Branner2003 - On Early Chinese Morphology and Its Intellectual History

On Early Chinese Morphology and Its Intellectual History.
Winner of the Barwis–Holliday

Award for 2001
Author(s): David Prager Branner
Source: Journal of the Royal Asiatic Society, Third Series, Vol. 13, No. 1 (Apr., 2003), pp. 45-
76
Published by: Cambridge University Press on behalf of the Royal Asiatic Society of Great Britain
and Ireland
Stable URL: http://www.jstor.org/stable/25188330 .
Accessed: 28/06/2014 13:42
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Cambridge University Press and Royal Asiatic Society of Great Britain and Ireland are collaborating with
JSTOR to digitize, preserve and extend access to Journal of the Royal Asiatic Society.
http://www.jstor.org
This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

All use subject to JSTOR Terms and Conditions
On Early Chinese Morphology and its
IntellectualHistory1
Winner of the Barwis-Holliday Award for 2001
DAVID PRAGER BRANNER
The medieval Chinese tradition tells us that a given Chinese character may change its meaning
when its reading is altered Modern scholars have for these

slightly. sought principles changes,
and from those have reconstructed a skeletal of Chinese
principles system early morphology
-
with such elements as derivation tone causative infixes,
by change, transitivising prefixes,
etc. Yet it is an arresting fact that some of China's most astute
pre-modern linguistically
scholars the multiple on which this research is based. seem
inveighed against readings They
to have held strong not made about how it is that
opinions, always explicit, precisely
Chinese characters represent These two views, modern and traditional,
language. represent
fundamentally different models of how early Chinese evolved into modern Chinese.
i. Introduction
-
This paper deals with some aspects of the of morphology in Chinese
question early
with its intellectual and practical in general concerns the
history application.2 Morphology
rules of word-formation, especially inflection and derivation. these processes are
Although
not considered in Chinese on any scale, a number of morphological
usually present large
functions have been posited for early Chinese and incorporated into reconstructions. Laurent
Sagart's important Roots of Old Chinese (1999) is a recent effort to assemble evidence for it.
Reconstructed morphology is of varied kinds, but the best attested form is that represented
by variant readings in the medieval phonological tradition. I hold the view that our chief
received sources on medieval Chinese the 'rime-books' of the "fetfiljt

phonology, Qieyun
tradition and the 'rime-tables' of the Yunjing iijl^ft tradition, embody a fundamentally
conservative and artificial literary ideal, and not the actual of any real time or
speech place.
The rationale for this viewpoint is discussed in Norman and Coblin (1995) and my own
set down in Branner (2000, pp. 147-174). Note that the medieval tradition is the
thoughts
earliest whole system we have for any type of Chinese; reconstructed
phonological early
Chinese is, conceptually, derived in large part from the medieval system, with the addition
1
An earlier version of this paper was read 17 January, 2000, at the University of Hong Kong, and a still earlier
version as "Did Early Chinese Really Have Morphology"? (delivered 1November, 1998, Annual Meeting of the
Western Branch of the American Oriental Society, Seattle). My thanks to South Coblin, Victor Mair, Tsu-Lin Mei,
Martin Kern, Margaret Chu, and Thomas Bartlett. A portion of the writing of this paper was done with support
from Victor Mair and the University of Pennsylvania.
I use the term "early Chinese" to refer to what is also called "Old Chinese" or "Archaic Chinese", because
those terms seem to suggest a clearly defined linguistic entity. In fact, early Chinese is imprecisely defined, and is
reconstructed using materials of greatly varying dates.
JRAS, Series 3, 13, 1 (2003), pp. 45-76 ? David Prager Branner 2003
DOI: 10.1017/S1356186302003000 Printed in the United Kingdom

46 David Prager Branner
of data from character structure, and other sources. in the discussion

rhyming, Consequently,
that follows, I illustrate in the main using medieval clothed
phonological points phonology,
in the direct system in Branner 1999b. forms are
transcription presented (Medieval always
placed in curly brackets {})

Below, I offer two well documented of pairs of medieval variant that
examples readings
are understood to derive from related words in early Chinese. The first are
morphologically
of a noun from a verb, in which the tone to
examples derived changes qusheng -??%? from
something other than the qusheng:
verb in non-qusheng noun in qusheng
& chu {tshyuQ3b} 'to dwell at' fa chu {tshyuH3y} 'place'

frzhi {tri3b} 'to know' &zhi {triH3b} 'knowledge' (=#)
fochen {dren3b} 'to set out, arrange' fazhen {drenH3b} 'battle formation' (=j$)
% cheng {zyeng3} 'to ride' % sheng {zyengH3} 'carriage with team of horses'
fachudn {druan3b} 'to transmit' fyzhuan {druanH3y} 'a record'
&shu {sruQ3c} 'to count' $ksnu {sruH3c} 'number'
Qusheng is indicated inmedieval transcription by the letter H at the end of a syllable. It is

some that this H may be a survival of a suffix, hence an
thought by nominalizing example
of whose trace survives in a tonal distinction.
early morphology only
When we read classical Chinese, it is conventional to pay attention to these variant
readings, and in traditional sishu Â^ education one of the teacher's chief responsibilities
was to train in when to read a character in its "basic" and when
pupils given pronunciation
to read it in its "changed" For instance, consider the line
pronunciation.
i^JU-#**?*#*>
is to make one's home in choosing, one does

[The finest thing among good people; if,
not dwell among the good, how can one gain knowledge?] (Analects 4: 1)
We are to read, inMandarin, // ren wei mei; ze bu chu ren, yan de zhi. The boldfaced
supposed
words chu 'to dwell and zhi 'knowledge'
are
special readings, and we must not read
among'
them *chu 'place' and *zhi 'to know' if we are to understand the passage grammatically.
In
medieval transcription this passage is nyen3b ghwi3y dreik2a pwet3a tshyuQjy nyen3y
{HQ3d miQ3C;
an3a tekt and the Jingdian shlwen entry for this passage duly supplies sound-glosses
triHjy},
on these two words:
> = rather than the usual

fa % &? {tshyuQ3y} (i.e., reading {tshyuH3b})
%, . -1^ = {triHjb} (i.e., rather than the usual reading [tri3b])
The in this particular passage are uncontroversial and known to all literate
readings people
in Chinese, some others of this type are far more recondite.
although
second set of consists of pairs of verbs, in which a stative or "inactive"
My examples (jing
is derived from a transitive or "active" sense, when a voiceless initial
jj$) meaning (dong ffj)
to voiced:
changes

On Early Chinese Morphology and its IntellectualHistory 47
active verb with voiceless initial inactive verb with voiced initial
& bai {peiH2C} 'to defeat' & bai {beiH2C} 'to be defeated'
#] hie {pat3b} 'to separate (tr.), distinguish' %\hie {bat3y} 'different, to depart'
? zhuo {trak3} 'to place on; to wear' ? zhu {drak3} 'to be attached to'
^jidn {kan4} 'to see' ^ xian {ghan4} 'to have audience,
be seen' (=^)
#f ji# {keiQ2a} 'to unite' fy xie {gheiQ2a} 'to be released,
relaxed' (as in xieddi %& 'sluggish')
# jl {keiH4} 'to tie' # xi {gheiH4} 'to be connected to'
<?kudi {kweiHit,} 'to bring together' ^ /imi{ghweiHjt,} 'to come together'
Here there is a small discrepancy between medieval and early phonology: {b} is the voiced
form of voiceless {p} in both systems, and {d} is the voiced form of voiceless {t}. Medieval
or laryngeal fricative [y] or [?] in the time of the
{gh} (phonetically probably a voiced velar
Qieyun) is believed to derive from a voiced stop [g] in early times, and it is that voiced g that
to the voiceless in the medieval forms shown.
corresponds {k}
Now consider the line
[The well-bred person assembles friends culture, and nurtures through

through goodness
(Analects 10: 24).
friendships.]
We customarily read this line junzi yi wen hut you, yi youfu ren ({kwen3a tsiQ3d yiQ3d mwen3a
kweiHib ghouQ3b, yiQ3d ghouQ3b buoQ3c nyen3y}). The meaning of^ is clearly a transitive
verb 'to bring because it takes ^ 'friend' as direct and so to
together' object, according
the received tradition it should be read kudi But for some reason the Mandarin
{kweiHjy}.
reading kudi for^ is now associated only with the word kudiji^%\ 'accounting' (literally,
"to assemble and Kudi as a for the character in the sense of "to
tally up"). reading literary ^
assemble" has out of modern
dropped reading practice.
The content of this paper is twofold. First, I review the of these variant
background
and introduce two modern views of them, one native to China and one the product
readings
of the western-Chinese synthesis in recent times. Second, I consider the evidence for and
against these two recent views and consider the place of variant and reconstructed
readings
morphology in the modern study of and research in classical Chinese.
2. The conservative tradition of variant

lexicographic readings
Chinese variant readings have been transmitted since antiquity in the native lexicographic
tradition. I see three main in the evolution of that tradition.
phases
The earliest are found in exegetic commentaries on classical texts. This
examples high
first phase was in full bloom by Eastern Han (25-220), and is typified by works such as the
Mao shijidn -^f^H of Zheng Xuan Jtp"ST (127-200). Zheng Xuan's commentary consists
of notes on the whole classical text, and both variants
running sound-glosses, embracing
and more usual readings,
are included. But phonological glosses occupy a very small part of
what is primarily devoted to discussion of content and meaning.

continue to appear in numbers in commentaries

Original phonological glosses large
at least times. But a second of the native tradition was attained
through Tang phase glossing
in compendia of variant readings, which flowered in the Six Dynasties period (222-589).
It seems likely that the growth of Buddhism favoured the Chinese interest in phonology,
even when the texts were non-Buddhist. far the best known of
being glossed By exemplar
this type is the Jingdian shiwenM^f-^SC of Lu Deming F^^tyf (c. 550-630). The Shiwen,
which was before 589, is a dense collection of earlier semantic and
completed phonological
without an index. It differs from the commentaries of Zheng Xuan's in that it
glosses, type
does not attempt to subordinate its glosses to a full and corrected version of the text itself.
Rather, it is intended to be used as an to the texts, and is arranged

adjunct original following
their order. Here commentary seems to be raised up at the expense of text; have
glosses
become the body of the work, like an entire meal made up of condiments.
But the Shiwen assembles thousands of from various sources, it

although sound-glosses
does not attempt to make an of the variant in those
orderly interpretation readings glosses.
The actual of the seems to have been undertaken rather later, mainly
interpretation readings
in the Song and after. For example, theJiyun %.i% (completed 1039) incorporated many of
the Shiwen s wholesale into the framework of the Qieyun,
glosses phonological generally
an textual passage for each unusual variant
adding exemplary reading.
More systematic interpretation of variant readings is found in the Qunjing yinbian
^&J?"il"$f: of Jia Changchao ^ %$ft (998-1065). The Yinbian seems to be based in large
part on Shiwen material, but it is organized as of individual characters, and indicates
study
which of a character are to be considered and which derived. The
readings given primary
concept of "derivation" is not explicitly stated, but Jia Changchao clearly presents what he
considers the main first and the derived reading second. Derivation is more clearly
reading
evident in brief comments of Huang Zhen ;|f Jt (1213-1280), found in his Hudngshi richao
Jt"J^<9 $y (Zheng and Mai 1964, p. 195). Huang attempts to assert a general principle
for to the is said to have a "inactive"
relating meanings readings: primary reading jing j&
and the derived a This contrast has a
meaning reading dong ^"active" meaning. dong-jing
long history in Chinese philosophy, but its importance in grammatical thinking is explicitly
attested only since Song times.
s work has been influential, and seems to have served as the

Jia Changchao immensely
basis for similar down until the of the period of native-western
presentations beginning
in Chinese around the end of the Manchu For instance, the
synthesis linguistics, period.
section "Dongzi bianyin $)]^fJf -?"" inMajianzhong's 1898Mdshi wentong J%J^ 3C ij?, owes
most of its material to Jia's book.
Overall, the native Chinese tradition of variant was conservative. Its

readings quite
exponents seemingly felt unable either to discard the tradition or to develop it beyond
and the Most never asked how
collating arranging examples. significantly, they apparently
the tradition related to the of China. That line of was taken up, in
language early inquiry
different ways, by two different groups of iconoclasts: the kdozhengîzi philologists of the
Manchu period, and westerners from the early missionary period onward.
Below I deal first with the western tradition, because its assumptions are more
widely
accepted today, and then with the Manchu-time philologists.

3. Western reconstructionism based on variant readings
Western has been concerned with the nature of Chinese and the ways
sinology long early
in which it differed from other of the world. Indeed, even before knew
languages they
about China, a about has been characteristic of European
larger curiosity foreign things
and other Mediterranean civilizations. Certainly from the time of Herodotus (c. 485
post 425 BC, the beginning of theWarring States period in China), western intellectuals
have been fascinated with describing and comparing the many different cultures and
encountered. Mediterranean civilization has understood itself to be

languages they long
a of cultures, some newer, some older, and many of them literate. The
neighbourhood
whole context of the Odyssey a world in which different cultures were accustomed
implies
to each other. This cultural memory of the meeting and mingling of peoples
encountering
is borne out even in records the modern world has retrieved from bronze age Minoan
Crete and Mycenae. It is very different from classical China's view of its own in
place
the world, even if it turns out that bronze age China was also a of many
meeting place
cultures.
The watershed in the of Chinese historical was the

European practice linguistics
of to the materials of the Chinese tradition,
application comparative-historical linguistics
most associated with Bernhard (1889-1978) work after World War I.
famously Karlgren's
But the history of the Western reconstruction of early Chinese actually
morphology predates
Karlgren; I see four stages in its

development.3
The
first stage, which I call the "metaphysical" view, I shall dispose of briefly. It
was exemplified by such different personalities as Gottfried Leibniz (1646-1716), Etienne
Fourmont (168 3-1745), and J. P. Abel-Remusat (1788-18 32). The feature of Chinese that
attracted the most attention in this period was the writing system, which seemed to many
savants to exist of any and even to represent a pure

independently spoken language, perhaps
and abstract akin to a mathematical notation of pure ideas. This
"philosophical language"
idealistic (and factually baseless) view was firmly debunked by Pierre (Peter) du Ponceau
(1760?1844), heir to Franklin and as President of the American
Jefferson Philosophical
and a intellectual force in the Du Ponceau's 1838 book
Society major early republic. begins
with a useful resume of the of the metaphysical view, in a but
exponents ending long
resounding rebuttal. Du Ponceau holds that any true written must be

language necessarily
based on
speech.
More is the next stage, the view, to which Chinese
significant "typological" according
was seen as the consummate of The
representative "primitive" monosyllabicity. prime
of this view was the founder of modern the polymath Wilhelm
exponent linguistic typology,
von Humboldt Von Humboldt is concerned with as a token of
(1767?1835). language
?
that is, with the relation between the way a nation and its "national"
cognition speaks
mental characteristics. He was the first to characterize human as into three
language falling
distinct of which he saw Chinese as one of the He cited Chinese,
types, prime specimens.
with Burmese, as an extreme of the of in which
together example isolating type language,
morphology is fundamentally absent.
In researching this section I have benefited from reading van Driem (1979).

Von Humboldt must have had access to information about Chinese,

reasonably good
because he is by no means numb to the aesthetic power of the classical nor is he
language,
without to Chinese culture. But he describes Chinese as one essential
sensitivity lacking
refinement of an advanced
language:
Niemand kann laugnen, dass das Chinesische des alten Styls dadurch, dass lauter gewichtige
unmittelbar an einander treten, eine Wiirde mit sich fiihrt und dadurch eine
Begriffe ergreifende
einfache Grosse erhalt, dass es gleichsam, mit Abwerfung aller unniitzen nur
Nebenbeziehungen,
zum reinen Gedanken vermittelst der Sprache zu entfliehen scheint. (1903 [1836], p. 164)
can deny that the old a stirring to the fact that

[Nobody style Chinese reveals dignity owing
ideas upon each other; it reveals a simple because, by
important impinge directly grandeur
discarding all useless secondary designations, it seems to take recourse in depicting pure thought
via language.] (tr. 1971, p. 124)
Wenn man daher auch gern dass die Form der Chinesischen als vielleicht
zugesteht, Sprache mehr,
irgend eine andere die Kraft des reinen Gedanken herausstellt und die Seele, gerade weil sie alle
kleinen, storenden Verbindungslaute abschneidet, ausschliesslicher und gespannter auf denselben
hinrichtet, wenn die Lesung auch nur weniger Chinesischer Texte diese Ueberzeugung biz zur
so diirften doch auch die entschiedensten Vertheidiger dieser Sprache

Bewunderung steigert,
schwerlich dass sie die geistige zu dem wahren hinlenkt, aus
behaupten, Thatigkeit Mittlepunkt
dem Dichtung und Philosophic, wissenschaftliche Forschung und beredter Vortrag gleich willig
emporbluhen. (1903 [1836], pp. 255-256)
even we are willing to admit that the form of the Chinese tongue more than
[Hence, though
perhaps any other brings out the power of the pure idea and directs the soul toward it more
exclusively and precisely because it lops off all small disturbing connecting phonemes, and even
if reading of but a few Chinese texts increases this conviction to a state of admiration, the most
resolute defendants of this language could scarcely claim that it guides intellectual activity to the
true central point from which poetry, philosophy, scientific research, eloquent recitation blossom
forth.] (tr. 1971, p. 196)
Von Humboldt feels that Chinese is primitive because it has failed to develop in an important
way. because it lacks derivational in his view it is inadequate for
Specifically, morphology,
certain delicate mental I do not wish to dwell on von Humboldt's ethnic
processes. possible
which are intrinsic to his work and which have been commented on since his
prejudices,
own interest here is rather his view that Chinese is primitive because in
day.4 My lacking
morphology it has failed to develop something essential.

As influential as von Humboldt was, his that Chinese reflected a
deservedly assumption
primitive stage of linguistic monosyllabicity did not persist unaltered. Between his day and
a number of western scholars advanced the that Chinese, even
Karlgren's opinion though
it lacked derivational affixes, must have descended from a language that did display some
form of morphology. This third of development, the was
stage "morphological" viewpoint,
first ennunciated the astute Karl Below are
apparently by phonetician Lepsius (1810-1884).
some passages from his 1861 monograph:
4 was an early critic see Sweet (1980, pp. 403-406) but compare
Du Ponceau and Aarsleff (1988, pp. lxi-lxv),
the clear-headed remarks of Sweet (1989) and Koerner (2000, p. 9).

wenn
[...] Es ware denkbar, dass auch die Europaischen Sprachen, je die geistige Fortentwicklung
der Volker unterbrochen werden konnte und mit ihr die Quellen auch des leiblichen
Verjungungsprocesses der Sprache, [...] bald auch auf das geistige Niveau der Chinesischen
Sprache herabsinken konnten.
[It would be conceivable that the European languages could also sink to the mental level of
the Chinese language in short order, if ever the mental development of the peoples could be
interrupted and, along with it, the wellsprings of the processes of material rejuvenation of the
language [...]]
Der Zeitpunkt in welchem eine Sprache schriftfahig wird, und das Volk eine Litteratur erhalt,
der entscheidendste fur die Richtung seiner zu sein,
pflegt Wendepunkt Sprachentwickelung
und da wir fast alle Sprachen erst seit dieser Zeit naher kennen lernen, so bleibt uns in der
die erste und wichtigste Halfte ihres Lebens, die des leiblichen wachsthums unbekannt.
Regel
Die Litteratur halt diese in ihrer lebenskraftigsten Entwickelung auf, bringt sie zum Stillstand,
dann zum
Riickgang.
[The moment at which it becomes possible for a language to be written and the people receives a
literature tends to be the most decisive turning point in the direction of its linguistic development.
And since we first become more with almost all languages
intimately acquainted only from this
? ?
point, the first and most important half of their life that of their material growth remains as
a rule unknown to us. Literature arrests this growth at the stage of itsmost vigorous development,
brings it to a halt, and then into decline.]
Die Chinesische Einsilbigkeit ist nicht die ursprungliche, sondern eine bereits von fruherer
Mehrsilbigkeit und in verharteter Einseitigkeit an der Grenze ihrer Entwickelung

herabgesunkene
angelangte.
[Chinese monosyllabicity is not the original monosyllabicity, but rather one which deteriorated
from an earlier and which arrived at the limit of its development in a state of
polysyllabicity
obdurate partiality.]
At first may seem to hold a contemptuous view of Chinese-speaking

glance Lepsius people,
but in fact he is arguing for the malleability and fundamental of all human
equivalence
Chinese its not he holds, nor are western
language. immutably monosyllabic, languages
derivational. An von Humboldt's is
immutably important development beyond position
his claim that Chinese must have become after a
monosyllabic only having passed through
stage, and hence it has lost something. Chinese is secondary, not

polysyllabic monosyllabicity
primary in the
history of world languages. Chinese is not primitive, but advanced, he feels:
it has in such a way as to lack necessary, as a lose

developed something just species might
a trait that had evolved earlier. A similar view was in the 1881 of Wilhelm
expressed essay
Grube (18 5 5-1908), which Karlgren knew.
apparently
and Grube held views of Chinese more accurate and than Leibniz or
Lepsius sophisticated
von Humboldt. But it strikes me that common to all is the conviction
deeply Indo-European
that morphology is something essential. For Chinese not to exhibit is, therefore,
morphology
a defect, and it must have been to to correct the defect, to restore or recover
tempting try
the missing morphology.

The idea of lost morphology in Chinese arose on at least two

identifying apparently
separate occasions. The earlier one was due to the Otto

general linguist Jespersen
He seems to have been the first person to propose that variant could
(i860-1943). readings
be treated as evidence for a now-lost system of derivational In his
reliquary morphology.
1894 book he cites the following examples:
[__] xvang 'king' wang 'to become king'

[%] lao 'work' lao 'to pay the work'
[&] cong 'to follow' zbng 'follower' zong 'footsteps'

[ft] hdo 'good' hao 'to love'
[%] shou 'to acquire' [^] shbu 'to give'

[^] mdi 'to buy [jr] mai 'to sell'
and continues:
[...] I see no reason why we should not set forth the provisional hypothesis that the above
were or flexional
mentioned pairs of Chinese words formerly distinguished by derivative syllables
and the like, which have now without leaving any traces behind them
endings disappeared,
except in the tones. This hypothesis is perhaps rendered more by what seems to be an
probable
?
established fact that one of the five tones, at least in the Nan-king pronunciation, has arisen
the dropping of final consonants (p, t, k).

through
That is not far from what we believe Less than a decade after Maurice
today.5 Jespersen,
Courant (1865-1935) proposed a similar principle (1903):
L'histoire dela langue montrera sans doute que, dans la plupart des cas, cette polyphonie vient,
soit de la confusion en un seul de caracteres primitivement differents, soit de l'extension prise
par une dialectale, soit de quelques autres causes, il faut retenir la

pronunciation parmi lesquelles
suivante: variation a une modalite d'un sens premier et rappelant la
phonetique correspondant
flexion usitee dans d'autres ^ syak (so) "cuiller"; *j cyak (co) "puiser avec une cuiller".
langues:
[The history of the will undoubtedly show that, in the majority of the cases, this
language
comes from confusion over one of the characters which would originally
polyphony possibly
have been different, or from the evolution of a dialectal or from
independent pronunciation,
various other causes, among which it is necessary to remember the following: phonetic variation
to amodulation of the primary meaning and suggesting the derivational inflection
corresponding
used in other *j syak (so) "spoon"; *j cyak (co) "to stir with a
languages: spoon".]
and Courant the first concrete attempts to connect variant readings

Jespersen represent
with lost morphology, in the fourth stage of western reconstructionism, in
ushering
which the actual phonetics of ancient language is hypothesized and ancient morphology
It is in that stage that we live and work today. Although details vary, the
incorporated.
has been in most reconstructions made in the West
morphological principle implemented
or in the period of western-Chinese since the time of Henri Maspero (1883?1945),
synthesis
5
The reference to Nanjing i$j^ dialect has to do with the fact that Nanjing preserves the medieval rushing A^
tone category, in which all words anciently ended in -p, -t, or -k. In modern Nanjing dialect, these oral stops are
lost, and Jespersen's point is that the Nanjing rusheng category could be seen as having been produced by the loss of
those oral stops. But that is not sound thinking. One problem is that rushengwords inNanjing are still distinguished
by a glottal stop [?],meaning that the ancient stop ending has not completely disappeared. A second problem is that
the Nanjing rusheng is fundamentally the same contrastive phonological category as the ancient rusheng, meaning
that in fact no new tone has arisen. Itwas nevertheless an impressive effort for a non-sinologist.

On Early Chinese Morphology and its IntellectualHistory 5 3
and it is equally important in most Sino-Tibetan work because it establish cognates and
helps
likeness between Chinese Tibeto-Burman. there are some dissidents,
typological Although
among traditionally educated scholars in China, this is now the modern majority
especially
view in Chinese historical lingustics.
At its best, reconstruction allows parsimonious explanations of large collections of data.
(It should be remembered, without prejudice, that by "explanation" linguists sometimes
mean "economical The dozens of of verbs

simply representation".) examples qusheng
to nouns, mentioned on p. 46, are as evidence
corresponding non-qusheng today interpreted
of a lost suffix -5which forms nouns from the suffix -s in is considered to
verbs; early Chinese
erode in such a way as to the tone category in medieval
produce qusheng phonology. Again,
the many of verbs mentioned on p. 47, in which an inactive is associated
pairs meaning
with a voiceless initial but an active with voiced initial, are as
meaning today interpreted
evidence of a lost fi- which forms inactive verbs from active verbs while
prefix producing
the effect of initial These two affixes are to be related to
voicing. particular widely thought
Tibeto-Burman and so constitute an of evidence
morphological processes, important piece
for the Sino-Tibetan (see particularly Bodman, 1980 and Mei, 1980).
hypothesis
The medieval tradition of our for the variant
sound-glosses, prime authority readings
on which is reconstructed, contains numbers of these in
early morphology great readings,
considerable diversity. Perhaps the most attractive promise of morphological research is that it
many of these them to have been variants

might explain readings by showing morphological
of a word, or members of a "word At however, that
single single family". present, promise
has not yet materialized.
If there is one that epitomizes this tradition most it is the West's
thing starkly, long
fascination with the absence of morphology in Chinese. Many western
nineteenth-century
intellectuals declared the Chinese to be because it lacked
language inadequately expressive
and while this view would seem to be at odds with the earlier idea of the
morphology,
as a universal both views build on the that the lack of derivational
script language, premise
processes is the most distinctive trait of Chinese. The more recent movement to reconstruct
lost morphology assumes, after all, that morphology is something necessary to a In

language.
this respect, it appears to be felt missing in Chinese. It is ironic
"restoring" something long
that one of the best known of reconstructed Chinese s
examples morphology, Karlgren
claim to have discovered a kind of ablaut in the Chinese
early pronoun system (1920),
a of what should look
betrays characteristically Indo-European conception morphology
like evidence was undermined in That
(Karlgren's irretrievably by George Kennedy 1956).
should remind us that any without derivation seems to
example language unnaturally plain
many in the West.
4. The Chinese purist school
What I call the Chinese purist school ismainly associated with the kdozheng^"ift philologists
of the Manchu Its hallmark is an opposition to the tradition of variant that
period.6 readings
came down from medieval scholiasts and were consecrated in 'rime-books' and standard
commentaries. Its ideal conception of Chinese is that one character has

writing only
6
In researching this section, I have consulted the superb collection of relevant materials in Zheng Dian and Mai
Meiqiao (1964, pp. 103-105, pp. 168-199).

one hence I term it purist. this movement was active in

reading, Although primarily
the seventeenth century and afterward, one of its was the
important predecessors early
medieval moralist Yan ZhltuI Jlf-^dj-- (531?591?), whose Ydn's Family Instructions (Ydnshi
jiaxun >$KJ^^l^l) contains an important essay on proper pronunciation, the "Yinci -?"1$^"
I deal with Yan first and turn to the Manchu-time scholars afterwards.
chapter.
Yan advocates great of the lessons he wishes to
generally philological sensitivity. Many
on the reader are illustrated with anecdotes from literature. his book,
impress Throughout
written to his sons in regulating their households, there runs an
ostensibly guide undisguised
river of pride in his of literate culture, and he does not miss
deep knowledge opportunities
to out the of scholars and officials from all over the Chinese world. Yan
point ignorance
lived in both north and south at a time when they were different countries with complex
ethnic mixtures and loyalties, and left us precious if sparse comments on the different literary
worlds of the two cultural centres in sixth century China.
Yet it is curious that in spite of his love of philological precision Yan ignores most of the
we use in
variant readings morphological study. He apparently approves only of the variants
hao and hao "to like" for the character and the parallel forms e "bad" and wu "to
"good" if,
hate" for two as
,^|.. These pairs of readings he regards having exegetic legitimacy, although
that are understood
he laments they poorly by northerners.
(i960, pp. I23a-b)

^j^
are naturally either fine or coarse; fineness and coarseness are called
Generally speaking, things
,% {akt} and "bad". minds either or accept
^ {hauQ3} "good" People's reject things; rejecting
and accepting are called fy "to like" and ,%{uoHt} "to hate". These are seen
{hauH}3 readings
in the glosses of Ge Hong $ #t and Xu Mao %j&. The scholars of the North read the Shangshu
passage "to love

living things and hate killing" as
sreng3 akt srat2a\ "[to love living things
{hauH3
and bad lolling]". This is an example of using the expression for a thing in one case, and using
the expression for a feeling in the other case. It is far from making sense.
Other than the cases of %} and he of variant among others

,?, disapproves readings, citing
the case of bai jj?, whose differ as inactive vs. active verb with voiceless vs.
specific readings
voiced initials on p. 4,
(shown above):
> - 5M J&tfIfc - ^T*A^ &&i%&? .*

J**#dHtil4* crftftX^ ?)4/lfti
(i960, pp. 124D-125C)

H^:
When scholars of the south read the Zubzhuan, they pass their traditions down orally, and make
their own general rules. "When an army is defeated of itself, it is called {beiH2c} [the ordinary
reading], and when they defeat someone else's army it is pronounced I have never seen
{peiH2c}".
this reading in any of the commentaries. Even in Xu Xianmin's edition of the Zuozhuan there
is only a single place with this pronunciation, and there, moreover, he does not talk about the
difference between an army "being defeated" and "defeating someone else". This is hair-splitting.

On Early Chinese Morphology and its IntellectualHistory 5 5
It is significant that Yan was active at the end of the period of greatest activity,
sound-glossing
he most of the alternate current at that time as the
yet regards readings spurious.7 Clearly
of the variant tradition is not we should take for it
authenticity reading something granted;
was
already challenged in its own time.
being
The greatest partisan of the purist school was the independent scholar Gu Yanwu $&& 3\
Gu doubted the existence not only of variant but actually of all tonal
(1613-1682). readings
distinctions in ancient times.8 He attributed variants to the of
morphological ignorance
medieval scholiasts after the Classical and seems to have believed that Chinese in its
period,
earliest form was a in which every character had a and distinctive
pristine language single
reading, a dingyin /^.if* (1966c [1667], 4/2a, on the Ode "Xiaorong 'JN-$<"of the "Qinfeng
In a way this recalls the view of the western admirers of the
-I^JSl")- "metaphysical" early
Chinese Gu was, in any case, no He was a fanatical
script (p. 49, above). ordinary philologist.
opponent of the Manchu government, who advocated what Thomas Bartlett has called
"Confucian fundamentalism" a clear in Gu's messianic

(1985). Phonology occupied place
vision:
The fact that Heaven has not abandoned us means that a Sage will surely arise again; he will raise
up our modern and return them to those of pure antiquity.

pronunciations
Gu cited most of his evidence from and from the tradition. His
rhyming glossing
chief statement on this subject is his short essay "Xianru liangsheng geyi zhi shuo bujin
ran iL^^^-S"^^^*^ sL$& [the theory of former scholars, that readings with two
different tones each have their own is not true]" He
meanings, necessarily (1966b[1667]).
by contradictory examples of characters with readings in early

begins offering multiple
poetry:
%&iL^m -***** - &#f*# -?A.^lS-^1%*^**^ >*?&* -?&
All characters in the shang, qu, or ru tones have two or three or even four readings. They can shift
around and into each other and even into the ping tone. People
turn in ancient times called this
zhudnzhu [change and confluence]. The way a word is actually used when you are face to face
with the writing is sometimes hazy and sometimes precise; it lies in an unconstrained state. And
so the scholars of former times said that when one character has two readings, each has ameaning.
For example, the character g. when it means 'to hate' is read in the qusheng, {uoHt}; when it
means 'bad' it is read in the rusheng, The says these readings in the
{akj} Ydnshijidxun originated
time of Ge Hong and Xu Mo, that is, since the Jin and Song [265-479].
now
Gu cites five examples from early poetry, in three of which ,% means "bad" yet must
be read in order to rather than and in two of which it means "to hate"
{uoHt rhyme, {akj},
7
Note that Zhou Zumo (1966b, pp. 425-426) contradicts Yan's assertion about bai |?, but his evidence may
postdate Yan. Modern philologists consider Yan's ordinary reading {beiH2c) to be the derived reading.
His theory of tones, which I do not detail here, is ennunciated in his essay "Guren sisheng yiguan -?"A?E9???"5T
[for people in antiquity the four tones were all interconnected]". (i966a[i667]).

56David Prager Branner
and yet must be read in order to rather than That is, actual
{ak,} rhyme, {uoH,}. early
rhyming practice contradicts the tonal assignments of the medieval scholiasts. He continues:
-*A*-?Lt -
il*!**??#*-**fi-?-*-##- ****** fcHBl***
So we know that the distinction between qusheng and rusheng

was more than between
nothing
and heavy pronunciation, and not a distinction as between separate of land.
light sharp pieces
Among characters that have two pronunciations in books, a great many are of this kind. Itwould
be hard to list them one one.
by
From the time when arose and Classical declined, when 'rime-books' were
glossing learning
current and old-style poetry was put aside, petty disputation has more and more, while
spread
the Great Way has disappeared day by day. Alas! many are the subtle words of the former sages
that have become confused in the mouths and ears of school-teachers! One can generalize from
these cases, and for that I look ahead to some Gentleman of the future.
"Gentleman of the future" is the none other than the Confucian messiah who will
(This
restore classical with true
pronunciation, together ancient-style government.)
Gu goes on to attack another commentator name:
by
- -******* - *#*# -?*?*4:*H*

##*-H.Jl#* 4îL^,?X#
&* - j*^* -
jfcJ_L#3?4'ft^#^
-
?***
- *f
???****.
The "Shij) zhengyi lunyin li" by Zhang Shoujie of the Tang says,
In quality there is fineness and coarseness, which we call -jtf- {hauQ3} and
"good" ,% {ak^
"bad"; the mind has loving and detesting, which we call -?J- {hauH3} "to like" and ,?. uoHj} "to
hate". If it involves appropriate form, it is%,%t~ "reputation", but if it involves

{meing3b yuoH3y}
moodiness, it is S?;* "condemnation and praise".9
{huHQ3b yuo3b}
If today we look for evidence in the Shijing, we find that the poems "Riyue" from Bei,
"Mugua" from Wei, and "Nu yue jiming" from Zheng all rhyme ^J-with ^ {pauHj},
these are cases in the "to love with the heart". In the
qusheng meaning "Hongfan" chapter
>
of the Shujing are the [rhyming] lines. "&^f'ffc^J {hauQ,} ifki-îf. {dauQj} [having no
personal likings, pursue the kingly path]"; this is in the shangsheng, meaning "to love with
the heart". As for the character ^, it appears three times in the Shijing: in "^^JL^ [I only
wish you happiness and joy]" of the poem "Juxia j?^" and "*4c$HS^ [for his longevity
the people praise him]" of the poem "Zhenlu #.*" it is in the qusheng. Only in the line.
"
$? ?# jUk^ [Hanji was overjoyed]" of the poem "Hanyi $?^?" is it in the pingsheng; how can
9
Gu is quoting from the "Shiji zhengyi lunyin h ^^LJlj&ifr'l*^" of the Tang scholiast Zhang Shoujie ^^frp
(preface dated 736; see I975[i959], p. 15). Zhang's lines apparently quote the preface to the Jingdian shiwen,
themselves clearly reminiscent of Yan Zhitui's remarks, cited above.

anyone call this "moodiness"? the classics to like this is what

Reading according principles
is known as "not illuminated, gotten stuck in the muck"10. Here the
being having again
commentator is claiming a consistent between sound and meaning in the
Tang relationship
variants of -?f- and ^, but Gu shows that that claim is contradicted by the rhyming
readings
evidence of classical texts.
After Yan Zhitui's entire note about the of J& in the Zuozhuan
quoting readings
Gu goes on to cite on a classical passage different
(above), contradictory glosses given by
commentators, that these medieval of sound to were not
showing assignments meaning
even in the pre-modern the examples he cites are the
universally agreed upon period. Among
readings {kwant} "coffin" vs. {kwanH^ "to encoffm", {ghwangH3} "king" vs. {ghwangH3}
"to crown and "to observe" vs. "to cause to observe".
king", {kwan}} {kwanH}}
SM -***** -
********
In the "Gao di ji" of the Han s/jm, it says "the counties supplied clothes and quilts, coffins and
burial implements". Ru Chun comments: ffi "coffin" is pronounced like * but

{kwanHf\,
[Yan] says, "At first they made coarse body-boxes, but when [the bodies] reached their
Shigu
home counties, they supplied fresh clothing and coffins, and prepared their burial objects. There
is no need to go to the trouble of changing the reading to "f|~
1}" .n
{kwanH
it says "Xiang Yu went back on his agreement and declared himself our ruler is
Again, king;
at Nan Zheng". [Yan] Shigu says, "The first X is read Liu Ban [of Song
king {ghwangH3}"}2
times] says, "I say, what iswrong with .?_ being treated according to its usual
reading"?13
on two there are two a

Based these items, where readings for single character with superfluous
and twisted [we can see that] among in the past there were
expositions explanations, people
already those who denied [these cases], having realized the truth.
Gu then turns the question of the need for dual and dual
glosses readings:
Sim - J***********.* -
*?*U*#
In the fifth year of Duke Zhao, in the Zuozhuan, it says "[He] displayed the troops on the hill
at Diji. The Jingdidn shiwen says, "jfc was formerly read {kwan1},\ but in reading the Eryd it is
10
Alluding to the xiangzhuan commentary to the fourth line of hexagram Zhen ;?.
11
Han shu 1962:65.
12
Han shii 1962:30.
13
Han shii 1955:40 (1 j_:28/b).

always {kwanHj}. In the Zhou li, in the title siguan *J# Zheng [Xuan] reads it as in "-f-g$i*X
[I am like a flame on
display]"; thus H, is qusheng.14
The Song scholar Wei Liaoweng [i 178-1237] said of the Yijing hexagram "Guan |fe
[Observation]",
"Here, on account of the principle of and confluence', in the text proper and the
'change
commentary $1 is re-d as in guanshi 'to exhibit for observation',
symbolic {kwanHt} [=guan]
while in the six line-texts it is read as inguanzhan 'to observe from a
{kwant} [=guan] height'.
What Iwonder is, before the four tones zn&fdnqie existed, how did people know these did not
have the same pronunciation"?
Indeed, if you examine this matter with respect to meaning, the two words can be treated as one.
If you consider this question with respect to the period from the Yijing and Shijing to the Eastern
Han, all rhyming passages were, after the time of Sun Yan and Shen Yue, always constrained
tones cannot as being
within the four and pinned down by fdnqie glosses. You speak of them of
the same moment in time.
In addition to
introducing evidence from classical rhyming and attacking early medieval
commentaries, Gu claims that in Tang many alternate coexist, without a
poetry readings
semantic distinction:
* ^ *
*?*A##_J|t* + * *ML-^ *>%%%*%? ****** *****
Let me cite one or two from among the strictest verse of the Tang:
just examples regulated
$fa is read sometimes in the ptngsheng and sometimes in the qu, % sometimes in the ping and
sometimes in the qu, j? and ^ sometimes ping and sometimes qu, iff sometimes ping and
? can you say these are cases where are two different meanings?
sometimes and there This is
shang
precisely what in the study of ancient scripts is called zhudnzhu [change and confluence]. The fact
that the rime-books admit such words in two or three different places indicates that the poets of
the world used these words differently depending

on whether they were spoken slowly or quickly,
or heavily.
lightly
In all, Gu has three main to the variant apart from polemic. First, the
objections readings,
assignment of the readings {woH,} and {ak^ to ,% in classical rhyming texts does not always
match its expected and there are similar cases such as and *, from which
meanings, ij-
he would have us the Second, in some cases scholiasts appear to be at
generalize problem.
variance over which of two is appropriate to a character, as in the cases of
readings given
and Ut. Third, some characters for which alternate seem to have no semantic
^ readings
distinction between the alternates, and both may occur in strictly poetry.
regulated Tang
His is that there were no tones in ancient times, different ways
explanation actually merely
to pronounce words. Those different ways, he feels, were down in an artificial way
pinned
by the medieval scholiasts.
14
I cannot find a sound-gloss by Zheng Xuan for j$, in the Zhouli zhushu (1980:831 j^, 843 f>). The line
appears in the Shujing "Pan'geng _?_",where it may also be understood to mean "as for me, it is like
"^?11,^"
gazing at a flame".

Gu was not the Manchu-time savant to attack the received tradition of variant
only
and to claim that Chinese had no tones. The idea to be
readings, early begins expressed
a number of voices after his time. For the textual critic Mao jl?
by example, Qiling ^?"?q
(1623-1716), in his Yunxue zhiydo fj|^^?f ^ (1991), and the poet and bon-vivant Yuan Mei
;|t$C (1716-1798), in his "Yinyi fanchong "if*^"^"^", in Suiyudn suibi^M^^f (i993),
both assert that dual to appear with the of 'rime-books'
readings only began compilation
after the Qi ffi and Liang ^ periods (479?557), and that semantic glosses were distinguished
tomatch the diverging phonetic glosses. Yuan Renlin 3$l4~fa (fl. c. 1700), in his Xuzi shuo
Jm.^rilL (Zheng and Mai 1964: 197) attempts to distinguish between characters that have
variant but not variant and characters for which the glosses vary
meanings readings, together
with sound.
from unornamented statements of opinion, there also appear a few other of

Apart pieces
textual evidence to boost Gu's. Here are remarks by Qian Daxin I^Tv^T (1728-1804):
> *.
[...] ##@#S*t# *T*A**P ?#&** (Q^n 1927-35: i/3a-b)
When people wrote glosses in antiquity, they put much information into the Each
readings.
character had meaning; there was not, at first, a distinction between "empty" and "full" words
or "active" and "inactive" verbs. such as %$ and ,% each different
Examples having meanings
started in Ge Honge's Before the Han there was no such distinction.
Ziyucin.
[... ] Scholars follow what has gone before them as if in a stupor. This is what is known as
"trusting the teaching of the most recent authority, and denying those who delve into the past".
now cites Wei

Qian Liaoweng, mentioned above by Gu Yanwu (see p. 58), and continues:
- *) >*b?T? - - - st^vst >?**? -?

* ?*# *#??# &**&
(Qian 1927-35: i/3b)
This man can be said to have

my very In the Ddxue, for the
really "anticipated thoughts".
character zhi fe
["well-governed"] as in "the state iswell governed", Lu [in the
Deming Jingdidn
shiwen] reads it {driH3a}; but in "first govern the state", as there is no reading given it is supposed
to be read pingsheng, as is really ridiculous. [In the same text, the famous pairs]
{dri3a}. This
"balance the household" and "the household is balanced",
"cultivate and "you are cultivated",

yourself
"make upright your mind" is upright",
and "your mind
"make honest
your thoughts" are honest",
and "your
thoughts
"categorize the things of the world" and "the things are categorized"
?
in all these cases, I have never heard of there being two readings. Only for ?? "well-governed"
is there a distinction. If you don't read the whole isn't it almost the same as "not being
paragraph,
able to distinguish beans from barley"?15
15
The allusion is to the story of the mentally deficient elder brother of Sun Zhou j$ J$, described in the Zuozhuan
?4% under the 18th year of Duke Cheng $,. Here Qian means only that the two "readings" are indistinguishable
to us.

60 David Branner
Prager
The relevant passage from the opening of the Daxue A^ is one of the most striking
in classical literature of the active and inactive forms of a series of verbs used
examples
in close juxtaposition. To Qian, it must have been all the more conspicuous because of
its prominent in children's education. Of the six verbs involved, the first, }&,
place only
has separate readings for both its active and inactive forms: {dri3d} (inMandarin, chi) and
{driH3d} (Mandarin zh\). If the alternation of chi 2nd zhi is legitimate, asksQian, then where
are the parallel for the five verbs? If they do exist, have not made
examples remaining they
their way into the commentaries. (I return to the
problem of restoring the exegetic tradition
on this passage; see p. 68, below.)

A different kind of evidence was introduced by the textual critic Lu Wenchao >?f_3C 3?
(1717-1796). In a note entitled "Ziyi bu sui yin qubie ^4$^$1"?~Ig$] [themeanings of
characters are not Lu cites evidence from some
distinguished by pronunciation]" interesting
early glossing traditions, including Erya tJf^- and Boyd "rr^, to the effect that the authors
of those books sometimes seem unaware that a character has variant with
given readings
different meanings (Lu Wenchao 1985: 1).
His first is from the Erya, one of the earliest
example surviving glossaries (traditionally
said to have been in existence in Confucius' which contains a line
time!),
>
c?M^-^ h^l ^^L (seeErya jiaozhu 1984: 9-10)
That is to say, the six characters ^?, Jj?, 7c, ^% Y, and PJy, appearing in early texts, may
defined as not a or it can be read
be "*J*". This -f* does represent single word reading: yu
to mean T, me' or yu to mean 'to give'. Lu has noticed that in three of
{yuo3b} {yuoQ3b}
the six cases in the Erya's entry, 'f* is defining words meaning T, me' and in the other three,
words meaning 'to give':
y(4r 'I^nie' (appears in the Shujing)
zhen J|? T~me' (Shujing)

yang f% T~me' (Guo Pu |p^ cites the Lu Shi'%H% and says it is a regional word)
lai$~ 'to give as a gift to an inferior' (Shujing)

bi^ 'to give' (Shijing)
bu I" 'to give' (Shijing)
The is that the of the Erya does not seem to have minded that
point compiler -J* represents
two different words with different pronunciations.
Lu finds a similar case in the Boy a (or Guangya ^^), a later work modelled after the
Erya:
N
i?#?fr*?L^Nf?f Jt& (seeGuangya gulin 1998: 192)
The seven characters i?, ^, vfr, ?^, ipr, ?f, and ?f, appearing in early texts, are defined as "jfc".
However, this % does not a word or It can be read and

represent single reading. {keingQ3a}
have the sense 'border' (thisword iswritten Jt?today), or it can be read {keingHsa} and has
the sense 'to come to an end':
sui l? 'ditch between fields' (appears in the Zhou Li ffl ;$ )

=
jiang ? Jgr 'boundary' (Zhou Li)

=
jie *fc ^ 'boundary,border' (Shijing)
gang ?^ 'border' (Shuowen, said to be a regionalism)
bi^ 'to be done, finished' (Zuozhuan)

= an end'
zhong $- ?f 'to die, come to (Shijing)
zu f$ = 4s* 'to die' (Shuowen)
the person who the Boyd did not seem to mind one character
Again, compiled gloss using
in two different senses, to which the medieval tradition two different
assigns pronunciations.
Lu summarizes his conclusions this way:
Before the four tones were distinguished, people did not make the ping-ze distinction in their
poetry, but used the two interchangeably. If it sometimes happened that meaning did
always
not match sound, people in later times, always suspecting jidjie, have not realized that meaning
fundamentally does not change with pronunciation. What sort of jidjie is that?
Lu's evidence was supplemented by Wang Yun jLl| (1784-1854), best known for his
Shuowen studies. He left one note on the headed "Gu bufen ben
subject, sisheng, gai
wu N the four tones were not
sisheng ye ^^^^3^ Jk/^^^-Sp^jL [in antiquity
and this was because there were no four
distinguished, probably originally tones]" (1985:
20). His evidence and enlarges Lu Wenchao's, and is taken from the Eryd. Below
duplicates
I tabulate his additional more than above:
examples concisely
=
fan she gub yi ke jie 'to defeat'
{syengH3}
fl& #*&&&* M * ?
jidozhu1984:8-9)
Jfc& (Eryd
gong jian kdn {syeng3} 'to bear (responsibility)'
xian jdo jido ={ghanH4} 'apparent, obvious'
M m ft M ft MJJL(EryajitozMi9H:iS)
?
jin di 'to have audience before'
{ghanH4}
In the case of the words glossed with |L>Wang notes, "I^ÎJH^ tytfl >
? *k$$Jk&
[although they are both read are two different
{ghanH4} they certainly meanings]".
chun chu = 'to make'

jian {tsuoH^
* #t 19*4:17)
JR 4$JL(Eryajiaozhu
}f M &
bo
ydo
th ?
di li = 'to be stirred up'
dbng {tsak3}
fei she = 'to discard'

{syeQ3}
? j& &&
*t (Erydjidozhu 1984: 18-19)
shut = {syeH3} 'rest house'
The remainder of Wang's note claims that scholia on cannot be

early pronunciation
related to the tonal distinctions of the received tradition. He cites the Han commentator He
Xiu's i*$fa (129-182) comments on a passage in the Gongydng zhuan 4^^rT^ for the 28th
year of Duke Zhuang $?:

#?**A##
WKii :ft A*** - >

#ft&?4: #A*&
:Aft4f"J&j_. " ^
^^>i l|#*L"t'4L #A#& (Gongydngzhuan 1992: 9/ia)
Chunqiu text: The people of Qi *|/a [sent an expedition against]" Wei.
Gongydng zhuan: In the Chunqiu, fa refers to the recipient.

He Xiii: The one who sends the expedition against others is the recipient [of provocation].
We read^j prolonging the sound. This is expressed from [the point of view of] the people
ofQi.
zhuan: Fa refers to the active party.
Gongydng
He Xiii: The one against whom the expedition is sent is the party active [in provoking the
other]. We read fa shortening the sound. This is expressed from [the point of view of] the
people of Qi.
The comments do not seem is meant to have two

Gongyangs comprehensible unless^ i\
different that are not apparent in the written text. But, Yun notes,
readings Wang
although he follows the Gongydng zhudns double mention of fa, [He Xiu] makes this explanation
without any basis for distinguishing them.
He concludes,
in antiquity of the "five we cannot convert this into our

Just because they spoke sounds", simply
four tones. So too, all the characters collected in the rime-books: if you examine them according
to their component structures, most of them do not fall into neat categories.
The purist view of early Chinese may be epitomized by Gu Yanwu's idealized conception,
in which each character has one and one For a hundred and years
reading meaning. fifty
after his death, and intellectuals averred that early Chinese had lacked
progressive philologists
tones, and that the received medieval tradition of variant was flawed.
readings hopelessly
But the study of early Chinese phonology in the period between Gu and Duan Yucai
fî^, (173 5-1815) was sorely limited by the tiny amount of data that could be put to
research. Duan's achievement was to realize that the phonetic
evidentiary ground-breaking
components of all
xingshengtf} op characters could also be taken as evidence,
opening up most
of the to direct research. Duan's allowed scholars to see
dictionary phonological discovery
that there was a much closer between the and than between the
relationship qusheng rusheng
other tones, and that led to new and less radical theories of Chinese tone. In fact, even
early
well before Duan's time, the movement away from Gu's austere model is already evident
in the thinking of Jiang Yong ?X^C (1681-1762), who wrote that "/H^ltf ^^^ J?
[before the Han the four tones were not known]" (t966[i8io]: 4b). As Doong Jongsy
jj| #&&l has argued, Jiang Yong claimed not that there were no tones in antiquity, merely
that tones "were not known" i.e., had not yet been discovered and made use of by scholars

(Doongjongsy 1988, pp. 79?101). This claim has persisted in the native tradition since that
time.
We should not assume that the purist movement was the creation of Gu's
entirely highly
distinctive beliefs. I have shown that Yan Zhitui, a millenium earlier, was almost
living
as More Gu was a native tradition in which
equally purist. significantly, anticipated by long
characters the same element were considered to be
xingsheng Jfy&p containing phonetic
-
semantically cognate the so-called youwen shuo jfcJCiJt, the by reference
"explanation
to the right side of the character" (see the seminal study of Shen Jianshi 1986(193 3]). The
earliest statement of this principle has been attributed to Yang Quan $? jjfc.(/?. 4th c?), from
whose Wulilun $}5%J?q the following passage is cited:16
- N
& &/? ??<: &^Ê7 3f &AEJff* (Tdipingyuldn 402: 4b)
In metal and stone it is called "firm"; in plants and trees it is called
{kan4}, {kenQ3y}, "tough";
in people, it is called "sageliness".
{ghan4},
These are understood to refer to the of the

cryptic phrases underlying meaning xiesheng
m&F element that the characters ?^, ^, and % share. The fundamental meaning and sound
of that element rise to the three distinct characters. Such a of an

give daughter conception
"root" form is related to the ideal of one character,
etymologically primary closely purist
one
reading.
The youwen shuo certainly outlived Gu's purist view of variant readings; Huang Yongwu
(1965) has collected hundreds of examples of this the use of this explanation in philological
notes into the twentieth century.
5. An experiment in the recovery of evidence
The arguments of the purists to for adherents of the "reconstructive"

point specific challenges
view to overcome. One is to search more for morphological variants, not
carefully merely
by combing the old dictionaries, but by actually texts. This cannot be stressed
reading point
enough.
As an I present below the text of the Ddxue cited Daxin
example, passage by Qian (p. 20,
above). It contains of a series of verbs which appear in both active and inactive which
usage,
are in boldface. The is: for how of these words can variant
printed question many readings
be found in the tradition, or reconstructed? feels that only the verbs zhi and
plausibly Qian
chi, both written >?, have any commentatory justification.
In the each line is printed first in characters,
presentation below, followed by Mandarin
medieval and then a translation. The verbs under
transcription, transcription, rough English
discussion are in boldface.
printed
1 * 4:L ? W W % * * T **,**
gu zhi yu ming ming de yu tian xid zhe xian chi qi gub
{kuoQi tsyi3d yuk3C rneing3a tneing3a tekj uox than4 ghaQ2 tsyaQ3 san4 driu gi3d kwekt}
In antiquity, those who wanted to make a in the world of their
shining example
bright virtue first bring order to their states;
Most modern references to this passage cite the Yiwen leiju H-:sc.^3?., but I do not find it there.

2 * * * #*,##?
yu chi qi guo zhe xian qi qi jid
[yuk3C dri3d gy kwek, tsyaQ3 san4 dzeiH4 gijd ka2)

those who wanted to bring order to their states first balanced their households;
3 * # * * * A, # * ?
yu qi qi jid zhe xian xiu qi shen
{yuk3c dzei4 gijd ka2 tsyaQ3 san4 sou3h gijd syenjb]

those who wanted to balance their households first cultivated their own selves;
4 ***** *. ? * ^
yu xiii qi shen zhe xian zheng qi xin
sou3b syen3b san4 tsyeingH3b sem3]

{yuk3c gi3d tsyaQ3 gi3h
those who wanted to cultivate their own selves first rectified their minds;
54&j? * .Cf * & 1& $r &

yu zheng qi xin zhe xian cheng qi yi
{yuk3c tsyeingH3h gijd sem3 tsyaQ3 san4 dzyeing3h gy iH3d]

those who wanted to rectify their minds first made honest their thoughts;
6 '& Ik * I * it It # ^
yu cheng qi yi zhe xian zhi qi zhi
[yukJC dzyeing3h gi3d iH3d tsyaQ3 san4 triH3c gijd triH3b]

those who wanted to make honest their first let wisdom be attained;
thoughts
7 U: *? i. & *b
zhi zhi zai ge wu
{triH3C triH3b dzeiQla keik2a mwet3a)
attaining wisdom resides in categorizing the things [of the world].
8 $7 te .fi ? 4* ?
wu ge er hbu zhi zhi
{mwet3a keik2a nyi3d ghouQ1 triH3y tsyiH3c]

After the things of the world are categorized, wisdom comes;
9 %> ? 1*7 & & IS,

zhi zhi er hbu yi cheng
{triH3b tsyiH3C nyijd ghouQt iH3d dzyeing3h\
after wisdom comes, one's thoughts become honest;
io -fe HJi i*7 J& _? ,v

er hbu xin
yi cheng zheng
{iH3d dzyeing3b nyi3d ghouQt sem3 tsyeingH3b]

after one's thoughts are honest, one's mind becomes upright;
ii ,c* SL ?*7 fc % fr
xin zheng er hbu shen xiu
{sem3 tsyeing H3b nyy ghouQ1 syenjb sou3b]

after one's mind is upright; one's person becomes cultivated;

12 # # A ? t- #
s/zen xt? er /tow jw qi
[syen3b sou3b nyijd ghouQt ka2 dzei4]

after one's person is cultivated, one's household becomes balanced;
13 % * * jg @ &
jw 41 er /tow ?wo zfet
{/ea2 dzet4 nyiy ghouQ1 kwekx driH3d)

after one's household is balanced the state becomes orderly;
14 8 >&?;&* T -f
?Wo zf? er /tow fwn xw jjwg
[kwekt driH3d nyi3d ghouQ1 than4 ghaQ2 being3a]

after the state is orderly, the world is at peace. "Daxue [The Great
(Liji ^tl?,, A^
Learning")
I identify and boldface twenty-one items in this passage, which can be reduced
pertinent
to the each of which embraces the active and inactive forms of a
following eight examples,
verb.
Example Lines Character Pinyin Medieval meaning

6 1. & zhi {triH3c} to allow to be attained
8 ? zhi {tsyiH3c} to arrive
1 2. }& chi {dri3d} to bring order to
13 j& zhi {driH3j} to become orderly
1 3. 9JJ ming to make a shining example
{wetngJa}
I aft ming {meing3a} bright
2 $ 4. qi {dzei^ to balance
12 ^ qi to become balanced
{dzei4}
3 5. % xiu [sou3b] to cultivate
II {? xiu {sou3y} to become cultivated
4 6. SL zheng {tsyeingH3b} to rectify

10 j? zheng to become upright
{tsyeingH3b}
5 7. -?? cheng {dzyeing3b} to make honest
9 i& c/tewg {dzyeing3b} to become honest
7 8. +S- ?? {keik2a} to categorize

^ 8ge {keik2a} to be categorized
There are two verbs for which the variant forms are well known in the
already quite
literature. Below I place the active verb forms on the left side of the page and the inactive
verb forms on the First

right. example:
>S (Jingdidnshiwen) ?& (Jingdidnshiwen)

'to put in order' 'in order'
{dri3d} [driH3d\
This is the solitary example admitted by the Jingdidn shiwen. The Mandarin reading chi for
the active form is now obsolete. it is the same as chi%% 'to administer,
Presumably morpheme

66 David Branner
Prager
Commentaries from the Shiwen onward instruct us to read cht qi guo H3 but
despatch'. }x*&
guo zhi HI >&.

Here is the second example:
^t (Gudngyun) J_ (Gudngyun)
{triH3c}[Baxter'sOC *trjits] {tsyiH^} [Baxters
OC *tjits]
'to cause to arrive' 'to arrive'
This alternation exhibits the causative -r- infix and

proposed by Pulleyblank championed
In fact, it is one of a very small number of attested of this
by Sagart (1993). clearly examples
infix. The forms are in Mandarin: zhi.
putative homophonous
I can propose another four cases that I think are
reasonably well justified. Here is the first
example:
$- (Gudngyun) $- (Gudngyun)
'to mix in proportion' 'balanced'
[dzeiH4] {dzei4}
Note that in the sense "to mixin proportion", the word {dzeiH4} (Mandarin ji, as in
ydojishi JSHfll^F "pharmacist") is usually written %\. Hence the text is perhaps to be read
jid qi %~yt but *ji qi jid yt&%*. However, placement of morphological H is the reverse
of the >? Mei's article proposes that some apparent
example. important (1980) examples
of derivation tone may have been late inventions; within the
by change analogical though
reconstructionist camp, his view recalls the claims of Gu Yanwu et al.
There are two received of jE, which is my second

readings example:
j? jE (Gudngyun)
jE Usyeing3b} 'to rectify' {tsyeingH3b}'upright'
The meaning 'to rectify' is traditionally associated with the reading {tsyeingH3b}. But this
word rhymes consistently as {tsyeing3b} in the Shijing (standing for the word now written
;j& "to carry out a i.e., 'corrective' attack"). The is still
punitive, military reading zheng
used today in the compound zhengyue, 'first lunar month', that is, the "rectified" month,
meaning the month at which the beginning of the new year is ritually recognized (recalling
the formula "ijE^j [the king rectified the month]" in the Spring and Autumn Annals). It
that the First of Qin was born on the first of the first lunar month
happens Emperor day
(zhengyue jE^) of the 48th year of King Zhao of Qin, and later scholia assert that he was
named after his birthdate (Shiji 1959: 223-224). The eighth century Sima Zhen ^J .87^
courtesy name, Zi -f* jE, contains the character in question) wrote that in Qin
(whose zheng
was called dudnyuesfiqf\ in taboo avoidance of the Emperor's
times the first lunar month
personal name Zheng jE (Shiji 1959: 766 Sudyin note on Qin entry).17
17
The only inconsistency is that another eighth century commentator, Zhang Shoujie $?^$5, asserts that the
true pronunciation of the Qin emperor's name is nevertheless zheng "The First Emperor, on account
{tsyeingH3b}.
of having been born at Zhao on the morning of the first day of the first lunar month, was named zheng $?.
Later, because itwas the First Emperor's taboo name, itwas read as zheng fc"'. There seems to be no doubt that
his name means 'first lunar month', which we now pronounce zheng; Zhang Shoujie's claim is that this modern
pronunciation ismerely the result of taboo avoidance and not an ancient reading. However, I discount his claim
because the meaning "to rectify" is clearly related to the word written ^t, and because taboo avoidance
{tsyeing3b},
is seen mainly in the substitution of graphs rather than the alteration of readings.

*
Hence our passage is perhaps to be read xin zheng ?\jj? but zheng qi xin JL* <<*
A third example is:
=
if- (Guangyun) % % (Guangyun)
{sou3y} [Baxters OC *slju] {souH3y [Baxter's OC *sljus]
'to cultivate, 'to become ripe (said of grain in the ear)'.
Hence perhaps xiu qi shenW$r% but shen *xiu %*&. Let me point out that it is usual
among to seek cognate words among characters in the same
practice philologists xiesheng
i$i$jz series (this is another aspect of the youwen shuo), but much less usual to look outside
the xiesheng series. If, however, we have confidence in our reconstructions, we should
really
not hesitate to rely on them in the practice of etymology. So the equating of ^ with the
qusheng-derivcd inactive form of \fy may be unfamiliar, but is quite sound phonologically.
The fourth example is
$t =
(Guangyun) fe $?.& (Guangyun)
{keik2a} [Baxter'sOC *krak] [lakj] [Baxter'sOC *grak]
'a standard' > 'to classify' 'to be classified'
The inactive forms I identify as possible correspondents of %? are, first, lub j$~ 'to fall' >
'to fall to [someone's domain]', 'dwelling place' (juluo $Hf%t)\ and, second, luo $&> 'to
It should be noted that the character ^ was written for the newer
encompass'. traditionally
character f?~, and was in that case read Hence our text should be wu
{lakj}. perhaps readme
&$} but wu *luo$7^

In the two cases I have not found candidates for the active
remaining acceptable
to the inactive verbs in the text. The first is
correspondents example
*{meingH } 'tomake bright' Bfj{meing } 'bright' (Guangyun)
The only character in the Guangyun to the is^,

corresponding reading {meingH3a} usually
"to command". I find no evidence of a usage "to make For now,
straightforward bright".
we must senses
leave both the active and inactive of 9^ to the reading meing3a. The second
example is
*{dzyeingHb} 'tomake honest' i&idzyeing b] 'honest'
There is no obvious form 'to make honest' attested in the sources I have
*{dzyeingHsb}
examined.
On balance, I conclude that the received medieval tradition preserves, or contains, signi
more evidence for morphology than Qian Daxin found in primary commentaries
ficantly
such as the Jingdian shiwen. As as a small amount of is
long etymological interpretation
allowed, five additional of verbs can be identified. But the received tradition
pairs apparently
does not preserve evidence to reconstruct forms we
enough morphological everywhere
would expect to see them. There remain two inactive verbs for which no active forms have
been found.

6. Contrasting the morphological and purist models
I have previously argued that, ifmorphology did once exist in Chinese, we have to assume
that a great typological change took place, and that essentially all of modern Chinese dates
from after the change (see Branner 2000, pp. 159-166). In another paper (2002) I have
discussed modern dialect evidence for the reconstruction in early Chinese, of morphology
concluding that itmust not have been present in the mainstream language of the lateWarring
States and Han. If that is so, our reconstructionist model of Chinese linguistic history looks
something like this:
genuinely pure
fuzzy
morphological type area isolating type
modern
^(individualN {'Common"^*}somewhat
times >^_Chinese ^Sanomalous
V^varieties)_^^
medieval I received \
f
times I I reading I
Jhevelopmen ts^r^
' primitve,
" I ^^^ to
/^^Vi.ev failed
^* "^^ ^f^
[antiquity] posited Chinese \ develop
f ^^____^J
\iflAe West^/ ^l written )
typological likeness ^*** ^^script^^
*^
to Tibeto-Burman
The purist model, in contrast, looks something like this:
"impure" !<<<J pure

types x& isolating type
modern wX/^restored ^^\

^(individual^N
timesV^_yarieties)__^
^^^^^^^2^^
medieval received \$y Chinese \

j f
times I reading Ke I written I
\taditionVwX Vscript /
vfr true, worthy
o?j
tradition
ignorance^R&
[antiquity] $8$i posited ^\
SgVby Gu et aliaJ *%*** b?
>\w ^*-_- ^^ writing system

both of these models cannot be accurate. Of the two, the

Obviously, simultaneously
model is probably somewhat sounder; there are serious with most
morphological problems
of the evidence cited by the purists.
Lu Wenchao and Wang Yun mean to show that ancient were not aware of
people
tones. The and are of a passage such as
Erya Gudngyd basically compendia early glosses;
> is not intended to equate the six characters izftk^t^T
"?$(^^ b^7 ^^" I"J%,
to group them As as the did not read
merely loosely together by gloss. long compilers always
aloud what they wrote, there is no reason
why they should not have confounded different
same The is clever

words written with the graph. material presented by Lu and interesting,
but to claim that tones were not or did not exist in antiquity, based on
distinguished solely
evidence of this kind, does not follow.
in rhyming are more

Tonal dissonances (cited by Gu Yanwu and Lu Wenchao) serious,
but reconstruction resolves many of them: ,? *ak > *aks > ~f* *yo
>
{ak}}, {uoHj}; {yuo3b},
*yo? > {yuoQ3y}. Although the two Mandarin and medieval readings of /?. (e and wu, {akj}
and {uoHt}) are quite dissimilar, it is known (from Duan Yucai's discovery) that rushing and
forms are connected in early Chinese. The Chinese reconstructions
qusheng frequently early
*ak > *aks > illustrate this closeness, allowing
us to
reinterpret the
{akt}, {uoHt} qusheng
form wu ~ {uoHj} as itself a type of rusheng) (*aks).When the two forms are spelled *ak
and *aks, does not seem so strange. were, after all, almost identical.
rhyming suddenly They
In the case of the tonal distinction between and has been
:f>, sharp pingsheng shangsheng
a In both then, the reconstructions of
replaced by simple glottal stop ending (?). examples,
the two forms are similar to be in rhyming, but
sufficiently interchangeable systematically
accomodate the variance that is characteristic of medieval and modern We also
readings.
now realize that ancient often treated tone more than in later
rhyming practice loosely
periods.
The last issue is Gu Yanwu s that medieval sometimes with
complaint exegetes disagreed
each other; Daxin went so far as to attack the to the Ddxue A
Qian glossing passage.
somewhat doctrinal reconstructionist answer to Qian would be that we have no reason
to assume that the medieval tradition all the evidence from the
preserved early period,
if morphology to an of
especially belonged essentially "pre-Chinese" period linguistic
As I have shown above more we can in fact find more evidence than he
history. concretely,
could.
I think it is plain that in our age the reconstructionist has the better support.
viewpoint
But the view has its own which we should not overlook: it describes
purist advantage,
Classical Chinese within the isolating typology of Chinese, as indeed it has been read for
many centuries.
On the level, means that Chinese does not

simplest being isolating systematically express
in the number or case of nouns, the of with nouns, or the
changes agreement adjectives
tense or mood of verbs. Verbs may be used as nouns without into a
changing special gerund
form. Beyond this, however, because most Chinese words do not alter their
phonology along
with in grammatical function, the very concept of the part of is formally
changes speech
indeterminate. Part of can be identified context.
speech only by
The reader of the classical has the sensation of manipulating words
language mentally,
noun to verb, verb to adverb, inactive to active verb, and so on as context demands.
turning

One may hold several words in mind, without them as to part of

tentatively fixing speech
until the whole meaning has become clear by gestalt.
way of illustration, below are two of that remain until

By examples phrases ambiguous
are to a
parts of speech assigned key word:
First possibility Second possibility
Adjust Noun VerbObject

tS?
i$J &i$J
southern face to turn southward one's face
This is perfectly Both are valid, the former is

phrase ambiguous. interpretations although
more common in ordinary modern But the verbal ndnmian means,
language. compound
idiomatically, "to be the ruler", deriving from the fact that ancient Chinese rulers always
held court south, with their backs to the north.18 The "to face south"
facing sitting meaning
survives today in synonymous phrases: ndnmian wei wdng i$Ji&,^ JL and ndnmian chenggu
i)w^#) but readers do not grasp the verbal sense of nan in ndnmian alone until
always
to consider it.
pressed
Another example, taken (out of context) from the Sunzi (Tongdidn 49:3807) is:
First possibility Second possibility
Number+Nj [poss] N2 Verb Object

-a 4: 4a - a^4?
one person s ears and eyes to focus [the attention of] ears and eyes
people's
The second is the correct meaning in the text, but the first
interpretation original
is more on first The reader must manage to see
interpretation natural-looking inspection.
the common numeral "one" as not a numeral but a verb, "to and then the meaning
unify",
of the whole phrase snaps into place.
called like nan and "ambs" because were ambivalent,
George Kennedy examples yi they
sometimes as verbs and sometimes as nouns (in
behaving (in accepting negation) serving
as to another noun; 1964: There are numbers of such words in the
adjunct 2>7?ff-)- large
Classical Chinese lexicon. For our purposes, what is most about "ambs" is that
interesting
for most of them no variant have come down to us in the received
special pronunciatons
tradition, to their parts of Indeed, among literate Chinese they
identify competing speech.19
are not of as distinct Skilled readers are able to
conventionally thought meanings. generally
relate many usages of a word to a fundamental semantic core, and there is a
disparate given
plain likeness between this semantic core and the purist phonological ideal that Gu calls the
dingyin ^."?* or "fixed reading". Although neurolinguists tell us that all language processing
involves of this kind, Chinese is distinctive in that there are no clues at all in the
manipulation
text and few or none in its sound when read aloud. This process of tentative manipulation
is a basic of the skill of to read the and, when mastered, a
part learning literary language
aesthetic I suspect that that aesthetic is the real, root
richly satisfying experience experience
inspiration that motivated the whole purist viewpoint.
18
Bei j_ "north" is an ancient loangraph whose original meaning was the word bei "the back~to turn the back
on", now written *j$.
19
In standard Mandarin, however, the second example might be distinguished tonally as yiren for "one person"
but yi ren... for "to focus the attention of people['s]... ".

Another kind of ambivalence is, among nouns, the whole relationship between subject,
object, and agent, which are unmarked in Classical Chinese, and must be determined by
word order and context. however, in the great majority of Tibeto-Burman
Interestingly,
there is extensive case marking, and their ancestor has been reconstructed as
languages,
to greater or lesser (see Bauman 1979, DeLancey 1990). Since Tibeto
ergative degrees
Burman and Chinese each have many that seem to show likeness with
morphemes family
the other, and since are believed to descend from a common ancestor, it is quite
they widely
interesting to consider that Chinese, isolating as it is, has linguistic kin that exhibit affixation
of considerable Did Chinese indeed descend from an inflected or from
complexity. language,
a with an If so, it appears to have been
language ergative case-marking system?20 stripped
of all its morphology at an so that it displays one of the major
early period, typological
characteristics of a Creole.21
Be that as it may, as we compare the reconstructionist and models of Chinese

purist
it is clear that the reconstructionist model will resolve many of the
linguistic history, today
s textual as well establish connections between Chinese and other
purist objections, plausible
Asian There is no any real doubt that the reconstructionist model, however
languages. longer
dubious its origins, has won out. But we should remember that the understood
purists
Chinese as a which is how we must treat the
profoundly isolating language, literary language
if we are to master it. Their was an to see that that could have
problem inability language
evolved from more as we can no avoid Gu Yanwu s
something complex, longer doing.
model still survives as a kind of ghost in many Chinese in the Chinese world.
departments
It is still common in the Chinese of Chinese, for to all variant
study early example, assign
of a to the same Chinese even if doing so violates
readings given graph early rime-group,
rules of regular phonological development.
7. Practical matters: How should we read Classical Chinese?
Apart from answering the objections of the purists, the other challenge for the reconstructivist
to meet is to find a accurate way to read and teach classical Chinese. There
philologically
are several options.
(a) We can read to the values of the school. That would mean
according purist trying
to use a for each character. Consider that if a great
only single reading typological change
did take before the medieval then the purist model describes not the
place period, perhaps
reading principles of high antiquity but, effectively, the tendency of the newer, isolating type
of Chinese that seems to have existed since at least Yan Zhitui's time. However, since the
20
John Cikoski (1978 [1977]) has attempted to reinterpret certain qualities of transitivity of the Chinese verb
with a property he calls "ergativity", terminology at odds with the conventional usage of that word. Ergativity
more usually refers to a system of case marking on nouns, inwhich (in the simplest and purest case) the semantic
subject and object are marked one way ("absolutive case"), and the semantic agent ismarked another ("ergative
case"). Cikoski, writing before Dixon's definitive 1979 study, does not work out systematically the relationship
between his ergativity and the conventional usage of the term, nor between ergativity and conventional transitivity
in Chinese.
I have considered this possibility briefly in Branner 2000: 160?166. Since we lack concrete historical information
about the languages and societies that might have been in contact to form such a Creole or Creole continuum, I see
little to do at this point other than speculate.

of the purist school to variant do not seem to be well

objections pronunciations justified,
for us to insist on smacks of laziness.
reading puristically
(b) We can read with full use of reconstruction. reconstructions into

Fully incorporating
the teaching and reading of Classical Chinese has been proposed many times but is probably
too to It is viable up to a but its most serious
complex implement pedagogically. point,
drawback is that we lack the kind of systematic evidence that would permit all texts to be
read this way with confidence.
can variant
(c) We read conservatively, retaining and promoting the traditional readings
in Mandarin. The are that the conservative tradition lacks the reductive
disadvantages
power of reconstruction, and gives the appearance of chaos. There is also

explanatory
the problem that many conservative have become unfamiliar to modern ears.
readings totally
Consider:
characters traditional common modern gloss, comments

reading reading
lidnlei,
%H lidnleilidnlei 'to implicate' (j|?originally lei$)
tdishou j^^ tdishou 'prefect' (Gudngyun reading)
diaondn diaondn 'to create difficulties [for]'
^$$.
(nan Kff 'difficulties')
tuzhuo _L^- tuzhu 'aboriginal, native'
(zhuo ^ 'to be attached to')
iiU$L shuifu shudfu 'to persuade'
(youshui _f|'IJ_ 'to canvass support for')
yan Yan in personal names to Beijing

j& alluding
This list could be extended over many pages. newer Mandarin dictionaries do
easily Many
not even list the traditional as variants, and are out of the public
readings they passing rapidly
mind. Gu's seems to be winning a latter day victory.
dingyin
In principle, I favour the reconstructive but I would like to see it connected
approach,
wherever with the conservative tradition as it has existed. The
possible reading actually
conservative tradition has the of not the student to read reconstruction
advantage requiring
aloud every time he or she opens a book. The main of reconstruction is that it allows
utility
us to visualize but it is often a great burden to the reader when
etymological relationships,
as
dogma. There is no question that literature itself is the best vehicle
presented phonetic
for and the promotion of literature that we read in a reasonably
teaching philology, requires
normal Mandarin accent.
Even when we know of variant there are difficulties in

readings, practical applying
N
them Consider the Analects passage on p. 47, above: Ja"~%~VX iCjtiL
extensively. quoted
assembles and nurtures
)^>?uMH-- [The well-bred person friends through culture, goodness
through friendships.] Not only is the fifth character ^ never read kudi today as itsmeaning
demands, but that pronunciation is itself irregular; the medieval reading {kweiHib} should
to a in Mandarin, which is not attested in the modern standard
correspond reading *gui
tradition. Should we introduce such a in order to be faithful to the medieval
reading glosses,
even if that means the tradition? Or should we read it kudi so as to be
altering existing
traditional, even we know that we are not true to the old

though being gloss?
Furthermore, the second-to-last character $$ 'to nurture' is read/w, but the medieval
demands a Mandarin not found Should we restore

reading {buoQ3c} reading *fu, today.

*Ju aswell as *gui or kudP. The danger is that our students will be unable to practice this
to read even more that as
rigour when left by themselves, and seriously they will be viewed
when encounter literate Chinese

ignoramuses they people.
To take the broad view, if the of Chinese was related to
spoken language early truly
Tibeto-Burman, then even the written of that was so far removed

language day probably
from modern Chinese as to be unrecognizable
to the modern ear. Even if we
typologically
invasively restore all the morphology we can, we will still not be able to
reproduce the full
system and sound of early Chinese. It seems, then, that we cannot

fully throw
grammatical
off the spare that has characterized Chinese since medieval times. The conservative
typology
tradition, disorganized though it is, offers us the best tool for introducing issues of historical
within a system of Mandarin
linguistics recognizable pronunciation.
References
Aarsleff, Hans. 1988. "Introduction". Wilhelm von Humboldt, On translation of von

language,
Humboldt 1836 by Peter Heath (Cambridge: Cambridge University Press), pp. vii-lxv.
la nature monosyllabique communement a la
Abel-Remusat, J. P.. 1825a. "Sur attribute langue
chinoise". Melanges asiatiques (Paris: Librairie Societe Asiatique), pp. 47?61.
-. 1825b. "Sur des formes (Paris: Librairie Societe
l'origine grammaticales". Melanges asiatiques
Asiatique), pp. 257-266.

-. 1857. Elemens de la grammaire chinoise. Nouvelle edition. Preface dated 1820. Paris: Mais
onneuve et Cie.
Bartlett, Thomas. 1985. Ku Yen-wu's response to 'the demise of human society'. Ph.D. dissertation, Princeton
University.
on ergativity in Tibeto-Burman".
Bauman, J. J. 1979. "An historical perspective In Frans Plank, ed.,
- a
Ergativity towards theory of grammatical relations (London: Academic Press), pp. 419?433.
Baxter, William H. & Laurent Sagart. 1997. "Morphology, phonology, and the lexicon in modern and
ancient Chinese". In New approaches to Chinese word formation, edited by Jerome L. Packard. Berlin:
Mouton de Gruyter.
Bodman, Nicholas C. 1980. "Proto-Chinese and Sino-Tibetan". In Frans van Coetsem and
Linda R. Waugh, eds., Contributions to historical linguistics: issues and materials (Leiden: E. J. Brill),
pp. 34-199.
Boltz, William G. 1994. The origin and early development of the Chinese writing system. New Haven:
American Oriental Society.
Branner, David Prager. 1999. "A neutral transcription system for teaching medieval Chinese". T'ang
Studies 17: 1-170.
-. -
2000. Problems in Comparative Chinese Dialectology the Classification ofMiin and Hakka. Trends
in Linguistics series, no. 123. Berlin: Mouton de Gruyter.
-. 2002. "Common Chinese and To in the Journal
early Chinese morphology". appear of the
American Oriental Society.
Chang Song-Hing [Zhang Shuangqing] fc^fe and Li Rulong ^-kojjl. 1992. "Minnanhuade 'yangru
duizhuan' paishengci $|i$ji?6\J "fÂft^" #>?^T- Dierjie Minfdngydn xueshu ydntdohui lunwenji
$-I.M^f#W#tîi- Canton: Ji'nan Daxue Chubanshe ?p&;^#&&??, PP- 76-81.
Cikoski, John S. 1978. "An outline sketch of sentence structures and word classes in Classical Chinese".
Computational Analyses of Asian & African Languages, 8: i6a-i52. Dated 1977.
Conrady, August. 1896. Eine Indochinesische Causativ-Denominativ Bildung und ihrZusammenhang mit den
Tonaccenten. Leipzig: Otto Harrassowitz.

Courant, Maurice. 1903. "L'existence, pour certains caracteres chinois, de deux lectures distinguees
par les finales k-n, t-n, p-m'\ Memoires de la Societe de Linguistique de Paris 12: 67-72.
DeLancey, Scott. 1990. "Sino-Tibetan Languages". In Bernard Cornrie, ed., The Worlds Major
Languages (New York, Oxford: Oxford University Press), pp. 797?810.
Dixon, R. M. W. 1979. "Ergativity". Language 55: 59?138.
-. 1994. Ergativity. Press.
Cambridge: Cambridge University
Doong Jongsy jf&^J. 1988. Jiang Yong shengyunxue pingshu ?r_;fc_j:{[iîf i?. Taipei: Wensshizhe
Chubanshe ?&%&)&&.
Downer, Gordon B. 1959. "Derivation by in Classical Chinese". BSOAS 22: 258-290.
tone-change
van Driem, 1997. "Sino-Bodic". BSOAS 60: 455-488.
George.
Erydjidojian $$$.&% 1984. Zhou Zumo $7_Lfg, ed. Jiangsu Jiaoyu Chubanshe ?l#jl 4$. "J&/(&??.
Escure, Genevieve. 1997. Creole and Dialect Continua. Amsterdam/Philadelphia: John Benjamins
Publishing Company.
Forrest, R. A. D. i960. "Les occlusives finales en chinois Bulletin de la Societe de Linguistique
archaique".
de Paris 55: 228-239.
Gongydng zhuan &4-i%. 1992. Chunqiu Gongydng zhuan He shijiegu ;#^^-?/$/f?]J^4^'f_\ Jin Pan &i$,
ed. SBBY. Repr. Taipei: Taiwan Zhonghua Shuju i'^^^^Jij.
Grube, Wilhelm. 1881. Die sprachgeschichtliche Stellung des Chinesischen. Leipzig: T O. Weigel.
Gu Yanwu IK?3^. I966a[i667]. "Guren sisheng yiguan -?rAE9 _??J[*". In Yinlun -g- i%, published in
Yinxue wushu -%r^3L'$, printed in Yinyunxue congshu -f|--g|^4? ^f (repr. Taipei: Guangwen Shuju
fclltfii), +/9a-i2b.
-. "Xianru zhi shuo bujin ran ^^J^^f-S^^lfc^ _?#*" In Yinlun
I966b[i667]. liangsheng geyi
-fjj*i?r,published in Yinxue wushu -JîL^, printed in Yinyunxue congshu -fMl^jEIr" (repr. Taipei:
Guangwen Shuju fcZL^M]), T/2a~4b
-. Shi Published in Yinxue wushu in Yinyunxue
1966c[1667]. benyin l^^-f". ^-^j? ~%, printed
congshujOrifi^Jli^r (repr.Taipei: Guangwen Shuju J%x?M]).

Guangya gulin #*.
}%%$.%? 1998. Xu Fu %fli, ed. Shanghai: Jiangsu Guji Chubanshe ?cj$.'?'|f tfcJ?&;f?
Han shu /H^. 1955. Wang Xianqian JL;fc|?, ed., Han shii buzhu /H^^viL. Taipei: Yiwen Yinshuguan
-. 1962. Peking: Zhonghua Shuju îfc^Mj.
[Huang Yongwu] ^"^C^,. 1965. Xingsheng duojian huiyi kao ty$fc ^^fc^^^- Taipei: Wenshizhe
Chubanshe xi^ ifcX&fi
von Humboldt, Wilhelm. 1836. Uber die Verschiedenheit des menschlichen und ihren
Sprachbaues
Einflufi auf die geistige Entwickelung des Menschengeschlechts. Berlin, Gedruckt in der Druckerei der
Akademie der Wissenschaften. Edition: Wilhelm von Humboldts

Koniglichen gesammelte Schriften,
ed. Albert Leitzmann. Berlin: B. Behr's Verlag, 1907; repr. Berlin: Walter de Gruyter & Co.,
1968.
-. 1971. and intellectual Translation of von Humboldt 1836
Linguistic variability development. by
George C. Buck and Frithjof A. Raven. Philadelphia: University of Pennsylvania Press.
Jespersen, Otto. 1894. Progress in language. London: Swan Sonnenschein & Co.
Jia Changchao % g^- 1985[1022]. Qunjing yinbidn ?f M-lM^. Series Congshu jicheng chubian
It^!c$?#&!, vol. 1208. Peking: Zhonghua Shuju ^^t^Mj.
Jiang Yong ?r_^- I966[i8i9]. "Bian sisheng Ê9 _?." In Yinxue bidnwei -?r^$f$L, printed in Yinyunxue
congshu -j^ll^jjE^ (repr. Taipei: Guangwen Shuju fcx^M), 4a~5a.
Jingdidn shiwen M#J-# 3C- 1988. Compiled by Lu Deming ft%GM- Deng Shiliang $$&%$. and Wong Kuan
Io [Huang Kunyao] %*$?>, eds., Xinjido Jingdidn shiwen &f&ji_L&!f 3fc. Taipei: Xuehai Chubanshe

Bernhard. 1920. "Le proto-chinois, flexionnelle". 11: 205-232.

Karlgren, langue Journal asiatique
-. 1933. "Word families in Chinese". BullMFEA 5: 9-120.
Kennedy, George Qin Shouzhuo 4tîk]- 1964(1956]. "Zai shuo wu wo il^^-^". Tr. [Li Baojun]
^4?*?j. Tien-yi Li, ed., Selected works of George A. Kennedy (New Haven: Far Eastern Publications),
pp. 434-442.
-. 1964. "Word Classes in Classical Chinese". Li, ed., Selected works
Tien-yi of George
A. Kennedy (New Haven: Far Eastern Publications), pp. 323-433.
Koerner, Konrad. 2000. in 19th and 20th century of language: a neglected aspect of
"Ideology study
historiography". Indogermanischeforschungen 105: 1-26.

linguistic
? a
Laufer, Berthold. 1967(1919]. Sino-Iranica Chinese contributions to the history of civilization in ancient
Iran, with special reference to the history of cultivated plants and products. Chicago: Field Museum of
Natural Publication 201. series, vol. XV, no. 3: 185-630. at

History. Anthropological Reprinted
Taipei: Ch'eng-Wen Pub. Co., 1967.
Lepsius, Karl Richard. 1861. "Uber die Umschrift und Lautverhaltnisse einiger hinterasiatischer
Sprachen, namentlich der Chinesischen und der Tibetischen". Ahhandlungen der Koniglichen Akademie
der Wissenschaften zu Berlin aus demjahre i860, pp. 449?496.
Lu Wenchao ?.3C%Z. 1985. "Ziyi bu sui yin qubie ^jfe^fit-frKW- In Zhongshan zh'ajx $&ditlit,
Series Congshu jicheng chubidn Jt^l^$?#!&, vol. 352 (Peking: Zhonghua Shuju ^^-^M]) Ch. 1, p.
1.
Ma Jianzhong ,!%??,$.. 1954(1898). "Dongzi bianyin fe^W^""- In Mdshi Wentong $j fi^SLiQ., Ch. 5
"Shizi jf ^", Part 2. Edited and with notes by Zhang Xichen :?$M^ (Peking: Zhonghua Shuju
+ *#^), pp. 249-263.
Mair, Victor H. & Tsu-lin Mei. 1991. "The Sanskrit origins of recent style prosody". HJAS 51:
375-470.
Mao Qiling ?>-?{$$. 1991. Yunxue zhiydo ||^||f^. Peking: Zhonghua Shuju +^^ M).
Henri. et derivation en chinois de la Societe de Linguistique
Maspero, 1930. "Prefixes archaique". Memoires
de Paris 23: 313-327.
Mei, Tsu-lin. 1970. "Tones and prosody in Middle Chinese and the origin of the rising tone". HJAS
30: 86-110.
-. 1980. de shijian E9J?#] Jk t ^J ^ W4J '?" Zhongguo yuwen 1980/6:
"Sisheng bieyi zhong cengci
427-443.
Norman, Jerry L. and W. South Coblin. 1995. "A new approach to Chinese historical linguistics".
JAOS 115: 576-584.
Pan Zhonggui >#-$?!, ed- 1983 (preface). Jingdidn shiwen yunhidn M##Xl|l^. 3 vols.
Taipei: Zhonghua Minguo Xingzhengyuan, Wenhua Jianshe Weiyuanhui ^^
du Ponceau, Peter S. 1838. A dissertation on the nature and character in a letter

of the Chinese system of writing
toJohn Vaughan, Esq. American Philosophical Society. Part of this book was
Philadelphia: published
in 1828.
Pulleyblank, Edwin G. 1973. "Some new word families in Chinese". 1:

hypotheses concerning JCL
111-125.
-. 2000. in Old Chinese". 28/1: 26-51.
"Morphology JCL
?
Qian Daxin j|*Jf. 1986. "Yizi liangdu ^tfc". In Shijid zhdi ydngxin lu+$#?#?!? (SBBY.
Repr. Taipei: Taiwan Zhonghua Shuju i'^f^^^Mj), 5/3a?4b.
Sagart, Laurent. 1993. "L'infixe -r- en chinois Bulletin de la Societe de linguistique de Paris
archaique".
88: 271-306.
-. 1999. The roots of Old Chinese. Amsterdam/Philadelphia: John Benjamins Publishing Company.

Shen Jianshi j&j)|b-?*. I986[i933]. "Youwen shuo zai xunguxue shang zhi yan'ge ji qi tuichan
%?i&&nib$X^fe$Jk&$W. Shen Ji'anshi xueshu lunwenji&**##*:?& (Peking:

Zhonghua Shuju), pp. 73-173.
Shiji ?i&. 1959. Peking: Zhonghua Shuju ^^^Mj.

Sweet, Paul R. 1980. Wilhelm von Humboldt. Columbus: Ohio State University Press. Vol 2.
-. 1989. Review of Aarsleff 1988. Historiographia 16: 387-392.
Linguistic
Taiping yulan $L^-$?%. 1894. Reproduction of a Song edition. Shanghai: Jl<%:${iXiJ& Mj. Pref. dated
1812.
Wang Li ?jj. 1982. Tongyudn zididn J*|$fc^&. Peking: Shangwu Yinshuguan.

Wang Yun i^. 1985. Ltiydu yishuo 3t>$jZ?ifc. Series Congshu jicheng chubian jjt^f HkfawtiOb. Peking:
Zhonghua Shuju ^^^f Mj.
Wu Chengshi ^.j^cft. 1986. Wu Jidnzhai yishu: Jingji jiuyin xulu, Jingji jiuyin bidnzheng
: -
**#?* JWIf *** ******. Peking: Zhonghua Shuju t**?
Yan ZhituiJ^#. i96o[A.D. c. 589]. Ydnshi jidxun huizhu $#,^SWl?&. Zhou $&iti> ed.
Fagao
Taipei: Academia Sinica. See also Zhou Zumo (1966b).

Yuan Mei ;?.;&. 1993. "Yinyi fanchong -|"jfe^$". in Suiyudn suibi fttiSJSt^, Ch. 24 "Shiwen zhushu
lei, part 1 t$3l3ti&$ft, _L", pp. 408-409. In Yuan Mei ?$k, vol. 5. Nanking:
qudnjiffifa Jiangsu Guji
Chubanshe >*IJL"*"J|ftJMfc.
Zhang Shoujie $S.^$p (fl. 8th c). 1975[1959]. "Lunyin liîHN". Shiji zhengyi?HL j?^, incorporated
"
into Shiji jfcjfc, compiled by Sima Qian sj J%i&(Peking: Zhonghua Shuju), vol. 10, Shiji jijie xu
?*&##", pp. 15-16.
1964.GwHanyu yufdzilido huibian -?rj?&&&4?

Zheng Dian Jfc.J|andMai Meiqiao <jMfH&. jjf$f?&|.
Peking: Zhonghua Shuju ^^t^Mj
Zhouli zhushu ^)&&$JL. 1980. Shisdnjing zhushu -f-_HL|?>?j|S,. Peking: Zhonghua Shuju Îf^y^.
Zhou Zumom&i?. 1966a. "Sisheng bieyi shili E9J?#]4feffW. Wenxue ji f^$?%. (Peking:Zhonghua
Shuju, 1966), pp. 81-119.
-. 1966b. "Ydnshi zhubu $&^3fll-^ltJg&;Mj"' Wenxue ji $#& (Peking:
jidxun Yincipian'
Zhonghua Shuju, 1966), pp. 405-433.


Branner2003 - On Early Chinese Morphology and Its Intellectual History

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Branner2003 - On Early Chinese Morphology and Its Intellectual History

Încărcat de

Drepturi de autor:

Formate disponibile

On Early Chinese Morphology and Its Intellectual History.

Winner of the Barwis–Holliday

This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

Winner of the Barwis-Holliday Award for 2001

DAVID PRAGER BRANNER

when its reading is altered Modern scholars have for these

received sources on medieval Chinese the 'rime-books' of the "fetfiljt

This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

of data from character structure, and other sources. in the discussion

placed in curly brackets {})

something other than the qusheng:

verb in non-qusheng noun in qusheng

& chu {tshyuQ3b} 'to dwell at' fa chu {tshyuH3y} 'place'

Qusheng is indicated inmedieval transcription by the letter H at the end of a syllable. It is

is to make one's home in choosing, one does

> = rather than the usual

%, . -1^ = {triHjb} (i.e., rather than the usual reading [tri3b])

This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

[The well-bred person assembles friends culture, and nurtures through

morphology in the modern study of and research in classical Chinese.

2. The conservative tradition of variant

what is primarily devoted to discussion of content and meaning.

This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

continue to appear in numbers in commentaries

Rather, it is intended to be used as an to the texts, and is arranged

But the Shiwen assembles thousands of from various sources, it

s work has been influential, and seems to have served as the

Overall, the native Chinese tradition of variant was conservative. Its

accepted today, and then with the Manchu-time philologists.

This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

3. Western reconstructionism based on variant readings

encountered. Mediterranean civilization has understood itself to be

The watershed in the of Chinese historical was the

Karlgren; I see four stages in its

savants to exist of any and even to represent a pure

resounding rebuttal. Du Ponceau holds that any true written must be

morphology is fundamentally absent.

This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

Von Humboldt must have had access to information about Chinese,

can deny that the old a stirring to the fact that

kleinen, storenden Verbindungslaute abschneidet, ausschliesslicher und gespannter auf denselben

so diirften doch auch die entschiedensten Vertheidiger dieser Sprache

emporbluhen. (1903 [1836], pp. 255-256)

forth.] (tr. 1971, p. 196)

morphology it has failed to develop something essential.

This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

Sprache herabsinken konnten.

Mehrsilbigkeit und in verharteter Einseitigkeit an der Grenze ihrer Entwickelung

At first may seem to hold a contemptuous view of Chinese-speaking

stage, and hence it has lost something. Chinese is secondary, not

it has in such a way as to lack necessary, as a lose

This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

The idea of lost morphology in Chinese arose on at least two

separate occasions. The earlier one was due to the Otto

1894 book he cites the following examples:

[__] xvang 'king' wang 'to become king'

[&] cong 'to follow' zbng 'follower' zong 'footsteps'

[%] shou 'to acquire' [^] shbu 'to give'

the dropping of final consonants (p, t, k).

par une dialectale, soit de quelques autres causes, il faut retenir la

and Courant the first concrete attempts to connect variant readings

This content downloaded from 193.142.30.220 on Sat, 28 Jun 2014 13:42:57 PM

(It should be remembered, without prejudice, that by "explanation" linguists sometimes

> - 5M J&tfIfc - ^TA^ &&i%&? .

%&iL^m -***** - &#f# -?A.^lS-^1%^**^ >?& -?&

- -******* - ## -??4:H

- ) >b?T? - - - st^vst >?**? -?