Is Chinese Run-on Sentence an Exception to the Iconicity of Sequence?

As one of the most active scholars in the field of Chinese run-on sentence (CRS for short), Wang and Zhao (2016) keenly realize that CRS displays distinctive traits of spatiality, namely, chunkiness, discreteness and reversibility, among which, the last trait and the iconicity of sequence/order (e.g. Haiman, 1984, 1985) seem to depict a diametrically opposite picture. In the present article, there would be an attempt to undertake an investigation of Wang, Zhao et al.’s ‘reversibility’ to see whether or not CRS is an exception to the iconicity of sequence/order. The main arguments are as follows. First, ‘reversibility’ is borne out to be local by some linguistic facts, especially in: (i) duyuju within CRS; and (ii) shuncheng CRS. Second, although the ‘reversibility’ sometimes exhibits a tendency to change the positions of clauses/syntagms in CRS, there is a clear correlation between the clause order of CRS and iconicity. The sequence/order principle in practice emerges as a cognitive mechanism emitting some effects in the clause order of CRS. Third, the Reversibility Condition is required to come into being so as to arrive at a detailed specification of the applicable scope of the ‘reversibility’. And finally, it is more preferable to ameliorate the spatiality of CRS as two traits, that is, chunkiness and discreteness.


Introduction
Chinese Run-on Sentence (CRS hereafter), being ruled out in (written) English grammar, appears to be a kind of unique phenomenon in Mandarin Chinese. This term is firstly thrust into the limelight by Lü (1979:27), who defines it as a kind of sentence where clauses follow one after another, and in many places they can be connected or simply cut. Meanwhile, in Lü's (1979:22) deliberation, CRS is overwhelmingly frequent in spoken Chinese (see also in Lian, 1992;Shen, 2012). The two eye-catching viewpoints of Lü (1979) mark the commence of CRS's detailed researches over the ensuing decades. The working definition of CRS, for instance, has experienced several upgrades (Gao, 1988;Wu & Liang, 1992;Yuan, 2000;Zhang, 2000;Guo, 2004;Shen, 2004;Shen, 2012;Zhang, 2015;Sheng, 2016;Wang & Zhao, 2016;Chen & Duan, 2020), which can be encapsulated as follows: Chinese sentences, different from the Indo-European language family, "usually take a chronicle style" (Lian, 1993:67), and CRS is always treated as "a mirror of some typical features of Chinese" (Wu & Liang, 1992:316). Since connective words are not absolute, CRS is often a composition of a cluster of juxtaposed phrasal and clausal expressions that share a peculiarity of "referentiality" (e.g. Shen, 2012). The co-occurrence of phrasal syntagms and clausal syntagms, the loose structure, and "the blurry line between main and subordinate clauses as well as between subjects and predicates" (Lian, 1992:4;Kong, 1997:283) are all results of the absence of explicit markings and structural representations of such running sentence. Consider the sentence: (1) a. ①走，②不早了，③只有二十五分钟，④叫他们把车子开出来，⑤走吧。 (曹禺《雷雨》 ) c. 'Come on. We haven't got much time. Only twenty-five minutes before the train goes. Tell them to get the car out. Come on'.
The sentence (1) contains five independent syntagms with no explicit connectives, rendering a loose structure. As well as being unique, all syntagms in such diffuse construction are shown as having no full subject or predicate. Namely, the dominant ingredient of the CRS in (1) is proved to be an incomplete subject-predicate construction, or namely, minor sentence 1 (Chao, 1968;Shen, 2012). Furthermore, note that the subjects of some syntagms like ①, ③, ④ and ⑤ are not clearly specified. It means that the ambiguity of subject reference is borne out to be a "habitual frequenter" in CRS.
In central papers such as Hu & Jin (1989), Wu & Liang (1992) and Shen (2012), much efforts to provide an explicit characterization of CRS's structural properties are already underway. More recently, some illustrious works with regard to CRS tend to be spatiality-oriented in the footstep of Wang and Zhao (2016), who are keenly aware of the fact that the construction of CCS is marked by three spatial traits, namely, chunkiness, discreteness and reversibility, among which, 'reversibility' underlies that clauses/syntagms in a sentence is organized in a such flexible way that they can be moved to different positions, and even reversed in their positions (e.g. Wang & Zhao, 2016;Wang & Liu, 2021). However, the view of this kind of reversed clause order seems to display an opposite behavior with the "iconicity of sequence/order" (e.g. Haiman, 1980Haiman, , 1984Haiman, , 1985. Therefore, on the basis of acknowledging Wang et al.'s discreteness and chunkiness, the purpose of this study is to investigate the rationale of 'reversibility' so as to see whether or not CRS is an exception to the iconicity of sequence/order. For this examination, the present author will firstly zero in on some contexts where the use of 'reversibility' is ruled out. Then, pages are devoted to further discuss the sequential order principles on CRS and provide some constraints on the appearance of 'reversibility' inside CRS.
In the same vein, based upon an inventive viewpoint that Chinese is a spatiality-dominant language, Wang and Zhao (2016) keenly realize that CRS, "as a mirror of Mandarin Chinese (Wu & Liang, 1992:316)", also turns out to be spatiality-oriented. Therefore, Wang, Zhao et al. (e.g. Wang & Zhao, 2016, 2017a, 2017b, 2017c, 2020Cui, 2017;Cui & Wang, 2019;Wang & Liu, 2021), with zeal and genius, pursue the cognizance that the construction of CRS is marked by chunkiness, discreteness and reversibility, among which the 'chunkiness' is in essence about the same with 'juxtaposition 2 ' given by Shen (e.g. 2012). In general, they share a firm belief that a stream of syntagms (/full sentences and minor sentences), in default of explicit associative words, are loosely strung together, or juxtaposed, giving rise to the second spatial trait, 'discreteness'. This property, together with the 'fragmentability 3 ', is reserved to describe the absence of cohesive devices such as connectives among different linguistic units (Wang & Liu, 2021:5), leading to the co-occurrence of phrasal and clausal syntagms, the loose structure, and ambiguity of subject reference (Wang & Zhao, 2016, 2017a, 2017c, 2020Cui, 2017;Wang & Liu, 2021). Something to also note is that it is difficult to pinpoint the full or null subjects among syntagms.
Furthermore, with the mutual effect of 'chunkiness' and 'discreteness', the 'reversibility' of CRS comes in (Wang & Zhao, 2016:18-19). In Wang and Zhao's (e.g. 2016) shrewd cognizance of 'reversibility', syntagms (/clauses and phrasal expressions) in CRS are organized in such a flexible way that they can be moved to different positions, and even reversed in their positions, without any shift in meaning. This is hinted in example (2) from Wang and Zhao (2016:20).
1 According to Chao (1968:83), sentences can be classified into full and minor sentences. The former consists of two parts, a subject and a predicate, while the latter usually is not in the subject-predicate form and occurs more often in oral speech, such as commands, vocatives, responses and exclamations. Moreover, Chao (1968:83) points out that, compared to the complete construction of a full sentence, most minor sentences are either verbal expressions or nominal expressions. 2 In Shen's (e.g. 2012Shen's (e.g. , 2017Shen's (e.g. , 2019 a series of landmark publications, 'juxtaposition', as a defining property of CRS, often depicts a fact that CRS is a string of 'minor sentences' (Chao, 1968) and full sentences without the presence of obvious connectives. 3 In the words of Wang and Liu (2021:5), 'fragmentability' is closely germane to the 'discreteness'. Basically, it, due to the lack of explicit lingkages, between different clauses, refers to the greater flexibility in the organization of linguistic units.
( What is worthy of special mention is that there is no bright-line distinction between the three spatial traits but an inclination to interact as mutual cause and effect (Wang & Zhao, 2016:18). Put another way, in Wang and Zhao's (2016) notes, in practice, the 'chunkiness', 'discreteness' and 'reversibility' turn out to be inseparably interwoven.

Iconicity of Sequence/Order
Currently, with a significant boost from functional and cognitive linguists, the notion of "iconicity" has prevailed over the "arbitrariness" that, despite overwhelming evidence to the contrary, traditionally is a seemingly immutable axiom. Opposed to the "arbitrariness" invoked by Saussure in his illustrious work Course in General Linguistics (1916), the launch of Jakobson's paper Quest for the Essence of Language (1965Language ( , 1971) appears to be a prelude to promising research prospect of the iconicity account. By way of eye-catching mention of the strong affinity between Peirce's semiotic concepts and research on grammar, this groundbreaking essay presents the first substantial challenge to the conventional dominance of "arbitrariness".
Peircean iconicity later is fervently advocated by Haiman (1980) who reduces Peircean typology to two varieties, that is, imagic iconicity and diagrammatic iconicity. The notion of diagrammatic iconicity, which is concerned with the structural/relational similarities between the sign and the referent, has been used in various functional and cognitive explanation of linguistic structure (cf. Jakobson, 1965Jakobson, , 1971Plank, 1979;Haiman, 1980Haiman, , 1984Haiman, , 1985Givón, 1985Givón, , 1991Taylor, 2002;Haspelmath, 2008). In this paper, I only concentrate on a particular subtype of diagrammatical iconicity, that is, iconicity of sequence/order. In Jakobson's reservoir of works, the allegedly iconicity of sequence/order can be explained by a chain of verbs "Veni, vidi, vici (I came, I saw, I conquered) (1971:350)", which inform us of that the word order of component clauses is employed to reproduce the sequence of a succession of reported events. This correspondence in sequence also finds its place in Greenberg's (1966:103) paper. He gives a relatively sophisticated account that "the order of elements in language parallels that in physical experience or that of knowledge".
A somewhat hackneyed but quintessential example from Greenberg (1963Greenberg ( , 1966 is the all-but-universal precedence of subjects over objects, such that OSV, OVS, and VOS languages are a very rare occurrence. This dominant order in practice matches the sequence of experiences (i.e., it follows the order of AGENT-ACTION-PATIENT). Based upon such order, there is an unbridgeable gulf between SVO languages, such as English and Chinese, and SOV languages (or OV in spoken languages) like Japanese, giving effable evidence for the fact that "the degree of syntactic iconicity varies with languages" (e.g. Shen, 1993:4). To be exact, the SVO languages are found to be particularly prone to signal a higher degree of iconicity. Among scholars who bear out this averment ardently, a significant place belongs to Tai (1985Tai ( , 1989, who reiterates that Chinese indicates a higher degree of iconicity than English. And, grounded on the evidences from contrastive studies, Chinse, English and Japanese make a concerted effort to show a gradual decline in the degree of iconicity (Wang, 1999:6).
Assiduously pursued by Givón (1991:92-93), two separate principles of natural sequential order used extensively in syntax are laid out.
(3) Semantic principle of linear order: "The sequence of clauses in coherent discourse will tend to correspond to the temporal order of the occurrence of the depicted events".
(4) Pragmatic principle of linear order: a. "More important or more urgent information tends to be placed first in the string".
b. "Less accessible or less predictable information tends to be placed first in the string".
The use of principle (3) is firstly reflected in the fact that the sequence of clauses is iconically motivated by the temporal sequence of events depicted in a connected discourse. For the principle (3), I have in mind such contrasting pairs (5a) and (5d) as follows: (5)  Another instance being often singled out is a near universality of the precedence of cause clauses over their paired effect clauses, and of condition clauses over their paired entailment clauses. Thus consider (Givón, 1991:92): (6) Frequent: She shot him, and he died.
Infrequent: He died. She had shot her.

Infrequent:
He died after she shot him.
The principle (4), in Givón's (1991:94) sentiments, can be condensed into an elegant statement that "unpredictable, less accessible, surprising information is likely to be more urgent than predictable, accessible information". In this sense, the string-initial or clause-initial positions tend to be reserved for less frequent, marked and new important information exposed by referents, that is, the "figure" (e.g. Langacker, 1991;Ungerer & Schmid, 1996;Talmy, 2000), while the predictable and accessible information is always seen as the "ground" (e.g. Langacker, 1991;Ungerer & Schmid, 1996;Talmy, 2000). Such "word order denoting the topicality of referents (Givón, 1991:93)" in terms of importance and accessibility can be detectable by the following topic constituents: (8) a. Neutral: Chelsea really likes coffee.
b. Cleft-focused topic: It's coffee that Chelsea really likes.
c. Dangling topic: Coffee, Chelsea really likes it.
d. Contrastive topic: Chelsea really likes tea, but coffee she does not like.

Further Discussion of Reversibility
In this part, I will start with two cases, duyuju in CRS and shuncheng CRS, where 'reversibility' is instantiated to be ruled out. Then, pages will be devoted to a discussion of a cognitive mechanism, namely, the sequence/order principle, governing the clause order of CRS. Lastly, the present article will seek to lift the veil of the applicable scope of 'reversibility'.

Duyuju in CRS
Apropos of duyuju, the nominal independent sentence, Wang (2018) holds that it is a kind of non-subject-predicate structure in which nominal words or phrases are juxtaposed without an explicit grammatical connection and a verbal construct. In Wang's (2018) perspicacious attempt at a classification of duyuju, a noun or noun phrase can function as an independent sentence, a syntagm in CRS, or a juxtaposition of nouns or noun phrases separated by commas, as illustrated in example (9-11) below.
(9) a. 笑声。同班同学的笑声。①天真无邪的笑声。②烂漫友善的笑声。 a-CL attractive picture a-CL fascinating flowing river c. 'With the clear water, a blue sky, a bluish green mountain peak and a green meadow, the Lijiang River presents an attractive picture and a fascinating flowing river.'.

d. ②蓝天，④绿原，①碧水，③青峰。呵，⑤漓江，⑥一条醉人的江，⑦一副迷人的画。
In example (9), a stream of nouns and nominal phrases, being independent, come together in a row, and then a "choppy sentence 1 " (Shen, 2012) generates. And (9d) shows that trading the place of syntagm ① for the spot of ② is allowed since there is a pure coordination between the two syntagms. Somewhat differently, a series of noun phrases ① to ④ make up a CRS in example (11) where the syntagms (i.e. noun phrases), due to the coordinate relations among them, can be moved to different positions. Herein, my concern is not to discuss these two types of duyuju, but rather to dig deeper into the second type, namely, duyujus as syntagms in CRS.
The picture in example (10) where the underlined duyujus (/nominal independent sentences) serve as syntagms in CRS, is quite opposite to (9) and (11). To be specific, for one thing, in (10a), it is easy to determine that two nominal phrases are put side by side with no connective words to spell out their relationships. Wang and Liu (2021) explain that, owing to the preference for spatiality and the pursuit of aesthetic expressions, the linkage between the elements tends to be optional and redundant. For another thing, in Wang and Liu's (2021) deliberations, the juxtaposition of two nominal expressions in example (10a) is borne out to be closely germane to, and even equivalent to, the paratactic juxtaposition, that is, two images juxtaposed in spatial relations. It means that Wang, Zhao et al.'s (e.g. 2016) 'chunkiness' (/Shen's 'juxtaposition') embodies two layers of juxtaposition in CRS, that is, a juxtaposition of syntagms in structure and a juxtaposition of images in spatial relations. And the former juxtaposition always hinges on the latter. To be proper, in the underlined part in (10a), the sentence builder firstly presents the image of a single-plank bridge, and then zooms in to focus on the tavern that is the place the subject(s) might head for. A picture can be vividly perceived: nearby there is a single-plank bridge, and by the bridge stands a small tavern. Therefore, following the sequence of visual representation in (10a), the order between syntagm ① and ② cannot be exchanged like example (10d).
Furthermore, by the same logic, the example (12a) below involving dujuyus as syntagms would be unacceptable if the series of nominal syntagms from ① to ⑦ are rearranged in a different or reverse order, as (12d) showcased. The unacceptability of (12d) comes down to the fact that the order of syntagm ① to syntagm ⑦ are required to correspond to the temporal order of depicted occurrences. It means that the syntagm ① to syntagm ⑦ turns out to be placed in a temporal order. Given the specific context, some audiences died after several gunshots, therefore the syntagms ① and ② fail to be transposed. The same is true for syntagms ⑤, ⑥ and ⑦, which jointly present a fast-moving succession of events. Perhaps, there is a distinct possibility that a sentence builder might exchange the position of syntagms ⑥ and ⑦ since a coordinate relation can be tracked down between them. However, albeit a remote chance, moving the syntagms ⑥ or ⑦ in front of the syntagm ⑤ cannot survive because it is easy to determine that hunluan (混乱 'chaos') and shangwang (伤亡 'injuries') are the consequences of bombs going off.
In view of the facts above, a problem ensues when the duyujus are put side by side by spatial relations in (10) and temporal relations in (12). To be exact, the aforesaid facts we have unearthed seem to directly contradict Zhao et al.'s (e.g. Wang &Zhao, 2016, 2020;Wang & Liu, 2021) 'reversibility', which gives its speakers unusual flexibility in moving and even reversing elements. Put another way, it seems that the 'reversibility' appears to be fast-fingered in examples (10) and (12).
However, I list some cases in (13-14), where duyujus also work as independent syntagms within CRS, appear to run counter to the situations in (10)  Analogous to the syntagms ⑥ or ⑦ in example (12a), the syntagms ① to ④ in (13a) and syntagms ① to ④ in (14a) can also be moved and even reversed to other positions since both of them exhibit purely coordinate relations. In this sense, the 'reversibility' appears to be "welcome" at (13a) and (14a). To some extent, the difference between the present cases (13) and (14) and aforesaid cases (10) and (12) comes down to their different constructional environments. To be exact, clearly the 'reversibility' is given a cordial welcome by the former cases, but is shown as undesirable in the latter.
To wrap up, there is no dispute that it is not really the case as Wang et al.'s (e.g. Wang & Zhao, 2016, 2017a, 2017b, 2020Cui, 2017;Cui & Wang, 2019;Wang & Liu, 2021) 'reversibility' depicted. Albeit the removed and reversed duyuju syntagms exist indeed in CRS, there are some counterexamples where permission is denied to remove syntagms due to the spatial/temporal relations. However, admittedly, this kind of nuanced cases (10) and (12) involving duyujus as syntagms seem to not carry much conviction. So, we have better zero in on another case within CRS.

Shuncheng Relations in CRS
The linguists (e.g. Hu, 1984;Wang, 1994;Sheng, 2016) of the recent past have puzzled their brains about the multi-tiered relations among syntagms in CCS and have yielded fruitful results. For instance, four basic varieties, that is, coordinate relation, shuncheng relation, jieshuo/commentary relation and clausal relation, are gathered by Sheng (2016). The second type, shuncheng CRS, is a kind of sentence where the occurrences/actions depicted in preceding syntagms come before those in subsequent syntagms. It means that the sequence of linguistic forms in shuncheng CRS matches the actual order of experiences. Sentence (15), for example, is thought to be illustrative (Sheng, 2016 It is shown that the example (15d) is not felicitous as (15a) in logic owing to their distinction in the word order of syntagms. Specifically speaking, Intuitively, the actions described in syntagms ①, ② and ③ occurred ahead of that in syntagm ④. Put another way, other things being equal, the order of statements, namely, syntagms ① to ④, in a narrative description corresponds to the temporal order of the events they describe. In contrast, the temporal sequence presented in example (15d) goes against the time stream of the sentence builder's mind. Meanwhile, a big question-mark hangs over the logical relations among component clauses in (15d). Hence, the sentence (15d) in which the 'reversibility' is applied is left to be unacceptable.
Other than denoting temporal order like (15) Sheng (2016), with zeal and genius, unveils the secrets of the word order of syntagms in shuncheng CRS. In Sheng's (2016) shrewd cognizance, the order of linguistic forms in shuncheng CRS tends to be stereotyped. Put slightly differently, due to this kind of fixed sequence, the syntagms in shuncheng CRS cannot be removed or reversed. Furthermore, the sequence of linguistic forms always mirrors the spatial/temporal/logical sequence of actions/events depicted in shuncheng CRS (Sheng, 2016:87). If, in shuncheng CRS, some removements or inversions of syntagms surface, the conventional logical relations among them would be wrecked and the word order would run counter to the sequence of experiences (Sheng, 2016:87). All in all, to the best of my present analysis, a mini-conclusion can be arrived at is that the contention of Wang et al.'s (e.g. Wang & Zhao, 2016, 2020Wang & Liu, 2021) 'reversibility' also appears to be excluded in shuncheng CRS, as examples (15-17) illustrated.
Something also to note is that the unacceptability of 'reversibility' is not absolutely in shencheng CRS. Put another way, Sheng (2016:87) adds that sometimes this kind of CRS may be well-formed even under the condition that the positions of some clauses undergo ( (18d) and (19d) are both intuitively correct and make sense in logic, even though the syntagms ① and ② in (18a) and the syntagms ① , ② and ③ in (19a) undergo the 'reversibility'. Nonetheless, Sheng (2016:88) constructively holds that this kind of rearrangement in (18d) and (19d) would lead to a shift of their semantic relations compared with the word order of original sentences (18a) and (19a). In other words, at first sight, Wang et al.'s (e.g. Wang & Zhao, 2016, 2017a, 2020Wang & Liu, 2021) 'reversibility' seems to be accepted in shuncheng CRS sometimes, but the semantic/logical relations among synatgms would experience some (huge) shifts. Therefore, we might as well address the elephant in the room rather than temporarily leave aside the differences between (18a-19a) and (18d-19d).
In a nutshell, shuncheng CRS, as an indicator of temporal/spatial/logical relations, to some extent can serve as a sort of filler and a way of suggesting that, in practice, the 'reversibility' fails to work in some cases of CRS. It should be pointed out that shuncheng CRS holds a majority in the family of CRS. Therefore, together with the cases of duyujus in CRS, it is well-founded to call into question the rationale of 'reversibility'. To be more exact, based upon our observation of duyujus in CRS and shuncheng CRS, Wang et al.'s (e.g. Wang & Zhao, 2016, 2017a, 2017c, 2020Wang & Liu, 2021) 'reversibility' in fact is borne out to be a local or partial property rather than near-universal. The knotty problem here is that, in the words of Zhao et al. (e.g. 2016, 2020), the existence of 'reversibility', which in theory has the same status as the other two traits, 'chunkiness' and 'discreteness', should have been universally recognized in CRS in practice. Put another way, the 'reversibility' has to showcase a near universality and should be able to work in all types of CRS and all sorts of syntagms in CRS. And all syntagms in all varieties of CRS can be removed and even reversed to other positions. In short, the ideal universality of 'reversibility' in CRS is demonstrated to be totally at variance with the evidences presented in this article.

Sequential Order Principles on Clause Order
Up to now, we have arrived at a tentative conclusion that there exist some linguistic environments, namely, duyujus in CRS and shuncheng CRS, where the 'reversibility' is deemed as an "unexpected visitor". Under our analysis, what we have revealed with regard to the application of 'reversibility', as examples (10), (12), and (15-17) displayed, is presumably motivated by the same cognitive factor, that is, the iconicity of sequence/order (Haiman, 1980(Haiman, , 1984(Haiman, , 1985Jakobson, 1965Jakobson, , 1971Tai, 1985Tai, , 1989Givón, 1991;Shen, 1993) that to a large degree rules out the use of 'reversibility' in CRS.
Apropos of the iconicity of sequence/order, it is one of the widely accepted motivating principles subsumed under the broad heading of iconicity. Previous research endeavors (e.g. Haiman, 1980Haiman, , 1984Haiman, , 1985Tai, 1985Tai, , 1989Givón, 1991) pertinent to the definition of this principle can be encapsulated as follow: the sequence of component clauses or linguistic forms in connected discourse is iconically motivated by, or corresponds to, the sequence of the narrated occurrences in conceptual world.

The Principle of Temporal Sequence
According to the 'semantic principle of linear order' given by Haiman (1980), and the 'Principle of Temporal Sequence (PTS)' by Tai (1985:50), "the relative word order between two syntactic units is determined by the temporal order of the states which they represent in the conceptual world". In view of this, the unacceptability of examples like (12d), (15d), (16d) and (17d) can be predicted. Accordingly, the 'reversibility' turns out to be unable to work in these sentences. Another representative example (20) below can also be adduced. In the case of (20), following Haiman's (e.g. 1980Haiman's (e.g. , 1984Haiman's (e.g. , 1985 line of thinking, if we place the syntagm ① to the wake position, it would go against the order of a series of occurrences and actions in conceptual world. In light of the 'semantic principle of linear order' and 'principle of temporal sequence (PTS)', the actions pinpointed by a succession of predicates that firstly take place should ideally be described firstly, and those happen at a later time should also come next in the sentence. However, the example (20d) adduced here that involves the 'reversibility' goes in the opposite direction of the ideal case brought light on by Haiman (1980Haiman ( , 1984Haiman ( , 1985 and Givón (1991). Therefore, that is the reason why 'reversibility' is proved ill-suited in examples listed in this article.

The Principle of Spatial Sequence
As well as following Haiman et al.'s (e.g. 1980Haiman et al.'s (e.g. , 1984Haiman et al.'s (e.g. , 1985 statements, the "Principle of Spatial Sequence (PSS)" can provide a convincing account for the disuse of 'reversibility' in examples (10) and (16). This kind of principle underlines that the spatial sequence of constructing linguistics forms matches their spatial perception about the world. Hence, there is a strong tendency of sentence builders to order syntagms according to a certain kind of convention. In Liu's (2006) notes, spatial sequence, just like the temporal sequence, tends to arrange syntagms based on the size, the width, and the distance of space. And the order of depicting some images with spatial relations is constrained by body experience and mainly rests on the visual perception. Conventionally, Chinese speakers would rather to express spatiality in a way from nearness to farness. That is to say, dependent on the point of visual perception, sometimes the nearer image may be deemed as the "figure" (e.g. Langacker, 1991;Ungerer & Schmid, 1996;Talmy, 2000), and the farther image appears to be less salient and thus is the "ground" (e.g. Langacker, 1991;Ungerer & Schmid, 1996;Talmy, 2000). Other than that, other principles of spatial sequence like same-direction principle also exist (Sheng, 2016:68). I have in mind an example as the following: c. 'To the north of the T-shaped one is a river with five white stone bridges standing side by side on the river; to the further north is the rampart, in the center of which the tower of the Gate of Heavenly Peace rises'.

d. *③再北面是城墙，①丁字形一横的北面是一道河，④城墙中央高高耸起天安门的城楼，②河上并 排架着五座白石桥。 (李普《开国大典》 )
In (21), the sentence builder, with an inclination to express images in a way from nearness to farness, presents the river firstly and then depicts a picture of wu-zuo bai shiqiao 'five white stone bridges', chengqiang 'the rampart', Tiananmen 'the tower of the Gate of Heavenly Peace' successively.
All told, although Wang, Zhao et al.'s (e.g. Wang & Zhao, 2016, 2020Wang & Liu, 2021) analysis of 'reversibility' seems appealing, the problem that it has to come to grips with is that 'reversibility' is expected to work in all sorts of CRS, but instead is demonstrated as not simply random without rules. Namely, so far at least, the duyujus in CRS and shuncheng CRS alike refuse to accept Wang et al.'s (e.g. Wang & Zhao, 2016) insistence of 'reversibility'. Also, one cannot give a cold shoulder to some cognitive principles, such as the principle of spatial sequence, the principle of temporal sequence and same-direction principle (Sheng, 2016), under the label of the iconicity of sequence/order (e.g. Haiman, 1980Haiman, , 1984Haiman, , 1985 governing the clause order of CRS. Namely, although the 'reversibility' sometimes exhibits a tendency to change the word order of syntagms in CRS, there is a clear correlation between the clause order of CRS and iconicity account. In this sense, CRS is borne out to be not an exception to the iconicity of sequence/order.

Constraints on Reversibility
To the best of the present analysis, we have revealed that the 'reversibility' fails to be used at random in all sorts of CRS. Namely, the application of 'reversibility' has some limitations, which cannot get short shrift. Under our analysis, a condition that can rule out the application of 'reversibility' is called Reversibility Filter Condition:

(22) The Reversibility Filter Condition
The 'reversibility' will be ruled out under the condition that: i. There exist some temporal/spatial relations among syntagms; or ii. There exist some certain logical relations among syntagms.
Analogously, some typical constructional environments where the 'reversibility' can be licensed can be further encapsulated as Reversibility Condition: (23) The Reversibility Condition The 'reversibility' can be licensed if and only if: i. There exist purely coordinate relations but no temporal/spatial relation among syntagms; or ii. There exist certain cause-effect relations among syntagms; or iii. There exist certain condition-entailment relations among syntagms.
We can think of this condition as being like a driver's license. Only when we get a license at the Department of Motor Vehicles, can we drive. Therefore, we have to go there to get the license. The same is true for Wang et al.'s (e.g. Wang & Zhao, 2016, 2017a, 2017b, 2020 'reversibility' that needs a license to be able to surface in CRS. The Reversibility Condition herein is required to serves as a license for the occurrence of 'reversibility' in CRS. If 'reversibility' appears in some cases beyond the Reversibility Condition listed above or has some intolerable dissociations from the core cases enumerated in the Reversibility Condition, the 'reversibility' would be incapable of "driving" without license. In short, the Reversibility Condition in practice becomes a requirement that 'reversibility' can succeed being applied in CRS. Therefore, such requirement stipulated in (23) can be invoked to account for the fact shown in (24) (repeating (13) above) that involves a pure coordination among the underlined syntagms, gongyuan 'park', beihai 'the North Sea', tiantan 'the Temple of Heaven' and wanshengyuan 'zoo'. Since the example (24) is able to satisfy the Reversibility Condition, the positions of syntagms from ① to ④ can be interchanged and even reversed.  Wang and Zhao's (e.g. 2016) 'reversibility' is merely a local property rather than a general or overall one. The reason why the clause order in CRS sometimes appears to be not fixed and even some reversed phenomena can surface are precisely because, on the one hand, some syntagms of CRS are in certain cause-effect and condition-entailment relationships, by virtue of which new orders such as "effect-cause" and "entailment-condition" ensue. On the other hand, if there is a pure coordination in CRS, there would be no dispute that the appearance of removed and reversed syntagms is allowed.

Conclusion
In this study, the present author does not deal with all three traits of Wang, Zhao et al.'s (Wang & Zhao, 2016, 2017a, 2017b, 2017c, 2020Cui, 2017;Cui & Wang, 2019;Wang & Liu, 2021) spatiality of CRS, such as 'chunkiness' and 'discreteness'; rather, I attend to a further discussion of 'reversibility' and an attempt at oppugning the rationale of 'reversibility' has been made.
On the whole, based upon the facts of duyujus in CRS and shuncheng CRS, it is argued here that, in most cases, the composition of CRS is still governed by the "iconicity of sequence/order" (e.g. Haiman, 1980Haiman, , 1984Haiman, , 1985, which generates the Reversibility Filter Condition in (22). Nonetheless, the 'reversibility' is also proved to be desirable in some scenarios listed in the Reversibility Condition (23). Given that, the present analysis has revealed that the 'reversibility' turns out to have no same near universality as other spatial traits, namely, 'chunkiness' and 'discreteness', but instead to be bound by the Reversibility Filter Condition (22). In view of that, two summings-up can be further reached. For one thing, the arguments of Wang et al. (e.g. Wang & Zhao, 2016, 2020Cui & Wang, 2019;Wang & Liu, 2021) with regard to 'reversibility', as a defining property of CRS, seems to be unable to hold water. When the 'reversibility' is borne out to be not peculiar to all types of syntagms (e.g. duyuju as syntagm) and all varieties of CRS (e.g. shuncheng CRS), we suggest that the spatiality of CRS should be ameliorated as two traits, that is, 'chunkiness' and 'discreteness'. For another thing, notwithstanding some exceptions wrapped up in (23), CRS has been demonstrated to be not an exception of the "iconicity of sequence/order" (e.g. Haiman, 1980Haiman, , 1984Haiman, , 1985. On the contrary, most of the time, the composition and clause order of CRS is iconically motivated by the temporal/spatial order in conceptual world. Therefore, CRS, "as a mirror of Mandarin Chinese" (Wu & Liang, 1992:316), together with other Chinese sentences, emerges to be a kind of iconicity-prominent sentence/utterance.
In spite of the major conclusions above, this article is far from an extensive paper with some limitations. Firstly, the cause-effect relation and condition-entailment relation in the Reversibility Condition (23) are left unaccounted for. Secondly, we believe that, besides the duyuju syntagm in CRS and shuncheng CRS, there must be other linguistic evidences for the unacceptability of 'reversibility' waiting to be presented in future studies. Thirdly, the constraints in the Reversibility Filter Condition (22) and applicable possibilities listed in the Reversibility Condition (23) maybe have not been exhausted. Lastly, the present author has merely taken the "iconicity of sequence/order" into consideration. Perhaps, there also exist other cognitive mechanisms for the use of 'reversibility'.