1. The lexicographical tradition
True Chinese grammar appears to have developed natively only under Western influence in the late 19th century, after the opening caused by the humiliation of the Opium Wars.
Two historical facts about Chinese predisposed indigenous grammar to late and slow development. First, the traditional logographic script conceals sub-syllabic morphological changes. Second, whatever morphology existed in early times ceased long ago to be productive. The result is that both speech and script display overwhelmingly isolating typology. We might expect native grammar to be syntactic in the main, but in fact early Chinese interest in language was rhetorical (emphasizing epistemology and propriety) and lexicographical.
Governments beginning with the Han dynasty (208 B.C.E. - 220 C.E.) justified their political authority by reference to Classical Chinese texts, and a kind of exegetical bureaucracy developed. 1st century scholiasts leave us the first consistent remarks on grammatical particles. Called ci "expressions" or yuzhu "language-helpers", they were said to "have sound externally, the meaning being internal" (Xu Shen, 30 C.E. - 124). "Internal" evidently means not apparent except in context. This is the earliest intimation of the classic dichotomy between shizi "full words" and xuzi "empty words" or particles. Texts such as the Mozi (5th-4th century B.C.E.) and the Gongyang zhuan (3rd century B.C.E.?) contain dozens of glosses on common particles. Syntax is never treated, but the sense of many individual particles is explained in context. By the medieval period, scholars such as Liu Xie (c. 466 C.E. -c. 539), Yan Shigu (581-645), and Liu Zongyuan (773-819) were attempting to group particles into simple classes by usage or meaning. Specialized studies of the particles appeared late, beginning with the didactic Zhuyuci "expressions helping language" of Lu Yiwei (fl. 1324) and reached a high level of development in the Manchu period (1644-1911). The study of Classical Chinese in Asia continues this lexicographic tradition, whereas in the West a syntactic approach has become dominant.
2. Syntactic awareness
Beginning with the influential scholiast Kong Yingda (574-648), we see close attention paid to the principles of breaking unpunctuated texts into sentences, a subject called judou "sentences and pauses". This may be the first true syntactic study in Chinese history.
However, syntactic parallelism has been basic to Chinese literature since high antiquity. From no later than Mongol times (1279-1368) there survive primers of poetics that provide models (without commentary) for parallelistic composition, but practical instruction must have vastly more ancient roots. Only in Song times (13th c.) do we see literary critics using the terminology of "full" and "empty" to comment explicitly on the syntactic element in parallel composition. Consider the comment of Ye Daqing (fl. 1205) about a couplet by Wang Wei (d. 761), where
Wei4 Qing1 bu2 bai4 || you2 Tian1 xing4
"That Wei Qing was not beaten || is because Heaven favored him;"
is composed in parallel with
Li3 Guang1 wu2 gong1 || yuan2 shu4 ji1
"that Li Guang was without merit || is due to 'his numbers coming up odd' " (he had bad luck).
The two lines are syntactically parallel; "Heaven" and "number" are both nouns. But Ye feels shu ji "numbers being odd" should be read shuo ji "several times uneven", and notes that in that case shuo, an empty word, cannot be parallel to Tian, a full one. Though mistaken about the reading, Ye is clearly aware of the syntactic element in parallelism.
Also of apparent Song date is mention of a systematic distinction between dong "active" and jing "quiescent" or "inactive" words, which we understand to mean something like transitive and intransitive verbs. (Intransitive verbs in Chinese include adjectives.) Still another matter of Song interest was the doublet readings that many modern scholars believe embody the traces of ancient morphological differences. Philologists had always known that Han exegetes had assigned different pronunciations to the same character in different senses. A Northern Song scholar Jia Changchao (998-1065) now collected a large number of examples of these and arranged them with the primary meaning first and the derived meaning second. The Southern Song scholar Huang Zhen (1213-1280) then tried to systematize the relationship between the readings of verbs in terms of their "active" and "inactive" qualities. The Song thus clearly saw the growth of a kind of grammatical awareness, a movement away from the Chinese traditional lexicographical bias.
The validity of morphological doublets was harshly attacked by the iconoclast Gu Yanwu (1613-1682) and some others, but most modern scholars, Chinese and Western , now accept them. In the 1990's, morphology became the focus of a trend in early Chinese reconstruction by Western scholars such as Laurent Sagart, William Baxter, Wolfgang Behr, and Edwin Pulleyblank. But the paucity of data suggests that our knowledge will remain incomplete until genuinely new sources are found.
3. Western influence
Some Chinese speakers in traditional times were certainly exposed to languages with non-trivial morphology and highly developed pedagogy. Sanskrit was being studied by the 4th c., and Latin was taught in the 16th century. The isolating nature of Chinese impeded formal study on these models. Christian missionaries, however, began producing descriptive grammars of Chinese from the 17th century onward. Preëminent is the Arte de la lengua mandarina (1703) of the Dominican father Francisco Varo (1627-1687). Although he explains the language in terms of Latin grammar, Varo notes that the Latin system of cases and conjugations does not fit Chinese linguistic reality: "no syllable by itself can be fully considered as a real case, or as truly a part of speech;" "just as the nouns are indeclinable, so the verbs are also inconjugable, since the syllable is invariable (in form)." Translations of common Latin and Portuguese words form a substantial part of the book. But unlike Chinese lexicographers, Varo always mentions word order explicitly.
Varo's influence on the next 150 years of generations of Western missionary linguists was enormous, and unrivalled until the 1881 Chinesische Grammatik of Georg von der Gabelentz (1840-1893), which remains quite usable today. Yale's George Kennedy (1901-1960) and linguists of the U.S. Army Language School also made great advances in practical grammar of modern and Classical Chinese in the 1940's, promoting the "sentence pattern" rather than the particle as the basic grammatical unit. The German Walter Simon (1893-1981) did comparable work on a smaller scale in Europe, and Yuen Ren Chao (1892-1982) of Harvard and Berkeley produced two major descriptive studies of spoken grammar. Beginning in the 1950's, Angus Graham and later John Cikoski deepened our understanding of aspect in Classical Chinese. This Western tradition remains poorly known in China.
Within China itself, however, the first Western -style grammar was that of Ma Jianzhong (1844-1900) and his brother Ma Liang (1840-1938), the 1898 Mashi wentong. The Ma brothers, Catholics with a strong interest in adopting Western science, introduced Chinese terminology for parts of speech based on Latin, and cited Classical passages extensively to document short statements about syntax. The book was revolutionary and remains a primary work. Two of the most important later contributions have been Zhongguo wenfa yaolüe (1942) of Lü Shuxiang (1904-1998), which treats Classical Chinese and Mandarin together in one system, and Zhongguo xiandai yufa (1943) of Wang Li (1900-1986), which uses late pre-modern vernacular fiction as a corpus for Mandarin grammar study.
Since the Communist revolution of 1949, much work has been done on linguistics of all kinds, often under official auspices. Dialect grammar came under scrutiny in the 1980's and 90's, and the fields of historical grammar and pedagogy continue to develop at a greatly accelerating pace.
From the Oxford Encyclopedia of Linguistics. This article © by David Prager Branner. Used by permission of the author.