (Li and Chang 1997; Li and Chang 2000). 16 Consider the following example from the Suzhou Wu dialect.


Suzhou Wu dialect


‘Unfortunately, the prisoner escaped yesterday.’

According to Liu (1997: 6), in the Suzhou Wu dialect such sentences as (14) are typically used to express unexpected adversity. The third person singular pronoun li is preceded by the ‘give’ morpheme pəʔ and is co-referential with the topic NP, i.e., kəʔvᴇȵin ‘the prisoner’. Liu also pointed out that such constructions often have an overt beneficiary and a covert affectee (Danqing Liu, p.c. January 2012). In (14), for example, the beneficiary in the “escape”-type unaccusative construction is the pronoun following the ‘give’ morpheme (i.e., li), which co-refers with the topic NP kəʔvᴇȵin ‘the prisoner’. The covert affectee can be the prison official or the speaker. In the latter case, it could happen if the speaker were a prison official and thus responsible for the prison break, or if the speaker is merely an ordinary member of the community and is anxious about a convict on the loose in the neighborhood. In either context, the speaker is always emotionally affected. This could be for various reasons, for example, empathy with those responsible for security matters, embarrassment if the speaker is the prison official considered responsible for this breach of prison security, or anxiety if the speaker is alarmed at the thought of being in an unsafe neighborhood. There is often an animacy hierarchy constraint as well, which helps explain the lack of pleonasticity of the third person pronoun in the Wu dialects. In the Suzhou Wu dialect and other northern Wu dialects, the beneficiary is typically animate, and inanimate objects such as ‘paper’ or ‘flower’ cannot occur in unaccusative constructions (Danqing Liu, p.c.). Thus, only Type 1 (‘escape’-type) but not Type 2 (‘die’-type) unaccusative ‘give’ constructions are attested in these Wu dialects.

Similarly, only the Type 1 unaccusative ‘give’ constructions are attested in other non-Mandarin Chinese dialects such as Hui, Hakka, and Cantonese (pers. comm.). 18 These dialects vary in the extent to which they express all three features of the unaccusative ‘give’ construction in Jieyang Chaozhou and Taiwan Southern Min noted earlier in Matthews et al. (2005) and Lin (2011) – namely, (i) adversity, (ii) non-referentiality of the third person pronoun following the ‘give’ morpheme, and (iii) change of state. For example, whereas Hui dialects (e.g., Shangzhuang Jixi Hui) express all three features, Hakka and Cantonese dialects express only two of these features – i.e., adversity and change of state. Neither Hakka nor Cantonese has developed a non-referential or expletive third person pronoun. As illustrated in the Hong Kong Cantonese example in (15) below, the third person singular pronoun keoi5 that follows the ‘give’ morpheme bei2 is not expletive or non-referential, since it can only co-occur with singular topic NPs such as go3 caak6zai2 ‘the thief’ in (15a) but not with plural ones such as go3di1 caak6zai2 ‘the thieves’ in (15b). This indicates that number agreement is still important for the Type 1 unaccusative ‘give’ constructions in dialects such as Cantonese – and similarly, Hakka.


Hong Kong Cantonese


‘Unfortunately, the thief got away.’

個啲賊仔佢哋 / *佢走甩咗
go3di1caak6zai2bei2keoi 5dei6 / *keoi5zau2lat1zo2
CL.PLthief-DIMgive3PL /*3SGescape.PFV

‘Unfortunately, the thieves got away.’

Furthermore, there is a relatively strong animacy hierarchy constraint in Hakka and Cantonese dialects, as seen in the Cantonese example in (16), where the third person singular keoi5 ‘it’ cannot co-refer to an inanimate NP such as do2faa1 ‘the flower’, which explains why Type 2 unaccusative ‘give’ constructions are not attested in these dialects.


Hong Kong Cantonese


Intended meaning: ‘Unfortunately, the flower has died.’

Type 1 (‘escape’-type) unaccusative ‘give’ constructions like (15a) and (15b) in Hong Kong Cantonese may sometimes be a fortunate event for the topic NP (in this case, go3 caak6zai2 ‘the thief’ or go3di1 caak6zai2 ‘the thieves’ respectively), but these constructions nevertheless always signal that an unfortunate or unexpected event has happened that affects the speaker in some (usually negative) way. Crucially, then, even the Cantonese bei2 ‘give’ adversity marker – and likewise the Hakka pun (分) – conveys the speaker’s subjective stance, hence our use of the term ‘marker of speaker affectedness’, for the unaccusative ‘give’ construction. Unaccusative ‘give’ constructions in Mandarin Chinese varieties

So far, we have focused on the cross-dialectal distribution and semantic features of Type 1 (‘escape’-type) and Type 2 (‘die’-type) unaccusative ‘give’ constructions in non-Mandarin Chinese dialects. In this section, we will focus on Mandarin Chinese, which we separate from other Chinese dialects because (i) the status of unaccusative ‘give’ constructions in Standard Mandarin is controversial, and (ii) the unaccusative ‘give’ constructions in some Mandarin varieties show unique features that are different from those in the non-Mandarin Chinese dialects mentioned above.

Our first question related to Mandarin Chinese is whether there exist unaccusative ‘give’ constructions in Standard Mandarin. No potential example of unaccusative ‘give’ constructions can be found in several important works on Mandarin Chinese, such as Ding et al. (1979), Li and Thompson (1981), Zhu (1982) and Lü (1980). This seems to suggest that the unaccusative ‘give’ construction is at best marginal in Standard Mandarin. However, in a more recent study on the ‘gĕi VP’ construction in Standard Mandarin, Shen and Sybesma (2010) mentioned several examples of ‘give’ constructions that can be grouped among the unaccusative ‘give’ constructions discussed here. Our own investigations reveal that speakers differ as to whether unaccusative ‘give’ constructions such as (17) are acceptable or not in Standard Mandarin.


‘Unfortunately, the thief got away.’


‘Unfortunately, the flower has died.’

In general, in Standard Mandarin, Type 1 (‘escape’-type) constructions like (17a) tend to be more acceptable than Type 2 (‘die’-type) constructions like (17b). In addition, speakers tend to regard (17a) as a causative construction meaning ‘(someone) inadvertently let the thief run away’ (unintentional causative), or they tend to treat it as a derivative of the causative construction. Speakers also differ as to whether constructions like (17b) are widely used. For some speakers (e.g., from Xi’an in Shanxi and Harbin in Heilongjiang), a series of unaccusative/stative verbs such as 烂 làn ‘(be) rotten’ and 酸 suān ‘(be) sour’ can be used to form unaccusative ‘give’ constructions; whereas for some other speakers (e.g., from Baoding in Hebei and Xuzhou in Jiangsu), such constructions are understandable and acceptable, but very restricted and not commonly used.

It is possible that some examples of unaccusative ‘give’ constructions exist in Standard Mandarin. If so, it is likely that they are a recent development, and could be a result of influence from Southern Chinese dialects or non-standard Mandarin varieties. This may explain why the unaccusative ‘give’ construction was not reported in several major works on Mandarin Chinese mentioned above. Another reason why the unaccusative ‘give’ construction went unreported in previous works could be that unaccusative ‘give’ constructions typically occur in spoken language, rather than in written language.

Whatever the reason(s) for its lack of treatment in previous works, unaccusative ‘give’ constructions are in fact attested in some Mandarin varieties, such as the Beijing, Changchun, Harbin, and Xi’an dialects, with varying degrees of acceptability of intransitive verbs and stative verbs (see Table 1 below). Note that the distribution of Type 1 and Type 2 unaccusative ‘give’ constructions in Table 1 suggests an extension from unergative to unaccusative and then to stative verbs, though more cross-dialectal data is needed to further verify this generalization.

Another question related to Mandarin Chinese is whether unaccusative ‘give’ constructions in Mandarin varieties show any specific features that are not shared by other Chinese dialects. Similar to Southern Min varieties, in some Mandarin varieties such as the Xi’an and Harbin dialects, there are three types of verbs that may occur in unaccusative ‘give’ constructions, namely, unergative verbs such as 跑 păo ‘run’ and飞 fēi ‘fly’, unaccusative verbs such as 死 ‘die’, and stative verbs such as 烂 làn ‘(be) rotten’, although in other Mandarin varieties such as the Changchun dialect, only intransitive verbs (unergative and unaccusative, but not stative verbs) can occur in unaccusative ‘give’ constructions. Variations in the types of unaccusative ‘give’ constructions available in different Chinese dialects are shown in Table 2 below.

As seen in Table 2 above, in the Mandarin varieties, the unaccusative predicates are typically (though not only) formed with the help of the perfective marker le了. Somewhat different from Southern Min varieties such as the Hui’an dialect, the presence of gĕi (tā) 給(他) in Mandarin varieties such as the Harbin dialect is more inclined to imply the existence of a causer that is responsible for the event or the change of state (rather than implying the existence of a covert affectee). Shen and Sybesma (2010) likewise noted that the ‘gĕi VP’ construction in Standard Mandarin and most northern Chinese dialects contains a semantic component that they refer to as “external force”, which is equivalent to our notion here of “causer”. Consider the typical unaccusative construction and the ‘give’-type unaccusative construction in Harbin Mandarin in (18a) and (18b) below.


The typical unaccusative construction


‘The thief got away.’


The ‘give’-type unaccusative construction


‘Unfortunately, the thief got away.’

Example (18a) without gĕi (tā), simply describes an event that has already happened. Inclusion of gĕi (tā) as in (18b), however, implies that the thief got away because the police did not keep an eye on him, or the police were not able to catch him. No matter what the reason, in Harbin Mandarin, gĕi (tā) reminds the reader of the existence of a causer that is responsible for the thief getting away. The presence of gĕi (tā) also suggests that the event is contrary to the speaker’s expectation, e.g., the speaker feels that the thief should have been caught but the police had failed to apprehend him, and there is often a hint of blame assignment, in the sense that the speaker holds the police responsible for letting the thief escape. In this regard, as in the case of other Chinese dialects, unaccusative ‘give’ constructions in Harbin Mandarin are also used to express the speaker’s subjective stance.

Note that whereas there is greater preponderance toward a causative (hence agent-oriented) reading of the unaccusative ‘give’ constructions in some mainland Mandarin varieties (e.g., Harbin Mandarin), there is instead a greater preponderance toward a resultative (hence patient-oriented) reading in Southern Min varieties such as Hui’an. This stronger resultative orientation of the unaccusative ‘give’ constructions in the Southern Min varieties appears to be influenced by the more frequent use of various kinds of patient topic constructions (including the ka i disposal construction) in the Southern Min dialect group. This greater acceptability of using affected patient referents in topic and subtopic positions in Southern Min dialects applies to the typologically more restricted unaccusative ‘give’ constructions as well. We elaborate on the role of these patient-highlighting constructions in Section

In contrast to some Mandarin varieties in mainland China, in Taiwanese Mandarin the presence of gĕi in unaccusative constructions tends to imply the existence of an implicit affectee, rather than an implicit causer. According to Huang (2013: 111), this covert affectee in Taiwanese Mandarin is “the argument [i.e., the referent] that loses its argument position when subject suppression takes place but that still exists as a haunting phantom” [italics added]. Consider the gĕi constructions in Taiwanese Mandarin in (19a) and (19b) below.


‘The guard had the prisoner running away’

(Huang 2013: 109)


‘It happened that the prisoner ran away.’ (Huang 2013: 109)

In (19a), the subject kanshou ‘guard’ in Taiwanese Mandarin is generally understood to be the affectee. In (19b), although kanshou ‘guard’ is not overtly expressed, Huang (2013: 111) observes that the construction “can be understood implicitly as meaning that it happened to the guard that the prisoner ran away” (italics added). Huang further adds that when the context is underspecified, the implicit affectee in constructions such as (19b) could be either “the speaker or someone else whose identity is ‘somewhat slippery’”. In other words, according to Huang (2013), the implicit affectee in (19b) is someone who experiences the adversative event of the prisoner running away, and this “someone” could be the speaker or someone else, depending on the context. We suggest that this affected ‘someone’ is one with whom the speaker empathizes, hence the subjective, adversative reading of unaccusative ‘give’ constructions. Crucially, the unaccusative ‘give’ construction in Taiwanese Mandarin such as (19b) tends to imply the existence of an implicit affectee, rather than an implicit causer. This means that the semantics of the unaccusative ‘give’ construction in Taiwanese Mandarin is similar to that in Southern Min, rather than to that in some Mandarin varieties in mainland China. This is not surprising, given that Taiwanese Mandarin is strongly influenced by Taiwan Southern Min.

3 How did unaccusative ‘give’ constructions emerge?

Let us consider more closely the structural ambiguity that gives rise to a choice of causative, passive, or unaccusative readings depending on context. Agent defocusing can shift the focus of an utterance from a causative event to a passive one if the predicate is transitive, 19 as in (20), or to an unaccusative one if the predicate is intransitive, as in (21).


(i) ‘As for the thiefi, (the situation) enabled someonek to catch himi.’

(ii) ‘As for the thiefi, hei was caught by someonek.’


(i) ‘As for the thiefi, (the situation) enabled someonek to catch (himi).’

(ii) ‘The thiefi, (hei) was caught by someonek.’

(iii) ‘The thiefi was caught by someonek.’


‘The thiefi was caught.’


(i) ‘As for the thiefi, (someone/some situation) let himi get away.’

(ii) ‘The thiefi, alas, got away.’

It is evident from the above examples that both passive and unaccusative ‘give’ constructions are linked to causative ‘give’ constructions. The causative-to-passive development has been extensively discussed in previous studies. Less well-understood is how the unaccusative ‘give’ construction develops from the causative ‘give’ construction, and whether the emergence of the unaccusative ‘give’ construction is mediated by a passive ‘give’ construction. In what follows we will address this question of how unaccusative ‘give’ constructions emerge, which would shed light on the relationship between causativity, passivity and unaccusativity in Chinese.

3.1 Previous insights into the development of passive and unaccusative ‘give’ constructions

A causative-to-passive development involving the ‘give’ morpheme has been noted in a number of languages, not only in Chinese (e.g., Yue-Hashimoto 1976; Hashimoto 1986; Hashimoto 1988; Zhang 2000; Yap and Iwasaki 2003; Chin 2011) but also in neighboring languages such as Manchu-Tungusic (Nedjalkov 1993; Knott 1995). Some of these studies have further identified the reflexive ‘give’ construction as an intermediate stage between causative and passive. Nedjalkov (1993: 193), for example, has noted that causative suffixes in the Manchu-Tungusic languages have evolved a passive function via a reflexive link where the affected patient NP is co-referential with the subject NP. We illustrate this extension from causative to passive with examples from Cantonese in (22) below, where reflexive constructions (22b) and (22c) serve as “bridging” (or intermediate) structures facilitating the semantic extension and syntactic reanalysis of ‘give’ constructions from causative structures such as (22a) to passive ones such as (22d). This “Causative>Reflexive>Passive” grammaticalization pathway for ‘give’ constructions is highly robust in Chinese dialects (Yap and Iwasaki 2003).


Permissive causative


‘Mummy allows him to go to school by himself.’


Reflexive causative (with an intentional causer)


‘He allowed Mother to kiss him (one kiss).’


Reflexive passive (lacks an intentional causer → focuses on the affected patient)


‘He was kissed by someone.’


Passive (with focus on affected patient)


‘He was caught by a/the policeman.’


‘He was beaten by someone.’


‘He was bitten by a snake.’

In (22b), the second token of keoi5 ‘him’, which is the object of the verb sek3 ‘kiss’, is co-referential with the permissive causer NP keoi5 ‘he’ in subject position. 20 The ‘give’ verb bei2 here functions as a permissive causative verb meaning ‘allow, permit, let’. This is a case of reflexive causative (‘Hei allowed Mother to kiss himi’). In (22c), however, the patient NP keoi5 functions as a non-volitional subject (‘he’) as well as an affected patient (‘him’) of the predicate sek3 jat1 daam6 (lit. ‘kiss one peck’). This is a reflexive passive, with the postverbal patient NP elided and not expressed again overtly in object position, which has the effect of highlighting the affected patient in subject position. The availability of reflexive passive ‘give’ constructions such as (22c) facilitates the rise of adversative passive ‘give’ constructions such as (22d). This is typical of many other Chinese varieties as well, where the reflexive NP is often not copied or repeated again in the same clause. It is worth noting that the absence of the reflexive patient NP in postverbal position coincides with a shift in focus from a reflexive causative interpretation to a reflexive passive one. Crucially, without an overtly expressed patient NP in postverbal position, the prominence of the causee-agent NP is downgraded and the ‘give’ verb bei2 becomes reanalyzed as a marker of a defocused agent, otherwise also referred to as a passive marker.

As noted earlier in Section 2.2.2, in some Chinese varieties, such as the Jieyang Chaozhou variety of Southern Min discussed in Matthews et al. (2005), the ‘give’ constructions not only serve causative and passive functions, as in (23a) and (23b) respectively, but also unaccusative ones as well, as in (23c). Matthews et al. note that passive and unaccusative ‘give’ constructions – such as (23b) and (23c) respectively – share significant thematic and aspectual properties. Among these similarities, the internal argument occurs in subject (instead of object) position and both passive and unaccusative constructions are associated with the semantic feature of adversity. Both constructions also encode change of state, often signaled by the presence of a verbal complement accompanying the intransitive (or intransitivized) verb. 21 Matthews et al. suggest that these syntactic and semantic overlaps between the passive and unaccusative ‘give’ constructions could have facilitated the extension of passive ‘give’ constructions to unaccusative ones, along a Causative>Passive>Unaccusative pathway.


Permissive causative ‘allow’


‘He didn’t let me read this book.’ (Matthews et al. 2005: 270)




‘The book has been bought already.’ (Matthews et al. 2005: 271)




‘The wall fell down (by itself).’

(Matthews et al. 2005: 269)

In a subsequent study on another Southern Min variety, Lin (2011: 240) raises the possibility that the adversative (=our unaccusative) hoo7 ‘give’ construction in Taiwan Southern Min could also have developed in parallel with the passive hoo7 ‘give’ construction. More precisely, Lin left open two possible grammaticalization pathways for the emergence of the unaccusative hoo7 construction – one is the more direct permissive (causative) pathway as highlighted in (24) below, and the other a passive-mediated pathway, as highlighted in (25). However, Lin (2011) did not elaborate further on either pathway. Details of syntactic reanalysis and the pathway(s) from causative or passive to unaccusative thus still need to be further spelled out.


Permissive>Adversative (=our Unaccusative)


Permissive>Passive>Adversative (=our Unaccusative)

Recent work from Huang (2013) identifies unaccusative ‘give’ constructions as existential raising constructions, “which by virtue of the existence of an implicit affectee conveys a passive or passive-like meaning” (p. 112). Huang’s analysis highlights similarities between passive(-like) ‘give’ constructions and unaccusative ‘give’ constructions, and as noted earlier, he identifies Taiwanese Mandarin gĕi in unaccusative constructions as a marker of a ‘phantom’ (or implicit) affectee, which could refer to ‘others’ but as a default interpretation refers to the speaker. In this respect, Taiwanese Mandarin gei behaves more like its Southern Min counterparts rather than its counterparts in Mandarin varieties spoken in mainland China. 22

Huang uses the term ‘phantom’ affectee in the unaccusative ‘give’ construction to explain, in syntactic terms, how suppression of the causer allows the affected patient in the post-‘give’ predicate to be raised (via a fronting movement) to topic position. We here further suggest that suppression of the causer leaves the stranded causative ‘give’ verb open to ambiguous reinterpretations – induced through context – and facilitates its reanalysis as an adversity marker as well as a marker of speaker affectedness. This development is illustrated by the Taiwanese Mandarin gĕi ‘give’ examples in (26) below (from Huang 2013: 109, with additional interpretations for mainland Mandarin in (26a.i), (26b.i) and (26c.i) added by us). Unlike (26a.i) in mainland Mandarin, where the subject kanshou ‘guard’ could have either intentionally or unintentionally let the prisoner escape, in (26a.ii) in Taiwanese Mandarin, the subject kanshou ‘guard’ is helpless and lacks control over the unfortunate incident involving the escape of the prisoner. Huang (2013) identifies the use of gĕi ‘give’ in (26a.ii) as a “bystander” verb with a 2-place argument structure. In (26b), where the subject is suppressed, we can obtain an impersonal causer reading, typical for mainland Mandarin, where someone, something or some situation is seen to be responsible for the escape of the prisoner, as in (26b.i). However, as Huang suggests in (26b.ii), in scenarios that are more typical of Taiwanese Mandarin, such impersonal subject-suppressed constructions tend to be reanalyzed as existential raising constructions, with the gĕi ‘give’ morpheme reinterpreted as a semi-lexical unaccusative verb with the existential meaning ‘happen (to be)’, thus yielding 1-place predicate ‘give’ constructions with adversative readings such as ‘It happened (to be) that the prisoner ran away’. In (26c), raising the patient NP fanren ‘(the) prisoner’ to topic position has the effect of highlighting the affected patient, but as seen in (26c.i) and (26c.ii) respectively, the existence of a causer is still strongly implied in mainland Mandarin, while speakers of Taiwanese Mandarin tend to focus more on the existence of an affectee.

kanshougei[fanrenpao-le](mainland Mandarin)

‘The guard let the prisoner run away.’ (i.e., the guard as a causer)

kanshougei[fanrenpao-le] 23(Taiwanese Mandarin)

‘The guard had the prisoner running away.’ (i.e., the guard as an affectee)

[e]gei[fanrenpao-le](mainland Mandarin)

‘(Someone/Something/Some situation) had the prisoner running away.’ (subject elided)

[e]gei[fanrenpao-le](Taiwanese Mandarin)

‘It happened that the prisoner ran away.’ (subject suppressed)

fanrenigei[ti pao-le](mainland Mandarin)

‘The prisoner, alas, ran away.’ (implying the existence of a causer)

fanrenigei[ti pao-le](Taiwanese Mandarin)

‘It happened that the prisoner ran away.’ (implying the existence of an


3.2 Further insights into the emergence of unaccusative ‘give’ constructions

A major objective of this study is to examine in greater detail the grammaticalization pathways that give rise to unaccusative ‘give’ constructions in Mandarin and Southern Min, with an eye to shedding more light on the relationship between valence reduction phenomena and the expression of speakers’ subjective stance. To determine the viability of the two potential grammaticalization pathways to unaccusativity previously posited in the literature (e.g., Lin 2011), we will first examine the causative pathway mediated by unintentional causative ‘give’ constructions in mainland Mandarin varieties (Section 3.2.1), followed by the passive pathway mediated by ‘speaker-affected’ passive ‘give’ constructions in Southern Min varieties (Section 3.2.2).

3.2.1 From causative to unaccusative

In this section, we will show how ‘give’ constructions develop from causative to unaccusative constructions via an intermediate ‘unintentional causative’ stage, as illustrated by examples (27) to (31) below from the Harbin variety of Mandarin. We refer to this development as the causative pathway to unaccusativity.


Permissive causative


‘Mom allows our younger sister to go out and play.’


Permissive/unintentional causative

  1. ‘The police allowed the thief to get away.’ (Permissive causative)

  2. ‘The police let the thief get away.’ (Unintentional causative)

Whereas there is no ambiguity in meaning for the permissive causative construction in (27), the construction in (28) has two possible causative interpretations: permissive causative and unintentional causative. On the one hand, the police (jĭngchá) may intentionally let the thief (xiăotōu) get away, as in (28a); on the other hand, the police may in fact want to catch the thief, but unfortunately they failed to do so, as in (28b). These two interpretations are both possible, though the interpretation in (28b) is more natural for native speakers for pragmatic reasons, given that the police would not normally intentionally allow a thief to escape. This example constitutes a bridging context between permissive causative ‘permit, allow’ and unintentional causative ‘let’.

In the following, we will focus on the development from unintentional causative to unaccusative ‘give’ constructions. Note that the causer jĭngchá in (28b) can be omitted, as in (29) below.


‘Causer-elided’ unintentional causative


‘Unfortunately, (someone) let the thief get away.’ 24

As seen in (28b) and (29) above, unintentional causatives typically involve adversity, i.e., they involve an unfortunate and unexpected event. Omission of the causer as in (29) allows attention to be drawn to the adversative outcome of the event and its effect on the speaker. This facilitates the development of gĕi as an adversity marker.

Unintentional causative constructions sometimes involve topicalized patient NPs, as seen in (30) below, and the resumptive pronoun is often elided.


Unintentional causative with topicalization


‘As for the thief, (unfortunately) (someone) let him get away.’

In (31), unlike (30), there is no topic marker such as a 啊 following the topic NP xiǎotōu ‘thief’, nor is there a distinct pause between the topic NP and the gĕi ‘give’ morpheme, which suggests that the topic NP could have been reanalyzed as a grammatical subject. What is similar with the topicalized unintentional causative construction in (30) is the preferred elision of the third person singular pronoun and the strong adversity and speaker affectedness reading. This type of construction meets the criterion of an unaccusative ‘give’ construction. (31a) is a Type 1 unaccusative construction, given its reversible pǎo le ‘run away’ intransitive predicate, while (31b) and (31c) are Type 2 unaccusative constructions, given their irreversible intransitive predicates, sĭ le ‘die’ and suān le ‘turn sour’ (=‘turn bad’) respectively.


Type 1 unaccusative ‘give’ construction (causer-suppressed)


‘(Unfortunately) the thief got away.’


Type 2a unaccusative ‘give’ construction


‘(Unfortunately) the flower died.’


Type 2b unaccusative ‘give’ construction


‘(Unfortunately) the milk has turned sour (i.e., turned bad).’

In sum, we have seen that the speaker-affected unaccusative ‘give’ construction in the Harbin variety of Mandarin emerges via the causative pathway with the unintentional ‘give’ causative as an intermediate stage, as highlighted in Figure 1 below.

Development of unaccusative ‘give’ constructions in the Harbin variety of Mandarin via the causative pathway.
Figure 1:

Development of unaccusative ‘give’ constructions in the Harbin variety of Mandarin via the causative pathway.

This pathway may also be prominent among other Mandarin varieties as well as some other Chinese dialects (e.g., Shangzhuang Jixi Hui), since their unaccusative ‘give’ constructions also tend to focus on the existence of a causer, unlike similar constructions in the Southern Min varieties which tend to focus more on the resultative nuances (see Section 3.2.2 below).

3.2.2 From passive to unaccusative

In this section, we examine the possibility of how passive ‘give’ constructions pave the way for the rise of unaccusative ‘give’ constructions. Recall that Matthews et al. (2005) have suggested that passive morphosyntax can be extended to unaccusative predicates due to the parallels between passive ‘give’ constructions and unaccusative ‘give’ constructions in terms of syntactic structure and semantic properties, and Lin (2011) considers the developmental pathway from passive to unaccusative as potentially viable as well. In what follows, we will track this passive pathway to unaccusativity in Southern Min dialects, identifying the intermediate stages between passive and unaccusative with examples from Hui’an. Through this analysis, we will also identify stages in the grammaticalization process where the ‘give’ construction comes to increasingly express the speaker’s subjective stance – in contexts mediated by a strong presence of passive ‘give’ constructions.

Native speakers of Southern Min varieties such as Hui’an tend to associate the unaccusative ‘give’ construction more with the passive (rather than causative) ‘give’ construction. A typical example of passive ‘give’ construction in the Hui’an dialect is given in (32). 25


Typical passive ‘give’ construction


‘(I) got scolded by him.’

In this example, the first person singular pronoun ua3 in topic/subject position refers to the affected patient of the transitive verb 5 ‘scold’, while the third person singular pronoun i1 refers to the defocused agent. The ‘give’ morpheme khɔ5 is interpreted as a marker of the defocused agent, and is often also referred to as a passive marker. As noted earlier in Section 3.1, passive ‘give’ markers in Sinitic languages can be derived from permissive causative ‘give’ verbs via extensions into reflexive contexts; this includes the Hui’an passive ‘give’ marker khɔ5. In daily conversation, patient NPs in topic/subject position such as first person pronoun ua3 ‘I’ in (32) are often omitted, yielding a passive construction with the following structure: [(patient NP)TOPIC/SUBJECT+‘give’ passive marker+defocused agent NP+transitive predicate]. Given that the patient NP is typically a malefactee (rather than a beneficiary), as in the case of ua3 in (32) above, the passive ‘give’ construction in Hui’an has come to be strongly associated with an adversative reading, in that it often denotes an unexpected and unfortunate event. This adversative reading facilitates the rise of passive constructions with strong speaker-affectedness readings (i.e., utterances conveying the subjective (and often negative) evaluations of the speaker about the outcome of a particular situation). In the rest of this section, we will examine how passive ‘give’ constructions such as (33) facilitate the emergence of unaccusative ‘give’ constructions, focusing on the facilitative role of speaker-affected passive ‘give’ constructions as bridge-constructions.


Passive ‘give’ construction


‘Money was taken away by the thief worth 300 yuan.’

As seen in passive ‘give’ constructions such as (33) above, the topicalized patient NP (e.g., tsin2 ‘money’) can be elaborated with quantifying expressions (e.g., 1paʔ7−8 khɔ1 ‘(be worth) 300 yuan’). 26 The amount expressed conveys the speaker’s subjective evaluation of the value of the object (tsin2 ‘money’) stolen by the thief. It conveys not only the regular passive meaning of ‘money being taken away by a thief’ but also encodes the affective reading of ‘the speaker being adversely affected by an event involving the theft of a significant sum of 300 yuan.’

As seen in (34), when the defocused agent (e.g., tshat8−4-a3 ‘the thief’) is realized through the third person pronoun i1, ambiguity emerges as to whether the co-indexical referent is the defocused agent (=the thief) or the affected patient (=the money). This ambiguity further strengthens the reinterpretation of the ‘give’ morpheme as a dual agent-defocusing and patient-highlighting marker.


Passive ‘give+3SG’ construction

  1. ‘Unfortunately, money was taken away by him/her worth 300 yuan.’

  2. ?‘Unfortunately, money was taken away worth 300 yuan.’

  3. ?‘Unfortunately, money was taken away from me worth 300 yuan.’

Worth noting is that native speakers of Hui’an prefer the defocused agent interpretation when the predicate is transitive, as in (34.i), where the predicate thueʔ8−4khɯ0 ‘take go, steal’ still encodes two core arguments, namely, the internal argument tsin2 ‘money’ and the external argument i1 ‘him/her (=the thief)’. However, in the case of intransitive unaccusative predicates, which do not have external arguments, an even stronger patient-affectedness and speaker-affectedness reading will emerge, as we shall see later in (37) to (40).

As shown in (35a), the adversity and speaker-affectedness reading of the passive ‘give’ construction in Hui’an is further enhanced when the [‘give’+defocused agent] constituent is preposed to clause-initial position, leaving the undesirable outcome (tsin2 thueʔ8−4khɯ01paʔ7−8 khɔ1 ‘money being taken away (=stolen) worth 300 yuan’) in the spotlight in the information focus position (i.e., the position following the left-dislocated defocused agent). If the adversative situation involves money being stolen from the speaker, then the speaker is affected as a malefactee, but if the money was someone else’s, the speaker affectedness takes the form of empathy for someone else as the malefactee. The strong adversity and speaker-affectedness reading can be made even more explicit when the speaker is also overtly expressed as an affectee (in this case, a malefactee) in topic position (in the form of ua3 ‘I’) at the leftmost periphery of the utterance, as in (35b). Both preposed [‘give’+defocused agent] constructions in (35a) and (35b) have the effect of highlighting the speaker’s unexpectedness at what the thief has done, and elicits a strong reaction (e.g., one of shock) on the part of the speaker.


Speaker-affected passive ‘give’ construction (with preposed ‘give’ + defocused agent)


‘Unfortunately, money was taken away by the thief worth 300 yuan.’


Speaker-affected passive ‘give’ construction (with preposed ‘give’ + defocused agent, plus overt expression of the affected speaker)


‘Unfortunately, money was taken away by the thief worth 300 yuan,

and I am badly affected by this.’

Similar “speaker-affectedness” readings are obtained when the preposed [‘give’+defocused agent] constituent involves khɔ5-4 i1, where the high degree of pleonasticity of the third person pronoun i1 gives rise to ambiguity and could co-index a variety of referents, including not only a defocused agent as in (36a.i), but also potentially an affected patient as in (36a.ii) and implicit affectees that include the speaker as in (36a.iii). Explicit expression of an affected speaker (e.g., ua3 ‘I’), as in (36b), weakens the causative orientation and induces an even stronger resultative and passive reading.


(i) ‘Unfortunately, money was taken away by him/her worth 300 yuan.’

(ii) ?‘Unfortunately, money was taken away worth 300 yuan.’

(iii) ?‘Unfortunately, money was taken away from me worth 300 yuan.’


‘Unfortunately, money was taken away (by {him/her} / from me) worth

300 yuan, and I am badly affected by this.’

Note that passive ‘give’ constructions with strong speaker-affected readings, such as (35) and (36) above, abound in Southern Min varieties such as Hui’an, but are not found in mainland Mandarin varieties. These speaker-affected ‘give’ constructions involve complement structures that are morphosyntactically as well as semantically passive, with the patient NP (e.g., tsin2 ‘money’) occurring in preverbal position (e.g., tsin2 thueʔ8−4khɯ01paʔ7−8 khɔ1 lit. ‘money taken away 300 yuan’). These passive constructions are often further characterized by left-dislocated defocused agents, such as khɔ5-4 tshat8−4-a3 ‘by the thief’ in (35) and khɔ5-4 i1 ‘by him/her’ in (36), and often the speaker (ua3) is explicitly encoded as an affectee at the leftmost periphery. The abundance of these passive constructions in Southern Min (but crucially not in mainland Mandarin) helps to explain why native speakers of Southern Min tend to tilt more toward a result state and speaker-affected (often malefactee-oriented) reading in their use of ‘give’ constructions.

The adversity and ‘speaker-affectedness’ reading in passive ‘give’ constructions such as (33) to (36) above can be further extended in Hui’an to intransitive unaccusative predicates as well. This represents an extension, via analogy, from 2-place predicate environments to 1-place predicate environments. The results include an unaccusative ‘give’ construction without topicalization of the patient NP, as in (37), and an unaccusative ‘give’ construction with the patient NP topicalized, as in (38a). The latter type, which involves the presence of the pleonastic third person singular pronoun i1, is what we have referred to as the Type 1 unaccusative ‘give’ construction (see Section In Hui’an, both types of constructions – with or without topicalized patient NPs – yield a bias toward a result state and patient-affected reading with an implicit affectee that includes the speaker. This differs from mainland Mandarin, where ‘give’ constructions that are structurally similar to the Hui’an examples in (37) and (38) below yield a stronger “implicit causer” rather than “implicit affectee” reading for unaccusative ‘give’ constructions.


‘Escape’-type unaccusative ‘give’ construction [-patient NP topicalization]


‘Unfortunately, the thief got away.’


‘Unfortunately, s/he got away.’


Type 1 (‘escape’-type) unaccusative ‘give’ construction [+topicalized patient NP]


‘Unfortunately, the thief got away.’


‘Unfortunately, the thief got away.’

Note that in Type 1 (‘escape’-type) unaccusative ‘give’ constructions such as (38a), which are characterized by the presence of a patient NP in topic position, only khɔ5-4 i1 (but not khɔ5-4+NP) can be used in co-indexical fashion. Hence the acceptability of (38a) but not that of (38b). Crucially, not only does the khɔ5-4 i1 form have a referential function that co-indexes with the patient NP in topic position, it also has an evaluative and attitudinal function of marking the speaker’s affected stance as well (as reflected in the evaluative adverb unfortunately in the English translation). As noted earlier, this speaker-affectedness reading emerges from the strong association of the unaccusative predicate with an adversative context, and also from the pleonasticity of the third person pronoun i1 which could ambiguously refer, not only to an implicit causer or an affected patient in topic position, but also to implicit ‘other’ referents that include the speaker. As noted earlier, native speakers of Southern Min varieties such as Hui’an lean more toward an “implicit affectee” reading, suggesting that they are more empathy-oriented (e.g., ‘{Pity the guards / Pity us}, the thief got away’), while their mainland Mandarin counterparts tend to favor an “implicit causer” reading, sometimes involving covert criticism and assignment of blame (e.g., ‘It’s terrible, (how could it have happened / who let it happen) that the thief got away’).

The bias toward an implicit affectee (rather than implicit causer) reading can also be seen when the khɔ5-4 i1 form extends to ‘die’-type unaccusative ‘give’ constructions, either with or without patient NP topicalization, as in (39) and (40) respectively.


‘Die’-type unaccusative ‘give’ construction [-patient NP topicalization]


‘Unfortunately, the flower died.’


Type 2 (‘die’-type) unaccusative ‘give’ construction [+topicalized patient NP]

伊 /(*與 花)(=6b)
hue1khɔ5-4i 1 /(*khɔ5-4 hue1)si3khɯ0
flowergive3SG /(*give flower)diego

‘Unfortunately, the flower died.’

Thus, in addition to the more direct causative>unintentional causative>unaccusative pathway noted earlier for mainland Mandarin varieties (see Section 3.2.1), in Southern Min varieties such as Hui’an we also see a grammaticalization pathway mediated by passive ‘give’ constructions. Some of these Southern Min passive constructions are not attested in mainland Mandarin varieties – for example, the passive constructions with left-dislocated defocused agents such as (35) and (36) above.

Worth noting is that these passive constructions are patient-highlighting strategies involving adversative outcomes that inevitably evoke strong affective responses not only from the patient NP (if animate) but also from other referents including the speaker. Southern Min varieties such as Hui’an are known to be extremely rich in patient-highlighting constructions, including the ka i disposal constructions and other subtopic constructions (see Section for further elaboration). These passive and other passive-like patient-highlighting constructions contribute to a bias in Southern Min toward a covert affectee reading, which in the case of unaccusative ‘give’ constructions always involves the speaker. In other words, for native speakers of Southern Min varieties such as the Hui’an dialect, the unaccusative ‘give’ constructions have thus come to subtly yet unmistakably signal the speaker’s subjective evaluation of an event as ‘something unexpected and unfortunate’.

Thus, in this section we see that in addition to the unintentional causative pathway for unaccusative ‘give’ constructions typically noted in Mandarin varieties discussed earlier in Section 3.2.1, in Southern Min varieties (e.g., Hui’an) there is also the passive-mediated pathway for unaccusative ‘give’ constructions, as shown in Figure 2 below. Crucially, the mediation takes the form of analogy rather than derivation.

Development of unaccusative ‘give’ constructions in Hui’an via the passive pathway.
Figure 2:

Development of unaccusative ‘give’ constructions in Hui’an via the passive pathway.

To sum up our analysis thus far, in addition to marking a defocused agent, passive marker ‘give’ can also mark speaker affectedness by virtue of the frequent association of passive constructions with adversative situations or outcomes. This speaker- affectedness marking function then extends from passive ‘give’ constructions to unaccusative ‘give’ constructions in some Chinese dialects, in particular Southern Min varieties (e.g., Hui’an, Jieyang Chaozhou, and Taiwan Southern Min). Cross-dialectal data (see Table 2 in Section indicate that this extension initially affected ‘escape-type’ unaccusatives and subsequently ‘die’-type unaccusatives. An important factor for this extension from Type 1 to Type 2 unaccusative ‘give’ constructions appears to be the weakening of animacy constraints in some dialects. Essentially, the mediation of the passive ‘give’ constructions in the rise of the unaccusative ‘give’ constructions is one via semantic extension from 2-place (transitive passive) predicates to 1-place (intransitive unaccusative) predicates, essentially via analogy. Within the intransitive predicate domain, the extension proceeded from Type 1 to Type 2 unaccusative constructions. Both the passive and unaccusative constructions can be traced back to the causative ‘give’ construction as the derivational source.

3.2.3 Comparison of unintentional causative vs. passive-mediated pathways to unaccusativity

Given that the causative ‘give’ construction is a common source for both the causative and passive ‘give’ pathways to unaccusativity, it is highly likely that the emergence of the unaccusative ‘give’ construction in Hui’an and other Southern Min varieties has also been facilitated by the unintentional causative ‘give’ construction. Nevertheless there is strong evidence of a bias toward a resultative (rather than causative) reading in the Southern Min unaccusative ‘give’ construction, and this bias correlates with a strong preponderance for passive (and passive-like) ‘give’ constructions in the Southern Min dialects.

In this section, we will first highlight some salient similarities and differences between unintentional causative ‘give’ constructions and passive ‘give’ constructions (Section We then follow up with a brief discussion of various types of constructions that highlight the affected patient NP, among them the ka5-4 i1 disposal construction commonly found in Southern Min dialects (see Section This will shed light on how and why unaccusative ‘give’ constructions in Southern Min differ slightly in meaning from those in mainland Mandarin. Unintentional causative ‘give’ vs. passive ‘give’

We have seen from Sections 3.2.1 and 3.2.2 that both the unintentional causative pathway and the passive-mediated pathway can facilitate the emergence of unaccusative ‘give’ constructions. As highlighted in (41) and (42) below, in mainland Mandarin, an unintentional causative ‘give’ construction (41a) can develop into a causative-oriented unaccusative ‘give’ construction (41b) by retaining an ‘implicit causer’ reading, while in Hui’an, a passive ‘give’ construction (42a) can facilitate the rise of a resultative-oriented unaccusative ‘give’ construction (42b) by defocusing the agent and highlighting the affected patient and affected others (including the speaker). In both types of development, we see an extension in the domains of use of the ‘give’ construction to 1-place intransitive predicates. An important difference is that a stronger ‘implicit causer’ reading is retained along the unintentional causative>unaccusative pathway, while a stronger ‘implicit affectee’ reading emerges in the causative>passive>unaccusative pathway. This distinction over which argument type receives greater attention (i.e., ‘implicit causer’ vs. ‘implicit affectee’) is preserved even as the domains of use are extended from Type 1 (‘escape’-type) constructions to Type 2 (‘die’-type) constructions, as seen in the subtly different readings of (41c) and (42c).


Unintentional causative with topicalization (mainland Mandarin)


‘As for the thief, (unfortunately) (someone) let him get away.’


Type 1 unaccusative ‘give’ construction (mainland Mandarin)

小偷(他)了 (=31a)

‘(Unfortunately) the thief got away (e.g., because someone was negligent).’


Type 2 unaccusative ‘give’ construction (mainland Mandarin, e.g., Harbin)

花儿(他)了 (=31a)

‘(Unfortunately) the flower died (because someone didn’t take care of it).’


Passive construction (Hui’an)

tsin2khɔ5-4i1thueʔ8-4khɯ01paʔ7−8 khɔ1

‘Unfortunately, money was taken away by him/her worth 300 yuan.’


Type 1 unaccusative ‘give’ construction (Hui’an)


‘Unfortunately, the thief got away (and I’m worried about it).’


Type 2 unaccusative ‘give’ construction (Hui’an)


‘Unfortunately, the flower has died (and I’m upset about it).’

There is also evidence from other languages indicating that different types of ‘give’ constructions tend to profile different phases of a causative event. For example, some of the -v- adversative constructions in Even, a Manchu-Tungusic language, are more amenable to a causative interpretation while others are instead more amenable to a passive interpretation (Malchukov 1993). 27

In this section we have seen that languages – and dialects or varieties within a given language family – can vary in the different phases of a causative event that each is inclined to profile. In our analysis of unaccusative ‘give’ constructions in different varieties of Chinese, we have found that native speakers of mainland Mandarin tend to lean toward the enabling phase of a causative event and hence typically arrives at an ‘implicit causer’ reading, while native speakers of Southern Min varieties such as Hui’an lean more toward the result state phase and typically favor an ‘implicit affectee’ reading instead. In terms of grammaticalization, we have seen that unintentional causative ‘give’ constructions can facilitate the rise of unaccusative ‘give’ constructions with a bias toward an ‘implicit causer’ interpretation, as noted in the case of mainland Mandarin varieties such as Harbin Mandarin. On the other hand, in Southern Min varieties such as Hui’an, passive ‘give’ constructions mediate the rise of unaccusative ‘give’ constructions with a strong result-state and ‘implicit affectee’ interpretation. To a large extent, this is possible because the passive construction is a patient-highlighting construction. In the next section we will show how other patient-highlighting constructions found in Southern Min dialects such as Hui’an also helps to explain why there is a strong ‘implicit affectee’ bias within this dialectal group. Other patient-highlighting constructions in Southern Min

Extensions from transitive predicate contexts to unaccusative intransitive contexts through analogy are not automatic and are not attested for all Chinese varieties. There need to be favorable conditions. Examples of these facilitative conditions for Hui’an have been discussed in Section 3.2.2 earlier, with the availability of various subtopic constructions such as (36) and (37) respectively. Besides these, there are also other types of patient-highlighting subtopic constructions. These include the basic subtopic construction, with the patient NP in preverbal (as opposed to postverbal position) as shown in (43a) with its [(agent NP)+patient NP+verb] configuration. There is also a slightly more elaborate subtopic construction that is accompanied by a quantifying expression after the verb, as in (43b), with the configuration [(agent NP)+patient NP+verb+quantifying expression]. Both these patient-highlighting constructions come without a patient marker, relying only on the strategy of the patient NP being preposed to a position between an elidible subject and the verb. The use of these subtopic constructions is rather restricted in mainland Mandarin.


‘(S/he) has drunk the soup.’


‘(You) fry some cauliflower.’

Some subtopic constructions come with an overt ka-type patient marker. These ka-subtopic constructions are a type of ‘disposal construction’ in Southern Min. 28 As highlighted in (44), the disposal marker ka5 is deployed to yield an [(agent NP)+patient NP+ka5-4 i1+VP] configuration. Note that the Hui’an (as well as other Southern Min) ka-type disposal marker introduces a resumptive patient pronoun to further highlight the preposed patient NP, and thus the disposal construction constitutes a patient-highlighting subtopic construction


Hui’an ka5-4 i1 disposal construction


‘Close the door.’ Lit. ‘(You) (the door) for it close off.’

The frequent use of this patient-highlighting disposal construction in Southern Min dialects (e.g., Hui’an), which focuses on the patient NP and the result state (or outcome) of a causative event, helps explain why their unaccusative ‘give’ constructions tilt toward an “implicit affectee” reading.

The pervasive use of these (and other) patient-highlighting constructions in Southern Min varieties helps to create an environment that is conducive to the emergence of the unaccusative ‘give’ construction in this dialect group. It also explains the asymmetry whereby mainland Mandarin varieties rely predominantly on the causative pathway, via the unintentional causative construction, to develop their unaccusative ‘give’ construction that favors an ‘implicit causer’ reading, while the Southern Min dialects have additional recourse to a passive-mediated pathway that induces a stronger “implicit affectee” reading for its particular brand of unaccusative ‘give’ constructions.

4 Conclusion

This paper has examined some defining characteristics of the unaccusative ‘give’ construction in various Chinese dialects, with special reference to its function as an adversity and speaker-affectedness marker in Southern Min and Mandarin varieties. We distinguish between two types of unaccusative ‘give’ constructions: Type 1 (the ‘escape’-type) involves a reversible situation, while Type 2 (the ‘die’-type) involves an irreversible situation. Type 1 constructions typically involve unaccusative predicates derived from the combination of unergative verbs and resultative verbal complements (i.e., “unaccusativized” predicates), whereas Type 2 constructions typically involve pure unaccusative verbs, as well as stative verbs (or adjectives) that typically denote adversative resultant states (e.g., ‘something turning bad/rotten’). Type 1 unaccusative ‘give’ constructions are attested in many Chinese varieties, such as Mandarin, Min, Wu, Hui, Hakka and Cantonese, whereas Type 2 unaccusative ‘give’ constructions are more restricted and thus far have mainly been noted in Southern Min and Mandarin varieties.

Some typological differences have also been identified, namely, that some dialects (particularly Mandarin) prefer to elide the post-‘give’ third person pronoun, while other dialects (for example, the Southern Min varieties) prefer to retain the pronoun and have grammaticalized it into a pleonastic form that is no longer sensitive to person and number agreement. This semantic bleaching of the pleonastic pronoun increases its versatility, and is crucial to its development beyond marking a defocused agent in passive constructions. Thus, in unaccusative constructions, the pleonastic third person pronoun can also mark a resumptive patient, which co-indexes with the patient NP in topic position. In addition, this pleonastic pronoun can also co-index implicit affectees that include the speaker, and in this way evolve into a pragmatic marker of speaker affectedness, often induced by adversative contexts and thus often yielding a malefactee-oriented reading. In this regard, the unaccusative ‘give’ construction constitutes a subjectification mechanism that, although typologically relatively rare, provides some Chinese dialects (e.g., Southern Min and Mandarin) with a means to express the speaker’s subjective stance.

Our analysis further reveals that there are at least two major pathways in the development of unaccusative ‘give’ constructions, namely, the unintentional causative pathway and the passive-mediated pathway, as highlighted in (45.i) and (45.ii) below. Both pathways are derived from the causative ‘give’ construction. However, unaccusative ‘give’ constructions in some Chinese dialects (such as the mainland Mandarin varieties) involve a route via unintentional causative ‘give’ constructions. This helps account for their stronger “implicit causer” reading. Other dialects (such as Southern Min varieties) are more strongly influenced by a rich array of patient-highlighting constructions. Among them are the passive ‘give’ construction and various types of passive-like subtopic constructions (including the special ka i disposal construction), which helps explain their bias toward an “implicit affectee” reading. In fact, in Southern Min varieties, the unaccusative ‘give’ construction is not unique in developing into a marker of speaker affectedness; there is evidence of unaccusative ka ua constructions in Southern Min that also serve speaker affectedness-marking functions, which we will further investigate in forthcoming work.


  1. The causative pathway

    Lexical ‘give’>causative>unintentional causative>

    ‘speaker-affected’ unaccusative (with ‘implicit causer’ reading)

  2. The passive-mediated pathway

Lexical ‘give’>causative>passive>‘speaker-affected’ unaccusative

(with ‘implicit affectee’ reading)

Essentially, our analysis suggests that when the erstwhile trivalent ‘give’ verb comes to serve as a marker of a covert affectee (or “phantom affectee” [Huang 2013]) that includes the speaker, we are witnessing the emergence of a pragmatic marker, more specifically a marker of the speaker’s subjective and affective evaluation of a causative event or situation, and usually one that is adversative in nature.

From a crosslinguistic perspective, it is worth noting that speaker-affectedness markers such as Hui’an khɔ5-4 i1 and Mandarin gěi (tā) in unaccusative ‘give’ constructions occupy a “clause-medial” position. 29 An interesting question for future research is to determine in what ways these non-periphery speaker stance markers are similar to or different from their left and right periphery counterparts (i.e., pragmatic particles in clause-initial and clause-final positions), which also frequently express speaker stance. 30 This question is also intimately linked to an issue raised earlier: to what extent is the speaker-affectedness marker autonomous or integrated within the intransitive unaccusative ‘give’ construction. Further work on the prosody and syntax of unaccusative ‘give’ constructions in these Chinese dialects may also help to shed more light on clause-medial (or non-periphery)-type subjective and intersubjective constructions in other languages as well.


This study was supported by a research grant from the Department of English, Hong Kong Polytechnic University for a pilot project entitled “Valence-Changing Phenomena and Implications for Argument Focus in Discourse” (HKPU 4-ZZAP) and by the Research Grants Council of Hong Kong for a larger-scale crosslinguistic project entitled “Valence-reducing Phenomena and the Emergence of Markers of Speaker Affectedness” (GRF PolyU Project No. 154055/14H). The research leading to these results has also received partial funding from the project sponsored by the Scientific Research Foundation for Returned Overseas Chinese Scholars, State Education Ministry (No. 49), and from the Major Program of the National Social Science Foundation of China (Grant No. 14ZDB098). An earlier version of this study was presented at the Annual Research Forum (ARF-2011) of the Linguistic Society of Hong Kong on December 3, 2011, the Spring Seminars of Sinotype, CRLAO, EHESS in Paris on March 20, 2012, the 22nd Annual Conference of the Southeast Asian Linguistics Society (SEALS 22) at Agay, France on June 2, 2012, the 20th Annual Conference of the International Association of Chinese Linguistics (IACL-20) in Hong Kong on August 30, 2012, and the Workshop on Structural Changes in Heritage Languages at Noordwijkerhout, the Netherlands on January 25, 2013, and we wish to thank the participants for their valuable feedback. We also wish to thank the editor and reviewers, and Prof. Stephen Matthews, Prof. Hilary Chappell, Dr. Marco Caboara, Dr. Jian Wang, Dr. Xuping Li, Tak-sum Wong, Shanshan Lv and Na Song for their insightful comments. We also wish to thank the following people for providing us useful information on various Chinese dialects: Prof. Zeping Chen, Prof. Rulong Li, Prof. Danqing Liu, Prof. Wuyun Pan, Prof. I-wen Su, Prof. Rujie You, Prof. Changcai Zhao, Dr. Xiaoying Hou, Dr. Tingting Huang, Dr. Xuping Li, Dr. Zhengda Tang, Dr. Jian Wang, Dr. Ruiyuan Xu, Ariel Chan, Yu-ying Chuang, Songdan Guo, Lily He, Jianguo Li, Jiuyang Li, Naihsin Li, Kathy Lin, Liming Liu, Melody Lu, Shanshan Lǚ, Ernest Pan, Yi Shen, Na Song, Susie Sui, Jessie Wang, Thomas Wang, Yan Wang, Tak-sum Wong, Andrew Wu, Vivien Yang, Yuqing Yang, Jerome Zhang, Peijia Zhang, and Yolanda Zhang.


  • Burzio, Luigi. 1986. Italian syntax: A government-binding approach. Dordrecht: Kluwer. Google Scholar

  • Chao, Yuen-Ren. 1968. A grammar of spoken Chinese. Berkeley, CA: University of California Press. Google Scholar

  • Chappell, Hilary, Alain Peyraube & Yunji Wu. 2011. A comitative source for object markers in Sinitic languages: 跟 kai55 in Waxiang and 共 kang7 in Southern Min. Journal of East Asian Linguistics 20(4). 291–338. CrossrefGoogle Scholar

  • Chen, I-Hsuan. 2009. Causative variants and related passives in Southern Min: A case study of grammaticalization. Taiwan: National Tsing Hua University MA Thesis. Google Scholar

  • Chen, Weirong. 2011. The Southern Min dialect of Hui’an: Morphosyntax and grammaticalization. Hong Kong: The University of Hong Kong dissertation. Google Scholar

  • Cheng, L.-S. Lisa, C.-T. James Huang, Y.-H. Audrey Li & C.-C. Jane Tang. 1999. Hoo, hoo, hoo: Syntax of the causative, dative and passive construction in Taiwanese. In Pang-Hsin Ting (ed.), Contemporary studies on the Min dialects. Journal of Chinese Linguistics (Monograph series 14), 146–203. Hong Kong: Chinese University Press. Google Scholar

  • Chin, Andy. 2011. Grammaticalization of the Cantonese double object verb [pei35] 畀 in typological and areal perspectives. Language and Linguistics 12(3). 529–563. Google Scholar

  • Chor, Winnie. 2010. Inclusion of the outsider – Grammaticalization of the verbal particle maai in Cantonese. Taiwan Journal of Linguistics 8(2). 33–62. Google Scholar

  • Chor, Winnie. 2013. From ‘direction’ to ‘positive evaluation’: On the grammaticalization, subjectification and intersubjectification of faan1 ‘return’ in Cantonese. Language and Linguistics 14(1). 91–134. Google Scholar

  • Ding, Shengshu, Shuxiang Lǚ, Rong Li, Dexuan Sun, Xiechu Guan, Jing Fu, Shengzhang Huang & Zhiwen Chen. 1979. Xiandai hanyu yufa jianghua [Lectures on modern Chinese grammar]. Beijing: Commercial Press. Google Scholar

  • Hashimoto, Mantaro. 1986. The Altaicization of northern Chinese. In John F. McCoy & Timothy Light (eds.), Contribution to Sino-Tibetan studies, 76–97. Leiden: Brill. Google Scholar

  • Hashimoto, Mantaro. 1988. The structure and typology of the Chinese passive construction. In Masayoshi Shibatani (eds.), Passive and voice, 329–354. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Huang, C.-T. James. 2013. Variations in non-canonical passives. In Artemis Alexiadou & Florian Schäfer (eds.), Non-canonical passives, 95–114. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Jiang, Shaoyu. 2002. “gei”ziju、 “jiao”ziju biao beidong de laiyuan – Jiantan yufahua、leitui he gongneng kuozhan [The origin of passive sentence with gei and jiao – Grammaticalization, analogy and expansion of grammatical function]. Yuyanxue Luncong [Essays on Linguistics] 26. 159–177. Google Scholar

  • Knott, Judith. 1995. The causative-passive correlation. In David C. Bennett, Theodora Bynon & George B. Hewitt (eds.), Subject, voice, and ergativity: Selected essays, 53–59. London: SOAS, University of London. Google Scholar

  • Li, Charles N. & Sandra A. Thompson. 1981. Mandarin Chinese: A functional reference grammar. Berkeley, CA: University of California Press. Google Scholar

  • Li, Rulong. 1997. Quanzhou fangyan de dongci weiyuju [Verb predicate sentences in the Suzhou dialect]. In Rulong Li & Song-Hing Chang (eds.), Dongci weiyuju [Verb predicate sentences], 121–135. Guangzhou: Ji’nan University Press. Google Scholar

  • Li, Rulong & Song-Hing Chang (eds.). 1997. Dongci weiyuju [Verb predicate sentences]. Guangzhou: Ji’nan University Press. Google Scholar

  • Li, Rulong & Song-Hing Chang (eds.). 2000. Jieci [Adpositions]. Guangzhou: Ji’nan University Press. Google Scholar

  • Lin, Huei-Ling. 2011. Pure unaccusatives with HOO I in Taiwan Southern Min. Lingua 121. 2035–2047. CrossrefGoogle Scholar

  • Lin, Huei-Ling. 2012. A non-uniform analysis of ka7 constructions in Taiwan Southern Min. Language and Linguistics 13(4). 743–766. Google Scholar

  • Lin, Lunlun. 1996. Chenghai fangyan yanjiu [A study of the Chenghai dialect]. Shantou: Shantou University Press. Google Scholar

  • Liu, Danqing. 1997. Suzhou fangyan de dongci weiyuju [Verb predicate sentences in the Suzhou dialect]. In Rulong Li & Song Hing Chang (eds.), Dongci weiyuju [Verb predicate sentences], 1–20. Guangzhou: Ji’nan University Press. Google Scholar

  • Lord, Carol. 1993. Historical change in serial verb constructions (Typological Studies in Language 26). Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Lord, Carol, Foong-Ha Yap & Shoichi Iwasaki. 2002. Grammaticalization of ‘give’: African and Asian perspectives. In Ilse Wischer & Gabriele Diewald (eds.), New reflections on grammaticalization (Typological Studies in Language 49), 217–135. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Lü, Shuxiang. 1980. Xiandai hanyu babaici [800 words in modern Chinese]. Beijing: Commercial Press. Google Scholar

  • Malchukov, Andrej. 1993. Adversative constructions in Even in relation to passive and permissive. In Bernard Comrie & Maria Polinsky (eds.), Causatives and transitivity, 369–384. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Matthews, Stephen, Huiling Xu & Virginia Yip. 2005. Passive and unaccusative in Jieyang dialect of Chaozhou. Journal of East Asian Linguistics 4(4). 267–298. Google Scholar

  • Matthews, Stephen & Virginia Yip. 2008. Passive, unaccusative and pretransitive constructions in Chaozhou. In Redouane Djamouri, Barbara Meisterernst & Rint Sybesma (eds.), Chinese Linguistics in Leipzig, 163–174. Paris: Centre de Recherches Linguistiques sur l’Asie Orientale. Google Scholar

  • McCawley, James D. 1992. Justifying part-of-speech assignments in Mandarin Chinese. Journal of Chinese Linguistics 20(2). 211–246. Google Scholar

  • Mei, Tsu-Lin. 2005. Minnanhua hoo “geiyu” de benzi jiqi yufa gongneng de laiyuan [The origin of the ‘give’ verb hoo in Southern Min]. In Dah-An Ho & Ovid J. L Tzeng (eds.), Yongyuan de POLA : Wangshiyuan xiansheng qizhi shouqing lunwenji [POLA Forever : Festschrift in honor of Professor William S-Y. Wang on his 70th birthday], 163–173. Taipei: Academia Sinica. Google Scholar

  • Nedjalkov, Igor. 1993. Causative-passive polysemy of the Manchu-Tungusic -bu/-v(u). Linguistica Antverpiensa 27. 193–202. Google Scholar

  • Pan, Wuyun. 1997. Wenzhou fangyan de dongci weiyuju [Verb predicate sentences in the Suzhou dialect]. In Rulong Li & Song-Hing Chang (eds.), Dongci weiyuju [Verb predicate sentences], 58–75. Guangzhou: Ji’nan University Press. Google Scholar

  • Perlmutter, David M. 1978. Impersonal passives and the Unaccusative Hypothesis. In Jeri Jaeger, et al (eds.), Proceedings of the 4th Annual Meeting of the Berkeley Linguistics Society, 157–189. Berkeley, CA: Berkeley Linguistics Society. Google Scholar

  • Shen, Yang & Rint Sybesma. 2010. Jufa jiegou biaoji gei yu dongci jiegou de yansheng guanxi [On the marker gei and its relation to the derivation of VP structure]. Zhongguo Yuwen 3. 222–237. Google Scholar

  • Shi, Rujie. 1997. Gaochun fangyan de dongci weiyuju [Verb predicate sentences in the Gaochun dialect]. In Rulong Li & Song-Hing Chang (eds.), Dongci weiyuju [Verb predicate sentences], 21–38. Guangzhou: Ji’nan University Press. Google Scholar

  • Shibatani, Masayoshi. 1985. Passives and related constructions: A prototype analysis. Language 61(4). 821–848. CrossrefGoogle Scholar

  • Traugott, Elizabeth C. 1995. Subjectification in grammaticalization. In Dieter Stein & Susan Wright (eds.), Subjectivity and subjectivisation: Linguistic perspectives, 31–54. Cambridge: Cambridge University Press. Google Scholar

  • Traugott, Elizabeth C. 2003. From subjectification to intersubjectification. In Raymond Hickey (ed.), Motives for language change, 124–139. Cambridge: Cambridge University Press. Google Scholar

  • Traugott, Elizabeth C. 2010. Revisiting subjectification and intersubjectification. In Kristin Davidse, Lieven Vandelanotte & Hubert Cuyckens (eds.), Subjectification, intersubjectification and grammaticalization, 29–70. Berlin: De Gruyter Mouton. Google Scholar

  • Wan, Bo. 1997. Anyi fangyan de dongci weiyuju [Verb predicate sentences in the Anyi dialect]. In Rulong Li & Song-Hing Chang (eds.), Dongci Weiyuju [Verb predicate sentences], 229–246. Guangzhou: Ji’nan University Press. Google Scholar

  • Wong, Tak-Sum, Foong-Ha Yap & Weirong Chen. 2014. Valence-reducing phenomena and stance marking in Chinese: On the development of kangin Classical Chinese and Southern Min. Paper presented at the Conference on Language, Discourse and Cognition, National Taiwan University, Taipei, 3–4 May. Google Scholar

  • Xu, Liejiong & Danqing Liu. 2007. Huati de Jiegou yu Gongneng [Topic: Structural and functional analysis]. Shanghai: Shanghai Education Press. Google Scholar

  • Yap, Foong-Ha & Winnie Oi-Wan Chor. 2014. Epistemic, evidential and attitudinal markers in clause-medial position in Cantonese. In Elisabeth Leiss & Werner Abraham (eds.), Modes of modality: Modality, typology and universal Grammar (Studies in Language Companion Series 149), 219–260. Amsterdam & Philadelphia: John Benjamins. Google Scholar

  • Yap, Foong-Ha & Shoichi Iwasaki. 1998. ‘Give’ constructions in Malay, Thai and Mandarin Chinese: A polygrammaticization perspective. In M. Catherine Gruber, Derrick Higins, Kenneth S. Olson & Tamra Wysocki (eds.), CLS 34, Part 1: Papers from the main session, 421–437. Chicago: Chicago Linguistic Society. Google Scholar

  • Yap, Foong-Ha & Shoichi Iwasaki. 2003. From causative to passive: A passage in some East and Southeast Asian languages. In Eugene H. Casad & Gary B. Palmer (eds.), Cognitive linguistics and non-Indo-European languages (Cognitive Linguistics Research 18), 419–446. Berlin & New York: Mouton de Gruyter. Google Scholar

  • Yap, Foong-Ha & Shoichi Iwasaki. 2007. The emergence of ‘give’ passives in East and Southeast Asian languages. In Mark Alves, Paul Sidwell & David Gil (eds.), Proceedings of the Eighth Annual Meeting of the Southeast Asian Linguistics Society, 193–208. Canberra: Pacific Linguistics. Google Scholar

  • Yap, Foong-Ha, Ying Yang & Tak-Sum Wong. 2014. On the development of sentence final particles (and utterance tags) in Chinese. In Kate Beeching & Ulrich Detges (eds.), Discourse functions at the left and right periphery: Crosslinguistic investigations of language use and language change (Studies in Pragmatics 12), 179–220. Leiden: Brill. Google Scholar

  • Ye, Xiangling (ed.). 1993. Suzhou fangyan cidian [Suzhou dialect dictionary]. Nanking: Jiangsu Educational Publishing House. Google Scholar

  • Yue-Hashimoto, Anne. 1976. Southern Chinese dialects – The Tai connection. Computational Analyses of Asian and African Languages 6. 1–9. Google Scholar

  • Zhang, Bojiang. 1997. Xingzhi xingrongci de fanwei he cengci [The scope and scale of adjectives]. Yufa yanjiu he tansuo [Study and exploration of grammar] 8. 50–61. Google Scholar

  • Zhang, Min. 2000. The grammaticalization of ‘give’ in Chinese dialects: A cognitive approach. Paper presented at the 9th International Conference on Chinese Linguistics, Singapore, 25–27 June. Google Scholar

  • Zhu, Dexi. 1982. Yufa jiangyi [Grammar issues]. Beijing: Commercial Press. Google Scholar


  • 2

    The term unaccusative construction comes from Matthews et al. (2005). 

  • 4

    The Hui’an variety of Southern Min, spoken in Hui’an County in Fujian province in China, belongs to the Quan (Quanzhou)-Zhang (Zhangzhou) subgroup of Southern Min. The Chaozhou variety of Southern Min is spoken in the northeastern part of Guangdong province in China, “in the area including and surrounding Shantou, Chao’an and Jieyang; and in diasporic communities in Hong Kong and overseas Chinese communities, particularly in southeast Asia” (Matthews and Yip 2008: 163). 

  • 5

    People may wonder about the status of the patient/theme NP in unaccusative ‘give’ constructions: Is it a subject, topic or something else? As is well known, Chinese dialects are topic-prominent, and the subject typically involves an agent NP. In this paper, the patient/theme NP is provisionally regarded as a topic. However, we need to keep in mind that Chinese dialects vary in the degree of syntacticization of the topic (cf. Xu and Liu 2007). 

  • 7

    Similar unaccusative constructions are also attested in the Anyi variety of Gan (Wan 1997:233). In this variety, the t’au 討-passive construction is further extended to the unaccusative t’au construction, and the author suggests that the passive marker t’au, originally a verb meaning ‘ask for’, may be traced back to its ‘give’ meaning, since the ‘give’ meaning can be attested in the verb phrase t’au sau 討潲 ‘give pigswill, feed the pigs’ in Modern Anyi (Wan 1997:237–238). 

  • 8

    As noted in Matthews et al. (2005), morphosyntactic parallels between passive and unaccusative constructions – not necessarily involving the ‘give’ morpheme – can be found in a number of other languages as well, for example, Latin, Italian and Albanian, as well as English interlanguage grammars across a number of different L1 (i.e., first language) backgrounds. 

  • 9

    According to Liu (1997:6), the unaccusative ‘give’ construction in the Suzhou Wu dialect was regarded as a special type of pəʔ5 sentence (<pəʔ5 ‘give’), since it involves intransitive predicates, instead of taking the form of a typical pəʔ5-passive construction, i.e., ‘patient NP+passive marker pəʔ5+agent NP+transitive verb (+resultative element)’, and it does not express a typical passive meaning, though it shares some similar characteristics with pəʔ5-passive constructions. 

  • 11

    Note that the verbal complement khɯ°, derived from the motion verb khɯ5 ‘go’, shows tonal neutralization (a change from tone 5 to neutral tone 0), and is used to indicate a change of state. 

  • 12

    The term “stative verb” here basically corresponds to xìngzhì xíngróngcí “qualitative adjective” in previous works on Chinese linguistics such as Zhu (1982). The categorical status of adjectives in Chinese has been a controversial issue. Some linguists such as Chao (1968) and Li and Thompson (1981) suggest that adjectives in Chinese are a subtype of verbs, whereas McCawley (1992) argues that there is no adjective in Chinese. Recent works such as Zhang (1997) argue that some “qualitative adjectives” in Zhu (1982) should be grouped among the stative verbs, while others are indeed adjectives. 

  • 13

    For this study, we also conducted a cross-dialectal survey on the range of extended uses of ‘give’ constructions in a wide variety of Sinitic languages. With respect to Taiwan Southern Min, our native speaker consultants pointed out that elision of the third person singular pronoun i is also possible. This is at variance with unaccusative constructions in Hui’an and Jieyang Chaozhou varieties, where tone sandhi effects show clear evidence of i incorporation. It is also at variance with the Taiwan Southern Min data reported in Lin (2011). A possible explanation for this difference may be the strong influence from Mandarin, particularly among the younger generation, since elision of the post-gei (i.e., post-‘give’) third person pronoun ta is very common (and often preferred) in Mandarin. 

  • 14

    Unlike Lin (2011), who proposes that hoo7 in the hoo7 i1 pure unaccusative construction in Taiwan Southern Min serves as an adversative marker, this paper suggests that khɔ5-4 i1 as a whole, or khɔ5-1 (with i1 incorporated), functions as a marker of speaker affectedness. In our analysis here, the entire unaccusative khɔ5-4 i1 construction is pragmatically used to express the speaker’s subjective stance, and hence is speaker-based. 

  • 15

    Lin (2012) has also observed a similar development from referential to non-referential uses for ka i constructions in Taiwan Southern Min, i.e., in another type of transfer verb construction involving the disposal marker ka (otherwise referred to as an object marker). Chappell et al. (2011) noted that ka is derived from a transfer verb kang meaning ‘gather, share’, which then developed into a comitative marker ‘together with’. Based on a diachronic analysis spanning Old, Middle and Early Modern Chinese, Wong et al. (2014) also identified the early usage of kang/ka with the lexical verb meaning ‘to provide’ 

  • 16

    Both the Type 1 and Type 2 unaccusative ‘give’ constructions can be attested in the Fengshun variety of Hakka spoken in the County of Fengshun, Guangdong province. However, this is probably influenced by the Chaozhou variety of Southern Min, since (a) the unaccusative ‘give’ construction is quite limited in Meixian Hakka, the dominant Hakka variety; and (b) Fengshun County borders Jieyang and Chaozhou municipalities and is in fact situated within the Chaozhou-speaking area. The native speaker of Fengshun Hakka whom we consulted also points out that Fengshun Hakka is obviously influenced by Southern Min. 

  • 18

    We gratefully list the names of our native speaker informants in the acknowledgment section. 

  • 19

    See also Shibatani (1985). 

  • 20

    Cantonese, as well as other varieties of Chinese, does not make a formal (morphological) distinction between nominative and accusative forms (i.e., subject NPs and object NPs are realized by the same pronominal form keoi5 ‘he, him’). 

  • 21

    Unaccusative constructions necessarily involve intransitive predicates; passive constructions are derived from transitive predicates, which for discursive and pragmatic reasons have agents that are defocused or even elided, and particularly in the case of “agentless” passive constructions, the argument structure tends toward a valence-reduced interpretation that is somewhat similar to an unaccusative interpretation with its focus on a patient or theme NP. Nevertheless, a crucial distinction between the two is that, whereas unaccusative predicates ignore any reference to the agent, the passive construction still makes implicit reference to it. 

  • 22

    Some Chinese dialects (e.g., Wu) align with the Southern Min and Taiwanese Mandarin unaccusative ‘give’ constructions in terms of being more resultative-oriented and hence more passive-like in suppressing an agent reading, while some other Chinese dialects (e.g., Hui) align with the mainland Mandarin unaccusative ‘give’ constructions in being more causative-oriented. 

  • 25

    In Mandarin varieties, passive constructions could be expressed not only with passive marker gěi (derived from ‘give’), but also with other passive markers such as rang and bei (derived from ‘let’ and ‘suffer’ respectively). 

  • 26

    The yuan refers to the Chinese currency, and is also known as renmenbi. 

  • 27

    Although Malchukov did not specifically discuss these -v- constructions in terms of ‘give’ constructions, Igor Nedjalkov (1993) and others have suggested that the -bu-/-(v)u- causative and passive constructions in Manchu-Tungusic languages are derived from -bu- ‘give’ constructions. 

  • 28

    The term chǔzhìshì 處置式 ‘disposal construction’ is at the beginning used to refer to the ba 把/jiang 將-construction, which takes the form of ‘subject+ba/jiang+object+VP’, where the subject is followed by the function word ba/jiang which introduces the patient or direct object, which in turn is followed by a verb phrase expressing “disposal” of, or action upon, the object. 

  • 29

    For recent studies focusing on clause-medial constructions with subjective interpretations, see Chor (2010; 2013) and Yap and Chor (2014). 

  • 30

    For recent studies focusing on the development of sentence-final particles with subjective interpretations, see Yap et al. (2014). 

About the article

Published Online: 2018-01-05

Published in Print: 2018-01-26

Citation Information: Linguistics, Volume 56, Issue 1, Pages 19–68, ISSN (Online) 1613-396X, ISSN (Print) 0024-3949, DOI: https://doi.org/10.1515/ling-2017-0038.

Export Citation

© 2018 Walter de Gruyter GmbH, Berlin/Boston.Get Permission

Citing Articles

Here you can find all Crossref-listed publications in which this article is cited. If you would like to receive automatic email messages as soon as this article is cited in other publications, simply activate the “Citation Alert” on the top of this page.

Comments (0)

Please log in or register to comment.
Log in