22.1 The Momentum of Language Change
Language development receives impetus from momentum, an inertial power gained from the pre-existing system. A language is an ever-changing body that is energized by the collective cognition of the language community. Every speaker is subject to the impetus gained by this ever-changing body. Factors related to language acquisition, pragmatic inference, and language contacts cannot cause any change unless they are in accordance with the momentous development that the language is undergoing. Once a language gains momentum, it will develop in that direction for many centuries, and no individual can stop or alter its course. This point of view is illustrated by the following two cases.
First, in Late Old Chinese, momentum was gained that forced all non-resultative elements to be moved from the postverbal to preverbal position. This change lasted more than a millennium and involved many major syntactic constructions, such as the passive, comparative, locative, instrumental, and ditransitive constructions. During this process, a set of ordinary verbs that occurred in the first verb position of a serial verb construction were stimulated to develop into grammatical morphemes, taking over the functions of the former ones that had been in postverbal position. Consequently, the texture of the grammar was fundamentally altered. The grammaticalization processes of ordinary verbs must take place in specific contexts that are subject to the influence of semantic suitability, frequency, and pragmatic inference. As a whole, however, any concrete change must be in line with the general development of the grammatical system. Meanwhile, any deviation from the direction of development could not survive and was eventually abandoned. For example, in Late Medieval Chinese, the disposal construction had two major markers, jiāng and bǎ, but the former could occur only in postverbal position and the latter only in preverbal position. Clearly, the preposition phrases in the disposal construction were non-resultative; consequently, the jiāng construction was abandoned, and only the bǎ construction is still used today.
Second, considering the way of introducing an agent, the passive structure has undergone three major changes: in Old Chinese, no passive morphemes could introduce an agent noun in preverbal position; in Medieval Chinese, it was optional for passive morphemes to introduce an agent noun in preverbal position; and in Modern Chinese, it became obligatory for passive morphemes to introduce an agent noun in preverbal position. This momentum moved steadily in a given direction. This course determined the fates, lifespans, functions, and lexical sources of passive morphemes. Historically, there were nearly twenty passive morphemes with different lexical sources and numerous passive structures with different functions. These phenomena appear extremely complex but became truly regular from the perspective of the developmental momentum.
Our research indicates that none of these changes occurred in isolation or were simply caused by accidental factors. These facts call into question three widely known hypotheses about factors involved in language change; that is, (a) children’s language acquisition, (b) pragmatic inference, and (c) language contact, all of which assume that there is a static language system that may be changed by accidental factors. This is not the case. All individuals must be influenced by the momentous power of the ever-changing language from the moment they are born.
Therefore the big picture is that the evolution of the grammatical system is regular, is systematic, and has a direction, and each characteristic is illustrated as follows. By “regular,” we mean that grammatical changes are often free of exceptions and must proceed in line with certain rules. When the information-structuring principle of the predicate was established, all related constructions had to change, which involved the locative, passive, instrumental, comparative, and ditransitive constructions. In this case, for example, the comparative construction could not remain unchanged.
The term “systematic” here has two senses. First, the overall properties of the grammatical system at the time play a crucial role in motivating and controlling an individual change, which involves the recruitment of lexical sources, the life span and function of a grammatical morpheme, the concrete structure of a grammatical device, the direction and fate of a grammatical form, and so forth. Second, only pragmatically induced changes that are consistent with the trend of the grammatical development can survive and develop into a new grammatical category.
Note that the meaning of directionality here is different to grammaticalization theory, which defines a cline along which a lexical item develops toward a grammatical marker. The expression “has a direction” here refers to the fact that the grammatical system develops in a direction for hundreds of years and even millennia, and this development is not interrupted by so-called children’s language acquisition or different generations’ parameter settings. Considering the presence of an agent noun, for example, the passive markers in preverbal position have undergone three stages: they could not introduce the agent in Old Chinese, they could optionally introduce the agent in Medieval Chinese, and they must introduce the agent in Modern Chinese. There is no fluctuation with respect to this feature of the passive structure.
22.2 Changes in Cluster
As we analyzed in this book, there have been hundreds of changes in Chinese grammar in history, and none of them happened in isolation. We use the term “change in cluster” to describe the characteristics of the evolution of the grammatical system. This term can be illustrated by a Chinese proverb, “Pull one hair and you move the whole body,” which means that a slight move in one part may affect the whole situation and that a minor step may have major consequences. This reveals the most essential property of the diachronic system: all parts form an integral whole, similar to the human body or an ecological circumstance, where, if any individual part changes, it will affect the most closely related parts and even the whole system.
Thus far, no theoretical frameworks in historical linguistics can predict and explain the model called “language change in cluster,” a phenomenon that has happened repeatedly in the history of Chinese grammar. There are several hypotheses available in the literature and their predictions of the fashions of grammar changes are briefly stated here. According to the principle-parameter hypothesis, the set of principles remains stable and the parameters may change due to the settings of different generalizations (Reference Lightfoot, Joseph and JandaLightfoot 2013). This hypothesis predicts that the development of a grammatical system proceeds abruptly, and a parameter change can be applicable only to similar structures, such as those consisting of the modifier and the head. However, a cluster of changes usually includes various types of construction or grammatical morphemes that bear the least structural assemblies, such as the emergence of the copula shì and the typological change in the syntax of wh- words. More importantly, there is no evidence that any change is abrupt in pace over changes in generations.
According to the usage-based model, grammatical innovations are most likely to happen to high-frequency words, and a grammaticalization process is accompanied by an increase in token and type frequency (Reference Bybee, Hoffmann and TrousdaleBybee 2013). One key issue that must be resolved here is that, in many cases, the innovative form violates certain existing grammatical rules; for example, the emergence of the resultative construction broke the rule governing the co-ordinate verb construction. How could this new form come into existence, mature, and finally replace the former grammatical rule? No change is a matter only of its own usage, such as repetition and entrenchment of cognition, but must be motivated and guided by the overall properties of the grammatical system. Additionally, this model can only predict that any change in grammar occurs in isolation and cannot explain why change typically occurs in a cluster of constructions rather than individually.
In grammaticalization theory, the major factor in initiating a change is pragmatic inference (Reference Hopper and TraugottHopper and Traugott 2003: 71‒98). Theoretically speaking, every language user may have their own pragmatic inferences; thus the question is, what factor determines which pragmatic inference will be recognized by the whole language community and finally become a new grammatical device? More crucially, this hypothesis can only predict that any new grammatical form will happen individually and that each change has its own pragmatic inference. According to our research, only pragmatic inferences that are favored by the overall properties of the grammar in the particular period can cause a change in a grammatical device in a morphosyntactically specific context. It is true that any grammatical form must develop out of a proper context, unlike lexemes, which can be coined independently of any linguistic construction. The necessary condition guarantees that the new form is to a great extent structurally consistent with the existing grammatical system.
The history of Chinese grammar encompasses several major clusters of changes, from which we can see that the developments are highly regular. These clusters are summarized as follows.
Cluster 1: the emergence of the copula shì. The copula shì grammaticalized from its original demonstrative usage, in which it was often used as an anaphor between two nominal phrases. This development was encouraged by the SVO word order at the expense of the old copular construction that was marked by the sentence-final particle yě; in turn, the new copular construction marked by shì further strengthened the feature of word order. Later, the copula shì developed into a focus marker as an analytic device to replace the old method of highlighting by shifting the constituent order. Consequently, the old focus constructions disappeared; specifically, movement of three types of element ceased to be used, including wh- words, pronouns in negative construction, and the displacement of nouns. Furthermore, wh- words exhibited their inherent focus features by fusing with the focus morpheme shì, which resulted in the replacement of all interrogative pronouns.
Cluster 2: the emergence of the resultative construction. The emergence of the resultative construction may have been the most important event in the whole history of Chinese grammar, as it fundamentally changed the texture of the grammar and distinguished the system of Contemporary Chinese from that of Classical Chinese. This change was enabled by two high-level conditions, one intermediate-level condition, and three low-level conditions. At the high level, the disappearance of the conjunction ér created the serial verb construction in which two verbs occurred adjacently, and the disyllabification tendency fused two monosyllabic units that co-occurred frequently. At the intermediate level, the co-ordinate verb construction required that only two or more transitive verbs could precede an object and no other element could appear between the transitive verb and the object. This condition channeled the process from which the resultative construction developed. The establishment of the resultative construction was responsible for various changes, mainly including the disappearance of the co-ordinate verb construction; the emergence of the disposal, verb-copying, and SOV constructions; the emergence of the aspect system; the formation of verb reduplication; the boundedness of the predicate structure; and the formation of the negative system in Contemporary Chinese. Furthermore, due to its high frequency and wide scope of usage, the resultative construction assimilated other constructions, such as the ditransitive and some verb-plus-object constructions.
Cluster 3: the establishment of the principle governing the information structure of the predicate. This principle was established by analogy with the resultative construction, which required that only preposition phrases that expressed some kind of resultative could stay in postverbal position and all others that were non-resultative could appear only in preverbal position. Under the operation of this principle, many types of sentential structure underwent fundamental changes in constituent order, including locatives, comparatives, instrumentals, and ditransitives. During this change, a set of verbs in the first verb position of a serial verb construction became grammatical morphemes to take over the functions of the older forms. Under the operation of this series of changes, the relative clauses that formerly followed the head noun were restricted to prenominal position. Consequently, the constituent order within a nominal phrase developed from two types in Old Chinese, either head–modifier or modifier–head, into one “modifier–head” order.
Cluster 4: the establishment of the classifier system. The establishment of the classifier system was the outcome of the combined powers of the disyllabification tendency and the information-structuring principle of the predicate. The emergence of classifiers added a new word class to the grammar and profoundly affected the structure of nominal phrases: first, the numeral modifier had to be associated with its head by a classifier; second, by analogy, the demonstrative dǐ was triggered to develop into an associative morpheme to mark relative clauses, genitive phrases, and the adjectival phrases. The emergence of the classifier system made the old demonstratives unsuitable for the new nominal structures; at the same time, the most general classifiers zhè and gè developed into proximal demonstratives in the northern and southern dialects respectively. Additionally, the classifier system motivated the grammaticalization process of the plural -men from a classifier and the diminutive marker -er from the concept “child.”
22.3 Patterns of Grammatical Change
The patterns of grammatical change are far richer than those identified in the literature and expected by any framework. According to our research, these patterns include the following:
Type 1: the disappearance of grammatical forms. The history of Chinese grammar consists of two essentially related sides: the appearance and disappearance of grammatical forms. Grammatical change is similar to the metabolism of any living creature, with new forms being constantly added while old forms are abandoned. There is no language that just increases the number of its grammatical forms without removing old ones. To reveal the regularity of grammatical development, studying the disappearing phenomena is equally important because it may create a condition for the emergence of a new form (e.g. the disappearance of the conjunction ér made it possible for the verb and the resultative to become fused) and may control the course of grammaticalization (e.g. the development of the resultative construction was heavily influenced by the co-ordinate verb construction, and the disappearance of the latter was caused by the former).
Type 2: replacement of the old by new forms. Some historical grammatical forms simply disappeared without any replacement, such as the verb conjunction ér, which meant that the grammatical category no longer existed in the language. However, it is common for the grammatical category to remain the same while the form changes, which may involve replacing markers and altering structures. For example, the comparative construction underwent the replacement of the marker from yú to bǐ and the repositioning of the standard phrase from following the predicate to preceding the predicate.
Type 3: innovation in grammatical category and form. Historically, there were many innovations that did not exist before, such as the disposal construction, the verb-copying construction, the classifier system, the plural morpheme -men, and the diminutive particle.
Type 4: grammaticalization from lexical sources. A majority of grammatical markers developed out of lexical items. For example, the perfective aspect -le came from the verb “finish,” the plural marker -men from the classifier for enumerating families, and the diminutive -er from the concept “child.”
Type 5: phonological derivation. Not all grammatical markers must originate from a lexical source; phonological derivation is another way for them to develop. This has often happened in two complementary categories within a functional domain. The exemplar case is the proximal and distal demonstratives, where the proximal part developed out of a general classifier and the distal part was phonologically derived from the proximal.
Type 6: change in function. The same marker may acquire a new function over time. For example, the passive bèi could not introduce an agent for the first 600 years or so when it grammaticalized from the verb of suffering, but it later acquired that function.
Type 7: obligatory constituents. For the passive structure, for instance, the presence of the agent phrase was originally optional, but it became obligatory in Contemporary Chinese by analogy with the disposal construction. From the very beginning, the presence of the patient noun in the disposal construction was obligatory to make the construction well formed.
Type 8: change in syntactic category. In Old Chinese, the category of aspect was expressed by a set of auxiliary verbs, but since the beginning of Modern Chinese it has been expressed by a set of verb suffixes.
Type 9: change in the function of syntactic position. This change does not involve any overt formal feature. For example, in Medieval Chinese, a rule took place: a bare noun in the subject position was automatically assigned the feature definite, whereas a bare noun in the object position was assigned the feature indefinite. In turn, this rule had a profound effect on the emergence of the disposal construction and the extension of wh- words to universal and indefinite reference.
Type 10: the borrowing of a grammatical category. This change typically happened in some isolated dialects and is similar to loanwords. The purely innovated grammatical categories tended to be borrowed by isolated dialects. In the vast area of the northern dialect, the diminutive marker -er grammaticalized from the word ér “child” in Medieval Chinese. For geographic and social reasons, it could not spread to many isolated dialects. However, these dialects simply borrowed this new grammatical category by assigning it a phonological form, such as a high-pitched tone, a reduplicated form, or a nasal consonant.
The above list represents the major types of grammatical change but is not exhaustive.
22.4 Drivers of Language Change
From one perspective and to a certain extent, language is like any biological phenomenon, with any current state evolving from its former state. However, every human being undergoes the stages from infant to toddler, adolescent, and adult. Even when a person becomes an adult, the infant from whom the adult grew is still alive, but with a different shape. Similarly, Old Chinese, which existed more than three millennia ago, is still alive today and has evolved into Contemporary Chinese. Only a language that has ceased to be used can be regarded as dead; for example, Sumerian, the oldest-known written language, which dates back to at least 3500 BC but only lasted until 2000 BC, is a dead language. This metaphor is somehow misleading because it implies that a language in the early stages might be simplistic and destined to die. However, the grammatical system of a language at an earlier stage is by no means simpler than that at a later stage. A language will exist as long as it is used as a spoken language.
The life of a language consists of two parts: the cognition of the language community and the symbolic system for communication. The cognition of the language community determines the openness of human language, and the symbolic system renders the closeness of human language. Once again, this is something like the human body. To maintain life, humans must interact with the outside world, such as absorbing food and inhaling oxygen; meanwhile, the organs in their bodies need to work co-operatively, such as blood circulating and the heart beating. Therefore the motivations and mechanisms of the evolution of the grammar fall into two types, both outside and inside the grammatical system. Theoretically, any phenomena that can be the object of human cognition, which might be social, natural, and even imaginary, may affect the language system. The history of Chinese shows that there has been a massive inertial force of cognition within the language community that has overridden any influence from individual pragmatic inference or language acquisition across generations. Meanwhile, many changes are triggered simply by certain internal factors. Although the reduplicated forms of adjectives and nouns were already widely used in Old Chinese, verb reduplication did not come into existence until the fourteenth century AD, when the establishment of the resultative construction and the emergence of the aspect markings co-operatively created a new syntactic slot between the verb and the object.
Due to outside cognitive activities and inside systematic factors, any language is always in the course of evolution and many developments happen necessarily rather than possibly. Changes in grammar, far from being random, independent, or unpredictable, are rather regular, systematic, and predictable.
22.5 Construction Schema
Historically, the structures of passives, locatives, comparatives, instrumentals, and ditransitives have changed dramatically, but these changes happened in a consistent way, from which we can see that the construction schema was at work in the evolution of grammar. In Section 7.9, we discussed the formation of the principle of action–resultative ordering, which required that all non-resultative preposition phrases and adverbials disappear from postverbal position. These events resulted in a series of changes in sentence structures involving all the constructions mentioned above. As we saw previously, various passive morphemes and structures have existed throughout history. From a different perspective, this chapter aims to identify a new type of grammatical form, called the “construction schema,” that was at work throughout history, an essential concept for capturing the regularity of grammar development.
In all major theoretical frameworks, functional constructions, such as passives, comparisons, and ditransitives, are regarded as the largest linguistic unit or maximal schema (e.g. Reference ChomskyChomsky 1957, Reference LangackerLangacker 1987, Reference GoldbergGoldberg 1995, Reference CroftCroft 2001). In empirical studies of language, the maximal forms in linguistic typology are word orders, such as SVO and SOV (e.g. Reference Greenberg and GreenbergGreenberg 1966a, Reference ComrieComrie 1981, Reference CroftCroft 1996). For instance, Reference ChomskyChomsky (1957), Reference GoldbergGoldberg (2006), and Reference CroftCroft (2001) specifically used the passive as their largest construction (“mostly schematic,” in Croft’s words). They portrayed the passive construction in English:
(1)
(a)
NP2 – AUX + be + en – V – by + NP1 (Reference ChomskyChomsky 1957: 43), (b)
Subj AUX VPpp (PPby) (Reference GoldbergGoldberg 2006: 5), (c)
SBJ be-TNS VERB-en by OBL (Reference CroftCroft 2001: 5).
In the evolution of the Chinese language, however, we believe that the passive is not the most schematic construction and that a type of linguistic form, termed the “construction schema,” exists between these functional constructions and basic word orders. It is abstracted from a concrete construction, such as the passive, locative, and comparative, on the basis of their shared function and form. The evidence from the history of Chinese clearly shows that a construction schema has been at work in determining the diachronic changes in Chinese passives. Within a given period, the construction schema determines what must happen, what cannot happen, and in which direction the language might develop.
A construction schema is the outcome of a general form of human cognition – schematization (Reference HintzmanHintzman 1986, Reference Rimé, Philippot and CisamoloRimé et al. 1990). This type of cognition is at work at all levels of linguistic units, including syllables, words, phrases, and sentences. This concept has been applied to language acquisition (e.g. Reference BybeeBybee 1985, Reference Bybee, Bybee and Fleischman1995, Reference TomaselloTomasello 2003, Reference GoldbergGoldberg 2006: 220). According to Reference LangackerLangacker (2013: 167‒173) and in cognitive linguistics, grammar is regarded as a schema with different levels of internal complexity, in contrast to generative linguistics and OT theory, which concern rules or constraints. Recently, an increasing number of researchers have applied this concept to historical investigations, claiming that grammaticalization, especially constructionalization, is related to schematization (Reference NoëlNoël 2007, Reference TraugottTraugott 2007, Reference Trousdale, Trousdale and GisborneTrousdale 2008, Reference Traugott and TrousdaleTraugott and Trousdale 2013: 13‒16). In the literature, the most abstract schema is not beyond the functional construction; for instance, in Reference Traugott and TrousdaleTraugott and Trousdale (2013), the “macro construction” refers to the “give-gerund” (e.g. Tom gave John a push). In addition, all these studies suggest that schematization works in a “bottom-up” fashion, namely how a construction is created via high frequency or entrenchment. However, in this chapter, we discuss how schematization works in an “top-down” fashion – guiding the direction of grammar evolution.
Reference KiparskyKiparsky (2013) provided an extensive investigation of passives in numerous languages, arguing for a “null theory” in line with OT-based lexical decomposition grammar (Reference StiebelsStiebels 2002, Reference WunderlichWunderlich 1997, Reference Wunderlich, Bornkessel, Schlesewsky, Comrie and Friederici2006), a base-generated syntax that eliminates NP movement. His central viewpoint was that there is no specific construction exclusively for the passive and that the syntax of the passive is parallel to other syntactic structures of the same abstract schema. Within Kiparsky’s framework, the passive clauses in a language have no passive-specific syntactic properties, and the syntax is predictable from the active sentences of the language and from the argument structure of passive predicates. Therefore he concluded that “a learner who knows the grammar of active sentences of a language can predict the syntax of other diatheses” (Reference KiparskyKiparsky 2013: 7). In what follows, we demonstrate that the passive and many other active clauses in Chinese actually belong to the same construction schema and that they underwent parallel developments in history.
Construction schemas are universal, reflecting the nature of human language. For instance, the most general pattern of English may be abstracted as follows:
(2) The English passive: be + V-ed (past participle) + by NP.
In the generativist tradition, all components of the passive structure are meaningless, generated from so-called deep structures (typically an active form) through a set of formal rules (e.g. Reference ChomskyChomsky 1957: 61‒84). However, this English passive pattern is by no means universal; for example, the passive in Japanese does not have an auxiliary (copula) and uses the preposition ni to introduce the agent in preverbal position. These differences are determined by the overall properties of individual grammars.
Reference LangackerLangacker (2013: 121‒126) argued that no parts of the English passive are exclusively designed for the passive only and that all of them are meaningful. In his theory, only the copula and the past participle are the central parts; the by-phrase is marginal, closely related to its locative usage. Thus the concrete construction is licensed by a more abstract schema:
(3) AUX + V-ed + PRE + NP.
“Vi” could be either a present participle or a past participle. That is, the following sentences share the same construction schema as passives:
(4)
(a)
The window was broken by Tom. (b)
Bill was sleeping by the table. (c)
They had finished their homework in school. (d)
John is reading newspapers in the dining room.
Within a synchronic system, it may sometimes be difficult to determine which constructions are grouped to form a schema. As indicated in the next section, however, it is easy to identify how many concrete constructions a construction schema includes in the evolution of a grammatical system.
Reference LangackerLangacker (2013: 167) first proposed the term “constructional schema.” In his model, linguistic units, from simple words to complex constructions, are by nature symbolic assemblies that can be either specific or schematic. In our definition, in contrast, construction schemas refer only to those schematized from clauses or sentences. In the present analysis, a construction schema is understood as an abstract template that treats passive clauses and many other active clauses as members – a linguistic level that is between a functional sentential structure (e.g. passive, ditransitive) and so-called basic word orders (e.g. SVO, SOV).
(a) Word order refers to the linear arrangement of the three basic constituents – subject, verb, and object. Within a given language, the basic word order is stable, and there is usually some variation due to pragmatic factors. Word order is a matter of convention and hence does not express any semantic or pragmatic values.
(b) The construction schema refers to the linear orderings involving all elements composing a sentence, such as the three basic constituents, adjuncts (e.g. preposition phrases), morphological markers, and word classes. It is schematized from functional constructions, such as passives, ditransitives, and resultatives, as a clause-level form with an identifiable meaning and form.
(c) Functional construction refers to types of clause structure with different semantic or pragmatic values, such as passives, ditransitives, resultatives, topicalization, and questions. In this chapter, the term “construction” refers to clause structures, excluding words, morphological markers, and phrasal structures (for the definition of construction, see Reference GoldbergGoldberg 1995: 1‒5 and Reference CroftCroft 2001: 14‒28).
The identification of this linguistic form, namely the construction schema, will greatly benefit researchers in historical linguistics, as shown in the following analysis.
Grammatical schemas may exist at phrasal levels or syntactic levels. For example, due to the emergence of the classifier system and the associated particle de, the schema of the nominal phrase changed from “modifier + head” to “modifier + grammatical marker + head.” The establishment of the resultative construction created a syntactic schema “V + X + O,” where the X element is intransitive and refers to the state of the progression of the verb. It replaced the former schema “V1 + V2 + O,” where the two verbs had to be transitive and each governed the object individually. The information-structuring principle of the predicate caused a series of structural changes whose combined outcome was the formation of two subtypes of syntactic schema: (a) “Subj VO PPresultative” and (b) “Subj PPnon-resultative VO.”
Having identified the existence of the construction schema, we can see that the evolution of morphosyntactic history becomes highly regular, and the motivations for changes in the passive can be easily explained. This construction schema involves at least five major functional constructions: passive, comparative, instrumental, locative, and disposal. They were grouped together by the new information-organizing principle: all non-resultative preposition phrases had to occur in preverbal position, which caused the following three changes. First, all prepositions that introduced non-resultatives, mainly including yǐ (instrumental) and yú (passive, locative, comparative), disappeared from preverbal position. Second, many verbs grammaticalized in the first verbal position of a verbal serialization, expressing the functions of the passive. Third, the preposition phrases that introduced an agent noun were restricted to preverbal position. As a result, the passive morpheme jiàn was abandoned because it never developed the function of indicating an agent, bèi evolved a new function of introducing an agent, and the wèi … suǒ pattern was strengthened because it was able to introduce an agent in preverbal position even in the Old Chinese period.
Moreover, in the new construction schema, preposition phrases in preverbal position had to be followed by a nominal phrase to make the structures well formed. In particular, the disposal construction emerged around the eighth century AD, and the patient noun needed to occur to have a legitimate construction. They were functionally complementary, and the newly formed passive patterns required an agent noun to be grammatical. This requirement had the following consequences. First, the passive bèi quickly declined in use because it always optionally introduced an agent noun, although this passive pattern was used for more than 2,300 years. Second, the chī passive existed only for a very short period and disappeared entirely because it was never fully developed and was subject to elimination under the strong trend. Third, the newly grammaticalized passive morphemes, namely, jiào, ràng, and gěi, that were introduced into the language in the last 200 or 300 years shared the same syntax: the presence of an agent noun is obligatory in order to make the structure well formed. This usage is typologically peculiar.
Additionally, the lexical sources for grammaticalization to become passive morphemes were extremely diverse in Modern Chinese, and there were at least sixty-nine different ones among the 930 dialects, including “allow,” “call,” “give,” “wait,” “ask for,” and “want.” All these verbs, which are typically followed by objects, frequently occur in the first verbal position of a serial verb construction structure, the context responsible for the change. Thus their grammaticalization paths appear very natural if the internal construction network is taken into consideration.
Finally, the length of the life spans of grammatical morphemes and the selection of lexical candidates for certain grammatical domains are determined largely by the construction schema in a particular period. Throughout history, the life spans of different passive morphemes varied greatly; for instance, bèi has been used for more than 2,000 years, but chī existed for only roughly 600 years. In fact, their fates were determined by the construction schema as well as by their abilities to adapt to the overall grammatical property. The selection of lexical items in grammaticalization was also by no means accidental and was often determined by the overall grammatical property. The so-called frequency and pragmatic inferences in triggering grammaticalization, however, are entirely independent of the grammatical system.
From the perspective of the construction schema, it is safe to say that no changes are random or accidental. Beyond diachronic investigation, we can also see how the grammar of a language operates consistently. Additionally, the traditional distinction between passive and active disappears; thus all theories based on this distinction need to be reconsidered.
The identification of a construction schema also has profound implications for theoretical linguistics. Since the inception of generative linguistics, it has been assumed that every passive sentence is transformed from its corresponding active sentence. Reference ChomskyChomsky (1957: 43) explained it as follows:
(5)
If S1 is a grammatical sentence of the form NP1 – AUX – V – NP2 then the corresponding string takes the form NP2 – AUX + be + en – V – by + NP1 is also a grammatical sentence.
This hypothesis faces many empirical problems; for instance, many active sentences do not have corresponding passive forms, and the opposite is also true. This theory is powerless to explain why the passive form in a particular language has undergone fundamental changes. Our analysis has proven that the so-called passive and active forms actually belong to the same construction schema, which makes the grammar considerably simpler. We believe that we will gain a deeper understanding of grammatical evolution when our research agenda includes construction schemas from cross-linguistic perspectives.
