Syntactic causatives, also called analytic causatives or periphrastic causatives, involve a causative marker which is a free form, typically a verb with the meaning ‘cause’, ‘make’, ‘let’, ‘give’, etc. (Kulikov, 2001). Portuguese syntactic causatives are formed with causative verbs (Vcaus) such as fazer ‘to do, to make’, mandar ‘to order, to make’, and deixar ‘to let, to make’. It has been observed that these causative verbs can produce causatives with four different forms: Vcaus + finite (1a), Vcaus + inflected infinite (1b), ECM/Raising-to-object type (1c), and fazer-Inf (1d) (see Barbosa & Raposo, 2013; Gonçalves, 1999a; Gonçalves, 1999b; Raposo, 1981; Santos, Gonçalves, & Hyams, 2016):

These causative verbs are selective in terms of the caused event. For example, a caused unaccusative may occur with fazer but not mandar (2). Nevertheless, the sentence with mandar in (2b) may be acceptable in contexts such as children’s tale where the stones are personified and can be ordered to fall.

Another group of instances that are included in our discussion is the simple resultatives (terminology from Mateu, 2012) formed by a light verb (e.g., fazer ‘to do, to make’, pôr ‘to put’ or tornar ‘to turn’) and an adjective/participle, as exemplified in (3). Contrasting with (1) and (2), which denote the causation of an action or activity, the simple resultatives express the causation of a change of state.

In Chinese, syntactic causatives generally are formed with the causative verbs 使shi, 让rang, and 令ling, among which 使shi has the generic causative meaning ‘to make’, 让rang has the meaning ‘to let, to make’ (similar to Portuguese deixar), and 令ling has the meaning ‘to order, to make’ (similar to Portuguese mandar). The Chinese syntactic causatives surface as NP1 + Vcaus + NP2 + VP, as illustrated in (4):

Besides 让rang, 令 ling and 使shi, the causative verbs occurring in Chinese syntactic causatives may also be 叫jiao ‘to call, to make’ or 害hai ‘to harm, to make’. The choice depends on the speaker’s presupposition, the intended style of speech, and the nature of the caused event (Huang, 1974). For instance, 使shi tends to causativize a psychological state (4), but not a physical action, as shown in (5).

The causative verbs 使shi /让rang /令ling may also embed an AP, expressing the causation of change of state, surfacing as NP₁ + Vcaus + NP₂ + AP, as shown in (6). However, when 使shi serves as the causative verb, this sentence sounds a bit less natural than that with 让rang or 令ling.

From the observations above, we can see that both Portuguese and Chinese exhibit causative verbs, which can create causative constructions productively. Semantic variations are observed among the causative verbs, which are language-specific.

In morphological causatives, causativity is encoded in a causative morpheme instead of a verb. Productive morphological causatives are attested in various languages, such as Turkish, Sanskrit, Ewenki, Georgian, Hindi, Japanese, and Finnish (see Comrie, 1976; Falk, 1991; Montrul, 2001; Pylkkänen, 2002; a.o.). Consider the examples in (7):

As Kulikov (2001) pointed out, Romance languages generally lack productive morphological causatives. Some affixes can form verbs with causative meanings in Portuguese, such as -ificar and -ear, as in puro ‘pure’ – purificar ‘to purify’ and branco ‘white’ – branquear ‘to whiten’, where the verbs denote causation of a change of state. However, no causative morpheme can productively derive causatives by attaching to a base root in Portuguese, contrasting with languages such as Japanese and Turkish. Although many Portuguese verbs with -ar may involve causative meanings, such as those in (8), Matos (1999) points out that not all verbs with -ar involve causativity (9) and suggests that it is not the suffix but the semantic properties of the base roots that drive the causative interpretation in (8).

Morphological causatives existed in Old Chinese but underwent a fade-away during language development. In Old Chinese, causative meanings could be expressed through phonological or morphological changes such as tonal change and affixation (see Shi, 2002; Xu, 2006; Mei, 2012). For example, in (10), the prefix *s- alters the verb’s meaning from ‘to eat’ to ‘to feed’, bringing about the causative interpretation.

This morphological/phonological mechanism to create causatives faded away gradually in the history of Chinese, and only relics can be found in Modern Chinese. The morpheme -化hua might be the only existing morpheme with causative function in Modern Chinese, similar to the Portuguese morpheme -ificar (e.g., puro ‘pure’ – purificar ‘to purify’). As illustrated in (11), when -化hua attaches to the base root 美mei ‘beautiful, beauty’, the causative meaning ‘to beautify’ is produced. Other combinations include 绿化lü-hua ‘green-HUA: to make green’, 现代化xiandai-hua ‘modern-HUA: to modernize’ and 国际化guoji-hua ‘international-HUA: to internationalize’. In recent years, the causative formation with -化hua has become highly productive. One can literally attach -化hua to any nominal or adjective to produce causative meanings.

Based on the observations above, we can see that both Portuguese and Chinese exhibit some morphemes with causative meanings. However, it is worth noting that these morphemes can only apply to a non-agentive root to express causation of change of state (e.g., purificar ‘to purify’ and 美化mei-hua ‘to beautify’). They can never express the causation of an activity. That contrasts with languages such as Finnish and Turkish, where causation of an agentive event can also be expressed through morphological means (see Comrie, 1976; Montrul, 2001; Pylkkänen, 2002; a.o.).

Nevertheless, verbs that intrinsically express the causation of change of state seem to be more available in Portuguese than in Chinese. Many Portuguese verbs that contain a morphological constituent plus a state base, such as sujar ‘to stain’ and limpar ‘to clean’, do not have Chinese equivalents in simplex verb forms. For example, the Chinese counterpart of sujar ‘to stain’ takes a complex form, as shown in (12b, c), where the verbal component that denotes the causing activity (V1) precedes the one encoding the result (V2). Since the two Vs are in adjacency, we will call them Causative Resultative V-Vs (CR V-Vs) throughout this study.4 In the literature, other names used to refer to this construction include “resultative V-V compounds” (e.g., Li, 1990), “resultative compounds” (e.g., Huang, 1984; Cheng & Huang, 1994; Basciano, 2010), and “resultative verb compounds” (e.g., Thompson, 1973; Lu, 1977; Gao, 1997; Li, 2007).

Note that we do not categorize CR V-Vs as syntactic causatives (see Section 2.1) for a few reasons. First of all, although the V1 弄nong in (12b) denotes a generic meaning of ‘to do, to make’, it is not an alternative to the causative verbs found in the syntactic causatives presented in Section 2.1. In addition, the V1 in CR V-Vs may be a verb that refers to a concrete and specific activity, as in (12c), and there can be a non-exhaustive list of possible V1s. In other words, the causative meanings in CR V-Vs are not produced by any causative verb (Vcaus), contrasting to the syntactic causatives. When 弄nong ‘do, make’ serves as V1 in CR V-Vs, as in (12b), the causing activity is simply understood as “doing something”.5 A further difference is that CR V-Vs exhibit compound features and show tight integrity. A more detailed description of CR V-Vs will be provided in Section 2.4.

The Portuguese verb limpar ‘to clean’ also corresponds to Chinese expressions in the CR V-V form, where V1 may have a generic (13b) or specific (13c) meaning.

One may argue that the meaning ‘to clean’ can be expressed by a single verb in Chinese, namely 打扫dasao. However, note that this verb only denotes the activity of ‘doing cleaning’ without implying a result. For this reason, the semantic meaning of (14b) is not equivalent to the Portuguese sentence with limpar ‘to clean’ in (14a). To include the result meaning that ‘the room became clean’, a CR V-V form should be employed with the occurrence of an overt result component, as in (14c).

In contrast to morphological causatives, lexical causatives lack any regular and productive causative marker (Kulikov, 2001), such as brown, thin, smooth, dry, and narrow when used as verbs.

Lexical causatives may be subdivided into a suppletive type and a labile type (see Comrie, 1989; Kulikov, 2001): the suppletive subtype consists of verbs with causative meanings and are in a suppletive relation with their non-causative counterparts, e.g., kill – die; the labile subtype includes the lexical causatives which are not formally distinguishable from their non-causative counterparts, including but not limited to the so-called alternating verbs (i.e., verbs that allow causative/transitive alternation) such as break and open.

Both Portuguese and Chinese have suppletive verb pairs. They come to have a similar group of meanings (e.g., ‘kill’ – ‘die’, ‘teach’ – ‘learn’) for metalinguistic reasons (since people’s conceptualization and experience are generally the same). However, semantic differences are observed. Taking the pair ‘kill’ – ‘die’ as an example, the meaning of the Portuguese verb matar ‘to kill’ assumes a completed caused event denoted by morrer ‘to die’. As shown in (15a), the refutation is not allowed because it contradicts the first part of the sentence (i.e., ‘the girl was killed and thus died’). In contrast, the Chinese verb 杀sha ‘to kill’ does not necessarily assume the completion of 死si ‘to die’; therefore, refutation is possible (15b) (see Tai, 1984; Sybesma, 1997; a. o.).6 The Chinese equivalence of Portuguese matar ‘to kill’ should take a complex form, namely a CR V-V, as illustrated in (15c), where V2 explicitly expresses the result. As expected, this sentence does not allow refutation.

Verbs that allow causative alternation are attested in Portuguese, but it is quite common that the unaccusative clitic -se is used in the intransitive structure, as illustrated in (16).

However, there also exist verbs that do not require or disallow the occurrence of the clitic -se in the intransitive counterpart, as illustrated in (17)–(18) (see Matos, 1999).

The clitic -se here is usually treated as an anticausative marker, which permits an anticausative process, deriving intransitive configurations from a transitive verb. With regard to (17) and (18), where the occurrence of this clitic is not required or even disallowed, we follow Matos (1999) and assume that the anticausative morpheme is not phonetically realized, being null. In the case of non-occurrence of any anticausative marker, they can be considered labile lexical causatives (under the assumption that the notion of labile lexical causatives refers to verbs’ superficial forms).

The labile lexical causatives were very common in Old Chinese but are much less common in Modern Chinese. In Old Chinese, they could be formed from adjectives, nouns/nominals, or verbs – they are also called “zero causatives” since no overt causative marker is involved. As illustrated in (19), 树shu, a noun with the meaning of ‘tree’, turns to function as a transitive verb when preceding 之zhi ‘it’, producing the causative meaning ‘to make it have trees, to plant’. In (20), 厚hou is an adjective with the meaning ‘thick’; when preceding the NP 墙垣qiangyuan ‘wall’, it comes to have the causative meaning ‘to make the wall thick, to thicken’. Similarly, in (21), the intransitive verbs 进jin ‘go forward’ and 退tui ‘withdraw’ obtain causativity when preceding 之zhi ‘him’, producing the meanings ‘make him go forward’ and ‘make him withdraw’. The three examples are found in Xu (2006, p. 118), Shen and Huang (2017), and Xu (1998), respectively.

It has been observed that Chinese shifted from a fairly synthetic language to an analytic language (Huang, 2006, p. 25), and the causatives also underwent this shift: the morphological causatives and labile lexical causatives gradually decreased (see Xu, 2006; Shen & Huang, 2017). However, some instances of zero causatives can still be found in Modern Chinese, as shown in (22).

In Modern Chinese, few simplex verbs allow causative alternation. The verbs 开kai ‘open’ (23) and 关guan ‘close, turn off’ (24) are among the very few ones.

Consequently, the transitive and intransitive use of Portuguese alternating verbs may correspond to different verbal expressions in Chinese. For example, the Portuguese verb partir ‘to break’ can be used transitively and intransitively, as shown in (16a, c) above; in contrast, the Chinese verb 碎sui ‘to break’ is pure intransitive (25a), as shown by the ungrammaticality of the attempted transitive use in (25b). In this case, CR V-Vs should be employed to express causative meanings, with a verb denoting the causing activity to precede 碎sui ‘break’, as shown in (25c, d).

In the previous sections, it has been observed that quite many Portuguese verbs that intrinsically denote causative and resultative meanings, either as morphological or lexical causatives, correspond to a compound type of causatives in Chinese, namely the CR V-Vs. We will dedicate this section to a brief presentation of this Chinese construction.

The Chinese CR V-V has attracted a lot of attention in the past decades due to its complex and interesting properties (see Thompson, 1973; Lu, 1977; Huang, 1984, 1992; Li, 1990, 1995; Cheng & Huang, 1994; Gao, 1997; Sybesma, 1999; Lin, 2004; Her, 2007; Li, 2007; Basciano, 2010; Fan, 2013; Liu, 2019). First, they show verb-like properties and can further undergo passivization, as shown in (26a), which presents a passivized form of (25d). Each CR V-V is read as one intonational unit, and the two Vs cannot be intervened by an NP, an adverbial modifier or an aspect marker (26b). Despite these compound properties, CR V-Vs also show syntactic features and are highly productive. For example, there is a non-exhaustive list of possible V1s to form CR V-Vs with 碎sui ‘break’ (26c).

Semantically, CR V-Vs denote the meaning of a change of state, and thus the component at the V2 position is subject to semantic constraints. For instance, unergative verbs cannot occur at the V2 position. As shown in (27), while the verb 跑pao has both an unergative (i.e., ‘to run’) and an unaccusative use (i.e., ‘to escape’), only the unaccusative one is acceptable at the V2 position.

Moreover, some CR V-Vs may show semantic ambiguity, allowing two or more different readings (28).

From a diachronic perspective, it has been claimed that CR V-Vs only occurred at a later stage of Chinese language development, accompanied by the gradual decline of morphological and zero causatives (see Shi, 2002; Xu, 2006; a. o.). Various authors (e.g., Shi, 2002; Xu, 2006) suggest that the rise of CR V-Vs was driven by a disyllabification tendency of words that occurred in the development of Chinese. As reviewed by Shi (2002, pp. 71–72), before 200 BC, Chinese words were primarily monosyllabic, and disyllabic words represented only approximately 20% of the Chinese lexicon, but in Modern Chinese, more than 80% of the words are disyllabic. In Old Chinese, verbs were mainly monosyllabic, and some could intrinsically express a caused-result event. Later, due to the disyllabification tendency, two monosyllabic verbs adjoined each other to form a unit to express caused-result meanings, resulting in a V-V surface, with V1 overtly denoting the causing event and V2 the result.

Due to the complexity of CR V-V’s properties, authors hold different opinions regarding the nature of this construction (for example, whether they are formed in the lexicon or syntax), and various proposals can be found in the literature. We claim that CR V-Vs are syntactically derived compounds, whose syntactic structures will be presented in Section 3 of this study.

There is a group of transitive verbs in Portuguese that can derive participles (or deverbal adjectives), such as avariar ‘to damage’ – avariado ‘damaged’, cansar ‘to make tired’ – cansado ‘tired’, and inclinar ‘to incline’ – inclinado ‘inclined’ (see Duarte & Oliveira, 2010 for discussions on Portuguese participles). These pairs can be considered analogous to the causative-anticausative pairs presented in Section 2.3 since the causative member is in a basic form, and the non-causative counterpart (e.g., the participle) is further derived. Interestingly, these formally simplex but semantically complex (i.e., involving causative and resultative meanings) verbs may not have Chinese counterparts in simplex forms. For example, the meaning ‘to damage’ (denoted by the verb avariar in Portuguese) is expressed by CR V-Vs in Chinese, as in (29b), where the cause-denoting V1 precedes the result-denoting V2 坏huai ‘damaged’.

In (29b), the first expression 弄坏nong-huai ‘make-damaged: to damage’ is considered a CR V-V in our definition. The latter two have lexicalized and are expected to be listed in dictionaries. Regardless of the lexicalization process, all these three are analyzable verb compounds, with V1 denoting the causing eventuality and V2 the result. There is a non-exhaustive list of possible combinations of V1 + huai ‘damaged’ to produce the meaning of ‘to make something damaged through the activity denoted by V1’.

Furthermore, some of the Portuguese verbs that intrinsically involve caused-result meanings may correspond to syntactic causatives in Chinese. For example, no Chinese verb corresponds to the Portuguese verb interessar ‘to interest’, which involves a causative meaning (‘to make someone interested’) and can derive a participle (interessado ‘interested’). As illustrated in (30), the Chinese counterpart employs syntactic means (via the Vcaus 让rang ‘to let, to make’) to causativize the result-denoting expression 感兴趣ganxingqu ‘interested’.

Based on the observations in the previous sections, the causative forms that are applied to produce causation of activity and causation of change of state in Portuguese and Chinese are summarized in Table 1.

As shown in Table 1, both Portuguese and Chinese exhibit syntactic means to express the causation of an activity. However, these two languages exhibit quite a significant variation in the causation of change of state. With syntactic means available in both languages, Portuguese exhibits plenty of verbs (including morphological and lexical causatives) that intrinsically involve caused-result meanings, whereas the Chinese correspondents may appear in complex forms, such as compound causatives. In the next section, we will focus on the two languages’ differences in expressing the causation of change of state and attempt to provide an explanation.

In this study, I assume the general idea of Distributed Morphology (Halle & Marantz, 1993, 1994; Marantz, 1997) that the syntax is the only generative system responsible for both word structure and phrase structure and that the “Narrow Lexicon” consists of “lexical roots” and “bundles of grammatical features” (functional elements): the roots are acategorial, and their function is to provide encyclopedic meanings; they can only get categorized and interpreted by merging with a categorizing functional head little x, such as v, n or a, which can categorize a lexical root as a verb, noun or adjective respectively, as illustrated in (31).

In particular, the core structure of a verb phrase contains a little v head and a root, as illustrated in (31a): the little v semantically introduces an eventuality, and the root modifies this event by contributing semantic content. According to Folli and Harley (2005, 2007), the different nature of events (e.g., causative, unaccusative, stative, unergative) is not determined by the semantic information contained in the lexical roots but rather by the different “flavors” of v heads that contain specific event-semantic content, such as v_CAUSE, v_BECOME, v_DO. In particular, v_DO is an agentive v, which requires an animate Agent subject; v_DO can take a straightforward Incremental Theme as its Complement and is a true verb of creation, as exemplified in (32a). In contrast, v_CAUSE is a causative v, which only requires that the subject be a possible Cause; v_CAUSE takes a state as its Complement, creating a resultative structure essentially, as in (32b).

As presented in Section 2, Portuguese exhibits many verbs that express complex meanings, involving causation and result, such as matar ‘to kill’, sujar ‘to stain’, and avariar ‘to damage’. Their Chinese correspondents often take complex forms, usually CR V-Vs, and sometimes syntactic causatives. We posit that the semantic meanings of roots play a role here.

Chinese simplex verbs generally tend to express simplex meanings, either a pure activity or a (change of) state, but rarely both (see Tai, 1984; Sybesma, 1997; a. o.).11 We have already shown that the Chinese verb 杀sha ‘to kill’ only denotes the action of ‘killing’ without implying a result (15b); 碎sui ‘to break’ only denotes the inchoative meaning ‘become broken’ without involving causation (25a, b). In Chinese, we can find pairs of verbs showing a division of labor, with one member denoting a pure activity and the other the corresponding result (or change of state). One example is ‘to boil’. The corresponding verbs in English (boil) and Portuguese (ferver ‘to boil’) can denote both the causing activity (e.g., ‘someone boils water’) and the caused result (e.g., ‘the water is boiling/boiled’). However, in Chinese, the causative and resultative meanings are expressed separately by a pair of verbs, as illustrated in (33a). The meaning ‘he boiled the water’ with the result ‘the water boiled’ implied should be expressed by a CR V-V in Chinese, with the activity-denoting member preceding the result-denoting member, as in (33b).12

Only a few Chinese verbs may express both an activity and a result. They are typically the verbs that allow causative alternation (see Section 2.3), such as 开kai ‘to open’. They not only can convey the meaning of causation of change of state but may also occur in CR V-Vs as one component. Since they can express both an activity and a result, the same verb may simultaneously serve as V1 and V2, showing the division of labor. In (34), 开kai in the V1 position denotes the pure activity of ‘to open the door’, and the one in the V2 position indicates the result ‘the door opened.’

I attribute the contrast to the roots’ different semantics in these two languages. In general, Portuguese roots may be intrinsically complex in their semantics, involving both causative and resultative meanings. In contrast, Chinese verb roots tend to express only an activity or a (change of) state. Under the assumption that the semantic meaning of a root determines what head(s) it can be merged with, we can infer that the semantic differences between Portuguese and Chinese roots may impact the possible structures that the roots can fit in. For instance, the encyclopedic meaning of the Portuguese root √partir ‘break’ is agentive/causative, and therefore the root can merge with v_DO. Contrastingly, the meaning of the Chinese root √sui ‘break, broken’ is inchoative or stative, and therefore, this root can only be compatible with v_BECOME, v_BE, or a but not v_DO.14 As shown in (35a), by merging with v_DO, the Portuguese root √partir ‘break’ is turned into a verb with causative meanings (with complex event structure). However, such a structure is incompatible with the Chinese root √sui ‘break’. Instead, this Chinese root can merge with v_BECOME to express a change of state event without causation (with a simplex event structure). We would like to highlight that the Portuguese root √partir cannot express stative meanings intrinsically, as shown by the unacceptable attempted adjectives formed with √partir merged with an a head, *parto/a ‘broken’. The stative counterpart requires further derivation departing from [v_DO √partir], resulting as a participle (partido/a ‘broken’), which will be discussed later in Section 3.3.2.

Moreover, as shown in (35b), although the Portuguese root √matar ‘kill’ and the Chinese root √sha ‘kill’ are both intrinsically agentive and can merge with v_DO, the generated verbs express different semantic meanings (with different event structures). It is because, after the merge of a root and a head, the expressed semantic meanings are produced by the joint force of the encyclopedic meaning of the root and the eventuality introduced by the head. In other words, merging with the same head does not guarantee a similar event structure – the meaning of the root also plays an important role. If a root is semantically complex intrinsically, it comes to express a complex event when turned into a verb by merging with v_DO – this is the case with Portuguese matar ‘kill’, which contains both causative and resultative meanings. Contrastingly, the Chinese root √sha ‘kill’ denotes a simplex meaning, and by merging with v_DO, it comes to express a pure action. Therefore, the difference between the Portuguese and the Chinese ‘kill’ in (35b) does not originate from the structure but the semantics of the roots.

To sum up, Chinese verb roots tend to express only an action or a (change of) state. In the former case, it can merge with a v_DO, and the resulted verb conveys a pure activity without necessarily involving a result; in the latter case, it may merge with v_BE or a to express a state or a v_BECOME to expresses a change of state without causation. In contrast, Portuguese verb roots may contain complex semantic meanings intrinsically, and their corresponding verbs may express complex events involving both the causation and the result. For this reason, many Portuguese verbs that denote caused-result meanings do not find Chinese correspondents in simplex forms. They often correspond to syntactic causatives and CR V-Vs in Chinese. A set of examples are presented in the Appendix.

In fact, in Old Chinese, the caused-result meanings such as ‘to stain’ and ‘to cut’ could be expressed by simplex verbs, as shown in (36). During that time, the CR V-Vs did not exist yet – they only occurred at a later stage of language development (see Shi, 2002; Xu, 2006). This further confirms the correlation of the semantic depository of the verb roots with the complexity of the employed expressions.

In this section, we will show that in expressing change-of-state events, the causative and non-causative pairs may be derived in opposite directions in Portuguese and Chinese: in many cases, while the non-causative member is derived from the causative member in Portuguese, the causative member is derived from the non-causative member in Chinese.

The observed differences in expressing change-of-state events between Portuguese and Chinese are mainly due to the available semantics of the roots. Portuguese exhibits many result roots, which are intrinsically complex in semantics and can produce causative meanings when turned into verbs. However, these meanings may have to be expressed by complex forms in Chinese (e.g., CR V-Vs or syntactic causatives) because such result roots are not available. There are some apparent result roots such as √sha ‘kill’, but in fact, the result meaning is not necessarily implied. In general, Chinese verb roots tend to express a pure activity or a (change of) state without causation. On the other hand, while Chinese exhibits plenty of simple stative forms, they may correspond to complex forms in Portuguese, namely the deverbal adjectives or participles, because such property concept roots are not available in Portuguese.

Therefore, comparing the causative and non-causative expressions of change-of-state events, we observe opposite derivational directions in the two languages, as illustrated in Figure 1. Portuguese verbs can express complex meanings (i.e., caused-result meanings) and may undergo an anticausative process to derive non-causative counterparts. In contrast, Chinese verbs primarily denote simplex meanings, either a pure activity or a pure (change of) state. A causative process would be employed to produce causative meanings.

However, we are not claiming that this general trend applies to every change-of-state event. Both Portuguese and Chinese exhibit syntactic causatives (see Section 2.1), which clearly involve causative processes; moreover, the equipollent alternation is attested in both languages.