Has Chinese always been an analytic language? Effects of writing on language evolution

1 Introduction

It is common knowledge that Chinese is a typical analytic language. In the meantime, the writing system of Chinese, consisting of Chinese characters, is widely recognized as a representative ideographic writing system. When it comes to the relationship between the spoken language and the writing system, ever since Saussure, writing is assumed to be secondary, hence going with the spoken language. This mindset creates an impression that the analytic Chinese language deliberately chooses an ideographic writing system as it fits the analytic nature of the language (e.g., Istrin 2002: 138, 504). However, recent studies proposed various types of morphology in Old Chinese including affixation, inflection and alternation (to be reviewed at length in Section 2), thus fundamentally casting doubt on the assumed secondary position of writing to the spoken language: Old Chinese evidently did not pick a writing system that fits it. For this point, this paper calls for reverse thinking by shedding light on the effects of writing on language. At least in the case of Chinese, the use of an ideographic writing system plays a non-trivial role in shaping the path of language evolution. Section 2 briefly reviews the different types of morphological operations in Old Chinese proposed by linguists. Section 3 analyzes the role that writing plays in the disappearance of morphology, and then provides a new model of Chineseness to account for the language and the writing at the same time. Section 4 tries to extend the model to other languages by highlighting the effects of writing on language variation and change in general. Conclusive remarks are presented in Section 5.

2 Morphology in Old Chinese

2.7 Summary

Section 2 reviews linguists’ reconstructions of morphology in Old Chinese. Overall, those operations mentioned in philological works, i.e., voiceless-voiced alternation, vowel alternation, vowel length alternation and tonal alternation, are firmly believed to exist. In contrast, affixes that are speculated based on cross-linguistic comparisons but not mentioned in philological works are much more controversial, in spite of the fact that those affixes are suggested to be the origins of other operations. That being said, it becomes a perplexing problem to identify the functions of each operation. The *s- prefix, *-s suffix, *-r-/-l- infix, voiceless-voiced alternation, and tonal alternation are all suggested to be related to valency change. Besides, the *-s suffix, *p-(b-) prefix, and tonal alternation are all suggested to mark the perfective aspect. Little systematicity can be observed from discrete examples, and scholars oftentimes disagree with each other.

Despite the disagreements, the (probable) existence of all those morphological operations evidently shows that Old Chinese was not a thoroughgoing analytic language, at least not so analytic as Modern Mandarin. As Chinese evolved, a large part of morphology was lost, making this language more analytic than before. The speculated path of morphological evolution in Chinese can roughly be summarized below in Figure 1.

Figure 1:

Speculated path of the morphological evolution in Chinese.

In Figure 1, the yellow shade represents the use of affixation, the red shade represents tonal alternation, and the blue shade represents voiceless-voiced alternation. The darkness of the shade roughly represents the frequency of the specific operations. Gradual transitions can be observed from affixes to tonal alternation, and from prefixes to voiceless-voiced alternation, indicative of continua. When it comes to Modern Mandarin, only tonal alternation is partially preserved, while all other operations disappeared.

3 Ideographic Chinese characters and the analytic Chinese language

To understand the disagreements regarding the morphology of Old Chinese, one thing to note is the synchronic and diachronic variations of Chinese. It is well-known that the idea of “Chinese” is not defined on the basis of mutual intelligibility, as Norman (1988: 187) puts it:

The above paragraph explains the synchronic variations between Modern Mandarin and other Chinese dialects, and the regional and diachronic variations of Old Chinese before the Qin dynasty (221 BC – 206 BC) could only be bigger. In fact, before Qin’s unification of China, the control of the royal house over different states was rather loose. The development of each state was largely independent of each other. In Mencius, there is the following sentence to describe the southern accent spoken in Chu around the Yangtze River:

In the Han dynasty, 扬雄Yang Xiong (53 BC – 18 AD) compiled a dictionary called Fangyan (《方言》), containing over 9,000 characters, which collects synonyms used in different areas. For example:

The above examples all suggest that dialect differences were huge at that time: even basic vocabularies might have dramatic differences. Coupled with the excessively long time span of Old Chinese, a unitary phonological system of Old Chinese is simply unimaginable. The lack of systematicity in linguists’ reconstructions is therefore easily accounted for, and there is technically no way for researchers to pinpoint the phonological system used at a specific place and time. It is surely possible that some morphological operations only existed in a certain area during a certain period of time, and that some morphological operations evolved faster in a certain area than other areas.

If the Chinese language is not defined upon mutual intelligibility, then how is it defined? The answer to this question resides in writing. Although Chinese involves wildly divergent phonological systems, noteworthily, this language has been constantly using the same writing system consisting of Chinese characters. The common language identity arises from the common writing, and the characteristics of this writing system are able to account for the loss of morphology.

As an ideographic writing system, each Chinese character has the graphic form, the phonetic form, and meanings. For the phonetic form, each character is pronounced as one syllable in Modern Mandarin, but distinct from syllabaries, many characters may have the same pronunciation. In fact, there are about 400 syllables (1,300 with tones) in Modern Mandarin, but Xiàndài Hànyǔ Cídiǎn ‘Modern Chinese Dictionary’ (《现代汉语词典》, the official standard in Mainland China) collects over 13,000 characters. As an extreme example, over 200 characters are pronounced as yì, each with distinct meanings. From a diachronic perspective, texts written in Chinese characters are passed down from every historical period uninterruptedly, serving as handy dictionaries for writers of later historical periods. Writers’ use of each character generally conforms to its meanings and functions as illustrated in previous literature, and thereby the correspondence between the graphic form and the meanings has been relatively stable. In comparison, the association between the graphic form and the phonetic form is much looser, as ideographic characters never provide precise records of the phonetic forms. Essentially, sound is momentary, always limited by time and space, whereas written texts composed of characters can be delivered across time and space. The functional structure of ideographic characters can roughly be represented in Figure 2.

Figure 2:

Functional structure of ideographic Chinese characters.

According to linguists’ reconstruction of Old Chinese, as reviewed in Section 2, the same character might have various pronunciations contingent upon the context. Taken together with the path of the morphological evolution presented in Figure 1, the development of characters’ phonetic forms can roughly be represented in Table 2.

On Stage 1, the pronunciations of each character might contain various consonants in the beginning (prefixes), middle (infixes), or end (suffixes), rendering the boundaries between characters unclear. From Stage 1 to Stage 2, as affixes were eliminated, consonant clusters and most coda stops (except for /n/, /m/, /ŋ/, /p/, /t/ and /k/^[3]) disappeared from Chinese syllables, so that the pronunciation of each character became more clearly bounded. For example, a sequence that definitely cannot occur in Modern Mandarin, i.e., /lukliɑŋ/, was legal on Stage 1, representing two syllables, and the boundary between syllables could be between /k/ and /l/, or between /u/ and /k/.

From Stage 2 to Stage 4, new characters were created to take over some of the variants in the alternations, such as 智 zhì previously mentioned in Section 2.6. Besides, the majority of morphological operations were lost, leaving each character with only one pronunciation. For example, the character 饮 can only be pronounced as yǐn in Modern Mandarin, meaning ‘to drink’, but the variant yìn ‘to make … drink’ no longer exists: a particular causative verb 使 shǐ/叫 jiào/让 ràng ‘to make’ is needed to register the causative sense. Nonetheless, there are still a few characters (about 10 % of the top 3,500 high-frequency characters, see Zhang and Qin 2016) preserving multiple pronunciations with distinct meanings, such as 数 (shǔ ‘to count’ vs. shù ‘number’) and 好 (hǎo ‘nice; good’ vs. hào ‘to like’) previously mentioned in Section 2.6.

Overall, morphology has been evolving in a way that stabilizes the pronunciation of each character: the association between the graphic form and the phonetic form has constantly been strengthening. In this process, consonants and vowels were fixed before tones. In other words, the morphology of the language has been adapting to the characteristics of the ideographic writing system.

The stabilizing function of the ideographic Chinese characters is also observed from the synchronic perspective. Modern Mandarin and other Chinese dialects have similar semantic units and grammar rules. It is previously mentioned that in the Han dynasty, people in different areas used diverse words to call ‘the tiger’, but at the present time 虎 is the common semantic unit used everywhere in China. However, the phonological differences among dialects are still dramatic. For example, the character 虎 is pronounced as hǔ [xu²¹⁴] in Mandarin, [hoʊ⁵²] in Suzhounese, and [fu²] in Cantonese. Evidently, everything that is firmly encoded in the Chinese characters has been effectively unified, including the graphic-semantic association and the arrangement of characters (grammar rules other than morphology, mainly word order and the use of function words), but things that are not precisely recorded by the ideographic characters have always been subject to change, particularly the phonetic forms.

4 The effect of writing on language evolution

Besides Chinese, a general correspondence can be observed between writing systems and the languages chronically using them, especially if we set the scope on the situation before the 18th century when colonization made the Latin alphabet available to virtually all languages in the world.

Based on the linguistic unit that each sign corresponds to, writing systems in the world fall into four major categories: (1) ideographic writing wherein each sign is meaningful; (2) syllabary wherein each sign corresponds to a syllable; (3) abjad wherein each sign corresponds to a consonant; (4) alphabet wherein each sign corresponds to a phoneme. The features of the languages chronically using them can be summarized in Table 3.

To illustrate the above generalizations, transitive-intransitive verb pairs in different languages are exemplified in (11)–(13):

As previously mentioned in Section 2, transitive-intransitive verb pairs involved voiceless-voiced alternation and tonal alternation in Old Chinese, which are suggested to be descendants from affixation, but most of these operations are extinct in Modern Mandarin: a great number of verbs can be used transitively and intransitively in the same form in Modern Mandarin (Zhang 2019). Other languages tend to be richer in valence-changing morphology. The alphabet-using Gothic derives intransitive (anticausative) verbs from transitive verbs by adding a single consonant -n-, as shown in (11). The abjad-using Persian derives transitive (causative) verbs from intransitive verbs by changing the vowel, while the preceding consonant – the onset of the syllable – remains unchanged, as shown in (12), characterizing a typical inflective language. In the syllabary-using Japanese, no single phoneme has the function to derive verbs (single phonemes cannot be morphemes) – the same stem is conjugated with different affixes to form the transitive-intransitive counterparts, as shown in (13) – an “equipollent” system is employed (Haspelmath 1993).

Besides, the linguistic units that are clearly represented by written signs tend to be relatively stable across regional varieties of a language; while internal variations occur mainly in the linguistic units that are not clearly represented by written signs. For the languages using ideographic writings, the ideographic signs derived from pictures stabilize the basic semantic units of the language across all its varieties, pronounced as clearly bounded syllables, but the exact phonetic values are not clearly represented by the writing, so a considerable degree of phonetic variation is allowed. For the languages using syllabaries, the structures of syllables tend to be simple, and morphemes must be combinations of syllables instead of single phonemes. For the languages using abjads, consonants tend to be more stable than vowels across different regional varieties, as only consonants are clearly represented in the writing. For the languages using alphabets, morphemes can take the form of single phonemes or the combinations of phonemes: the concept of syllables is not clear; internal variations occur mainly in prosodic features as every phoneme is clearly represented in the writing.

The correlation presented in Table 3 is normally interpreted as evidence for the secondary position of writing, as if a language picks the writing system that best fits it, but the traceable history of writing clearly shows that no language ever picked writing based on linguistic features. In fact, as illustrated below in Table 4 (Morgan 1905), a striking resemblance can be observed from the earliest writing systems in human history, although they were used for typologically remote languages.

Table 4:

A comparison of the earliest writings from different places (Morgan 1905).

It is generally agreed that writing started from pictures (e.g., DeFrancis 1984: 151; Gelb 1952/1963: 11–12). Pictures have meanings in their own right. When they developed an association with the spoken language, they could be called writing. In other words, the earliest writings were inherently ideographic, while their associations with the spoken languages were gradually developed. The resemblance presented in Table 4 is therefore easily explained by the origin of writing. Divergences occurred only when those earliest writing systems were borrowed by foreign peoples. The borrowing of writing implies the existence of a language in need of a writing system. In this scenario, the borrowed writing is secondary to the language. Here came a problem that the borrowing of the ideographic signs could be cumbersome due to the large number of signs. A convenient solution to this problem was to borrow some signs only for the phonetic value, deprived of their meanings. For example, the character 以 yǐ originally meant ‘to use’ in Chinese, but when Japanese borrowed it, its cursive form い was used merely to represent the syllable yi, deprived of any meaning itself. In this way, the number of signs was reduced through phonetization, which further increases the borrowability of the writing system. The emergence of self-originated writings and the phonetization of borrowed writings are illustrated in Figure 3.

Figure 3:

The emergence and phonetization of writing.

If a writing system was borrowed and adapted many times, the ideographic nature would keep decreasing until the signs became completely phonetic. Importantly, the complete loss of ideographic signs only happened in borrowings: in history, self-originated writing systems did not automatically evolve to syllabaries, as Gelb (1952/1963: 165, 196) puts it:

Owing largely to the secluded geographical environment protected by Tibet to the west and the Pacific Ocean to the east, Sinosphere did not have much contact with other civilizations before the Age of Discovery, so the borrowings of writing systems only happened on a limited scale. Self-originated ideographic writings and the derived syllabaries managed to survive. In contrast, the histories of Europe and the Near East have witnessed the rises and falls of numerous peoples. From Egyptian hieroglyphs and/or the Sumerian cuneiform script, writing systems kept being borrowed and adapted, resulting in abjads and alphabets that completely replaced the original ideographic writings. As fully phonetized writing systems, alphabets and abjads are even more borrowable than syllabaries with fewer signs: there are always more syllables than phonemes in a language. Clearly, the evolution of writing was triggered mainly by borrowings, but not by conscious linguistic analysis. The essential motivation for the developments in writing was to facilitate borrowing: writing has been going with cultural contacts, but not the linguistic features of languages. Therefore, the correlation presented in Table 3 can only be explained by a coincidence or an effect of writing on language. Given the history of the Chinese language presented in Sections 1 and 2, wherein the language has been constantly adapting to the ideographic writing system, a hypothesis can hereby be posited:

Support for the hypothesis can not only be drawn from Chinese, but also be captured in the histories of many languages. Up until the conquest by Greeks in the 4th century BC, Ancient Egypt had been using the self-originated writing system of hieroglyphs, and meanwhile, the language had been evolving in a way similar to the history of Chinese: with the elimination of some suffixes and coda stops, e.g., the coda -r and the feminine marker -t, the boundaries between syllables became clearer, making the language increasingly analytic (Bendjaballah and Reintges 2009; Loprieno and Müller 2012). Both Vietnamese and Korean used Chinese characters for more than a millennium, and there is convincing evidence showing that consonant clusters were lost from these two languages during this period (Nghieu 2019; Park 2009; Tamura 1980). Persian and Arabic provide examples representing the languages using abjads. As an Indo-European language, unlike European languages, Persian has been using various types of abjad scripts for the recent 2,300 years. Accordingly, rich variations can be observed from vowels but not consonants. The diphthongs /ai (aj)/ and /au (aw)/ in Old Persian became /ei (ej)/ and /ou (ow)/ in Modern Persian, but the variety spoken in Afghanistan preserves the old forms (Miller 2013; Rahbar 2008). In Arabic, the diphthongs /au (aw)/ and /ai (aj)/ in Old Arabic are only maintained in Malta and some areas in Tunisia today. In other varieties, they have become /e:/ and /o:/, and in Maghreb dialects have further merged with /i:/ and /u:/.

In the evolution of writing, the association between the graphic form and the spoken language has generally been strengthening. In contexts where there were large-scale cultural contacts, this association was developed mainly by the phonetization of writing happened in borrowings and adaptations; but when there was little or no cultural contact, the language would adapt to the writing system: in this way, the writing system determines the evolutionary path of the language.

5 Concluding remarks

By reviewing the evolutionary history of Chinese morphology, coupled with the evidence from typologically remote languages, this paper calls for an attention to the part that writing plays in language evolution. It is hypothesized that single signs of a writing system stabilize the basic units of the language using it. Languages never picked writing systems based on linguistic features. Instead, features of the writing system affect the evolutionary path of the language.

A common counterargument for the hypothesis of writing determinism points to the large number of undocumented languages, but in fact, nobody can provide the exact number of undocumented languages (Chambers and Trudgill 1980: 3–4; Croft 2000: 13–20, 2003: 26). Stable borders can hardly be drawn between languages without recourse to writing systems: it is exactly writings that help to demarcate and stabilize the borders of languages. Without writing systems, languages may change so fast that persons of the present generation may have difficulty to converse with people three or four generations older (Gelb 1952/1963: 223–224).

Meanwhile, there are a few points that must be clarified. In the first place, it is still admitted that writing and speech are two separate symbolic systems, and that speech appeared before writing in human history-in the history of human evolution, it was easier to control the sound than to leave visual symbols with sophisticated forms-but this does not necessarily mean writing is secondary to speech. Secondly, the effects of writing on language evolution do not preclude the effects of other factors. For example, the simplification of English morphology is widely believed to be related to language contact (e.g., Givón 1979; Kusters 2003: 9; Trudgill 1998; Whinnom 1980), in which the writing system obviously did not play the major part. Accordingly, the hypothesis of writing determinism cannot be interpreted the other way around: it is impossible to infer whether writings existed for a particular language based on its linguistic features. For example, syllables in the Hawaiian language unexceptionally have a CV(V) structure-the boundaries between syllables are as clear as in Japanese-but this cannot be taken as evidence for the existence of a Hawaiian syllabary.

The implications of the present study extend to typological studies by connecting the multiple strata of language evolution: the effects of writing systems explain many puzzles in linguistic typology. Some languages are evidently affiliated sharing a considerable number of cognates, but morphological features appear to be quite different, such as Chinese and Tibetan; while some languages are not directly affiliated yet present similar morphological features, such as Chinese and Vietnamese. This mismatch can hardly be explained without recognizing the role of writing on language evolution. Therefore, two strata of language evolution can be recognized, i.e., the pre-writing evolution that lays the foundation of lexicon and syntax, and the post-writing evolution that stabilizes phonology and morphology. Furthermore, the genetic affiliations of some languages, including Japanese and Korean, have been contentious. Such contentions can also be accounted for by the present theory as genetically affiliated languages can take divergent evolutionary paths when different types of writing are adopted. Along this line, further research is called for to explore the interaction of writing and other factors in language evolution, to analyze the social and psychological mechanisms underlying the changes related to writing, and to test the present theory in various contexts.