ABSTRACT
DeFrancis (1989)
claims that all writing systems are similar in being phonetically based. Chinese script, commonly cited as an
exception, is according to DeFrancis essentially a syllabic phonographic
system. The present article argues
that this claim confuses diachrony with synchrony. It may be correct that the creation of a script always
involves phonetic considerations, but subsequent evolution of script and spoken
language can remove the phonetic basis of a writing system. It is difficult to agree that modern
Chinese writing is essentially phonetically-based; and it is certain that
phonetic motivation is not a necessary feature for a script.
1. Introduction.
John DeFrancis (1989)
has argued at length that all writing systems used now or in the past are
essentially similar in being based on a phonetic principle; and, in particular,
that the Chinese script does not represent a fundamentally different type of
system from scripts generally recognized as phonetically-based. DeFrancis’s argument has been widely
reviewed and discussed not only by Sinologists but by many commentators on the
comparative study of scripts and on the psychology of literacy, and to date the
clear consensus is that DeFrancis has successfully made his case for the
universality of a phonetic principle in writing systems: see for instance Krippes (1990), Wrenn
(1990), Daniels (1991: 838), Tzeng (1991), Burling (1992: 423),
Carello (1992: 212), Coulmas (1992: 254), Coe (1992: 31, 292),
Liberman (1992: 168-9), Mattingly (1992: 18). King (1991), while disagreeing with DeFrancis on certain
specific issues, accepts that ‘it would be unsurprising if DeF[rancis]’s thesis
proved to be correct’.
DeFrancis constructs
his argument largely by taking issue with various points made in Sampson
(1985).[1] I am puzzled to know why DeFrancis
attacks my exposition so vigorously, since it seems that on the issues that
concern him most deeply DeFrancis and I are explicitly arguing on the same
side. Both DeFrancis and I have
independently taken pains to rebut the idea, which continues to be put forward
periodically by various writers, that Chinese script is a primitive or
intrinsically inferior vehicle for intellectual communication by comparison
with alphabetic European writing (see e.g. DeFrancis 1989: 221, 244-5;
Sampson 1985: 160-5; Sampson 1991).
And both of us have stressed that ‘Chinese characters represent words
(or better morphemes) not ideas’, as DeFrancis put it (1984: 145): compare Sampson (1985: 149). Nevertheless, I believe that there are
real and linguistically interesting typological differences between scripts
which DeFrancis blurs, and that the consensus identified in the previous
paragraph is misguided.
Sampson (1985: 32)
drew a (by no means original) set of distinctions among scripts or script-like
systems, between what I called semasiographic
and glottographic systems (the former
relating visible marks to meaning directly without reference to any specific
spoken language, the latter using visible marks to represent forms of a spoken
language), and, among glottographic systems, between logographic and phonographic
systems (the former representing a spoken language by assigning distinctive
visible marks to linguistic elements of André Martinet’s ‘first articulation’
(Martinet 1949), i.e. morphemes or words, the latter achieving the same goal by
assigning marks to elements of the ‘second articulation’, e.g. phonemes, syllables). These are ideal types, and it is likely
that actual, complex writing systems will commonly display at least some
characteristics of more than one type.
Nevertheless I believe that many scripts can appropriately be viewed as
predominantly exemplifying one rather than another type, and I do believe that
modern Chinese script is a fairly good example of logographic writing, whereas
the written forms of many European languages are fairly good examples of
phonographic writing (though written English is too mixed to be described
confidently as clearly phonographic or clearly logographic).
DeFrancis, by
contrast, argues:
(i) There is no such
thing as semasiographic writing ã the examples often quoted of direct visual
representation of ideas, such as the American Indian pictorial messages
discussed e.g. by Gelb (1963: ch. 2), are primitive, limited affairs which
do not deserve even to be regarded as forerunners of full-scale writing
systems. Any full-scale script
capable (as a spoken language is capable) of expressing whatever can be thought
must necessarily do so by representing the elements of a particular spoken
language; ‘... all forms of partial writing [by which DeFrancis refers to
semasiographic writing], other than ... specifically speech-related examples,
... do not properly belong in a discussion of writing at all.’ (DeFrancis
1989: 57).
(ii) A fortiori, Chinese script is not
semasiographic.
(iii) Any script which represents a spoken
language does so chiefly by symbolizing phonetic units of that language, in
other words there is no such thing as logographic writing; ‘the heart of all
writing systems is its [sic] phonetic base’ (DeFrancis 1989: 56).
(iv) A fortiori, Chinese script is not
logographic. It is essentially a
syllabic phonographic script, though one of a rather elaborate, irregular
kind: ‘Chinese and other so-called
logo-syllabic scripts are not a separate type but a subcategory of syllabic’
(DeFrancis 1989: 253).
To my mind, (i) is
largely a matter of definition; (ii) is true; (iii) is false; (iv) unavoidably
involves an element of subjective judgment, but if it cannot definitively be
regarded as false it is at least a surprising way to think about Chinese
writing.
2. Semasiography.
It is indisputable
that there exist systems of communication by visible marks which are
independent of any particular spoken language. One example is the road sign system, which, for instance,
uses the contrast between circular and triangular shape to distinguish command
from warning, and displays a red and black car side by side to signal ‘No
overtaking’. Some of these signs
(such as the one just cited) are partly iconic, others (e.g. the white
horizontal bar on a red disc for ‘No entry’) are wholly arbitrary, but almost
all of them are entirely independent of spoken language. It makes no sense to ask whether the
first sign cited should be read as ‘No overtaking’ or as ‘Overtaking is
forbidden’, or whether it should be read as an English or as a German phrase.
I agree with DeFrancis
that no semasiographic script ever used in practice has approached the degree
of generality and flexibility possessed by all spoken languages. Each such system has been limited to
expressing messages relating to some narrow, limited domain, such as traffic
discipline. Whether this makes
semasiography so different from glottography that the word ‘writing’ is
inapplicable to the former, or whether rather one should call existing
semasiographic systems ‘writing’ of an unusual, limited type, is purely a
question of how one chooses to use the word ‘writing’ and as such, surely, is
not worth many moments’ discussion.[2]
In Sampson
(1985: 30-2) I speculated about whether there might ever be a
semasiographic system comparable in expressive power to a spoken language. I gave reasons for thinking that in
practice this is unlikely to happen, but argued that ‘logically speaking such
an outcome seems not absolutely excluded’. DeFrancis is unwilling to admit that a ‘full’ system of semasiography
could be even a logical possibility.[3] He believes that those examples which
do occur are necessarily limited to expressing very simple ideas ã concepts
that would be expressed in speech by a word or short phrase rather than a
multi-word sentence.
DeFrancis’s argument
to this effect turns on examination of an example quoted in Sampson
(1985: 28‑9) of purported complex semasiography, the ‘Yukaghir love
letter’. I had taken this example
from a well-known book on writing, Diringer (n.d.: 35), and I retailed
Diringer’s explanation of it without trying to check this. DeFrancis has done the discipline a
considerable service by investigating the history of the example in detail, and
it turns out to be something rather different from what Diringer and I
described, and arguably not an example of ‘communication’ at all.
If I had known the
facts about the Yukaghir love letter which DeFrancis has brought to light, I
would probably not have used it in my book.[4] However, loss of this particular
example does not establish the generalization that semasiography can never be
used for logically complex messages.
Sampson (1985: 31-2) illustrated a second example, a set of
instructions distributed with a Ford Escort car in 1982, in the form of two
rows of six stylized pictures expressing a message that is admittedly less
complex than that allegedly expressed by the Yukaghir example but is still
fairly complex, requiring several clauses to express in English. (I suggested the translation ‘When
starting from cold, turn on the ignition without touching the gas-pedal; if the
engine is warm, press the gas-pedal halfway down as you turn the ignition
key’.) In this case, I know the
provenance at first hand, and the document remains in my possession (I should
be happy to show it to enquirers).[5] Whether it would be possible in
principle to develop this kind of writing into a full-scale script, capable of
expressing everything that can be expressed in speech, remains to my mind an
open question.
3. Chinese writing is not semasiographic.
In the seventeenth
century it was supposed by a number of European philosophers (cf. Knowlson
1975: 25) that Chinese script was a real example of the kind of full
semasiographic system discussed as a hypothetical possibility in the preceding
section. One still sees this
concept expressed by misinformed writers today. It is quite wrong:
Chinese script was created as a means of representing visually a
particular spoken language, the Chinese language as it existed at the period
when the script was developed.
No-one familiar with the language and script could doubt this.
In Sampson (1985: 149) I list various considerations which
demonstrate the point, such as the fact that (as with any other natural
language) the morphemes of spoken Chinese are often polysemous and have ranges
of meaning which are arbitrary and idiosyncratic, and these meaning-ranges will
normally be common both to a spoken morpheme and to the written graph (or
‘character’) which represents that morpheme. Occasionally, it is true, a single polysemous spoken word
will have alternative graphs for separate subsections of its meaning-range,
rather as the single spoken English etymon /metl/ has alternative spellings metal and mettle for separate subsenses ã
and, as in the English case, the alternative writings are perceived as distinct
vocabulary items. For instance guo meaning ‘fruit’ and
hence also ‘result’ has distinct graphs for the ‘fruit’ sense and the ‘result’
sense. But even in such cases the
alternative Chinese graphs will between them cover the identical meaning-range
that is covered by the single spoken Chinese word; we do not find
meaning-boundaries between written Chinese graphs and meaning-boundaries
between spoken Chinese words overlapping and cutting across one another, as one
finds with meaning-boundaries between Chinese words and meaning-boundaries
between words of English or another spoken language.
There is, admittedly,
a special consideration in the case of written Chinese which might seem to put
its glottographic status in doubt.
Once a spoken language has acquired a written form, the two linguistic
systems may evolve independently so that the relationship between written and
spoken languages becomes increasingly remote. With Chinese this happened in a way for which I know no
parallel: during much of recent
history, before the reforms in written usage associated with the May Fourth
Movement initiated in 1919, the standard written language of China (wen yan, or
literary Chinese) was a language which when read aloud in contemporary
pronunciation could not be understood by a hearer, irrespective of how learned
he might be, because the spoken equivalent of a written text did not contain
sufficient information to determine the identities of the morphemes of which
the text was composed. There were
two reasons for this. Sound
changes during the long period since the creation of the script had removed
many phonological contrasts and thus introduced an extremely high incidence of
homophony among morphemes;[6]
and developments in literary usage had created many possibilities of
meaningfully combining morphemes in writing in ways that would never have
occurred in spoken Chinese at any period of its history, thus reducing the
chance of determining the intended morpheme among a set of homophone candidates
by reference to its morphemic environment.[7] Therefore literary Chinese not merely
did only function but could only function
as a written and read language, not as a spoken and heard language.[8] This might seem to bring literary
Chinese within the definition of semasiography. But any literary Chinese text has a perfectly specific
spoken form, composed of morphemes many of which occur in modern spoken Chinese
and all of which are etymologically identifiable with morphemes that have
occurred in spoken Chinese at some historical period: there is a well-defined way of reading a literary Chinese
text aloud (morphemes which are obsolete in spoken Chinese are given the
pronunciations that result by applying subsequent sound-laws to the
pronunciations the morphemes had when they were current in speech), even if
this activity achieves no communicative purpose. The situation is very different from a case such as the Ford
starting instructions, where one has to invent some form of words to express
orally the message of the text, and independent readers could reasonably select
different spoken word-sequences to express the ‘written’ message. Chinese script is glottographic.
4. Is Chinese writing logographic?
People with limited
awareness of linguistic concepts sometimes talk as if the statement that a
script represents a spoken language necessarily implies that the units of the
script must represent phonetic units, such as consonants, vowels, or
syllables: that is, they overlook
the logical possibility that a script might be logographic. But linguists know that any natural
language has units at many levels, and in particular that all human languages
exhibit a ‘double articulation’ into units carrying meaning, on the one hand,
and phonological units whose function is to serve as perceptually-distinctive
building blocks out of which meaningful units can be assembled, on the
other. This feature is universal
in human language: it applies as
much to languages having no written form as to languages which have scripts.[9] It is at least logically possible,
therefore, that a glottographic script might assign distinctive symbols to
elements of the first rather than of the second articulation. DeFrancis accepts, I think, that this
is a logical possibility, but he does not believe it is a practical possibility
or that Chinese writing should be regarded as such a script.
Consideration of the
status of Chinese script in this respect is complicated by an unusual property
of Chinese as a spoken language.
In European languages there is commonly little regular relationship
between the position of boundaries of units of the two articulations. The spoken English word analysed, for instance, comprises two morphemes,
one of which is realized by two complete syllables and part of a third, while
the other is realized by just the final consonant of the third syllable. In Chinese, by contrast, phonological
syllable-boundaries (which are always clearly marked ã Chinese phonology lacks
‘interludes’ (Hockett 1955: 52) such as the /t/ of English butter, which belong equally to the preceding
and the following syllable) almost always coincide with morpheme
boundaries: each morpheme is
represented by exactly one spoken syllable. (There are marginal exceptions, such as, in modern Mandarin,
the nominal suffixes -z,
-r, and, in all
varieties of Chinese, a set of disyllabic morphemes for unusual flora and
fauna, the classic example being shanhu ‘coral’ mentioned by DeFrancis (1989: 259); but
these cases are few enough not to be significant in the present context.)
Because, broadly, each
Chinese morpheme is realized as a syllable, one might suppose that for this
particular language there could be no distinction between a syllabic
phonographic script and a morpheme-based logographic script. Conceptually, however, there is a clear
distinction, because of the high incidence of homophony in Chinese already
mentioned. The great majority of
phonologically-possible Chinese syllables each represent several etymologically
distinct morphemes with unrelated meanings. (The standard dictionary of modern Chinese Xiandai Hanyu
Cidian (Peking, 1981)
lists on the order of 10,000 morphemes, and about 1280 distinct isolation-form
syllables, giving an average of about eight morphemes per syllable.) Suppose English were written in a
script whose graphic units corresponded to word-sized units of the spoken
language: there would be a clear
conceptual distinction between a script which used a given graph for all
occurrences of words pronounced /s n/ irrespective of meaning, and a
script which assigned unrelated graphs to /s n/ ‘male child’ and
/s n/ ‘star which our planet orbits’. If English had numerous homophones and if one of these two
alternative principles were adopted consistently in each case, we would call
the script phonographic if homophones were written alike, and logographic if
each etymologically distinct word had its own written form graphically
unrelated to the written forms of its homophones.
The case of Chinese
script is certainly not as clearcut as either of these two hypothetical cases.
Chinese graphs come
close to standing in a one-to-one relationship with morphemes, so that
etymologically distinct morphemes will normally have distinct graphs. There is a significant incidence of
distinct morphemes sharing a common graph ã in some cases the morphemes written
alike are homophones (e.g. bie ‘other’ and bie
‘don’t’), in other cases they have distinct (but usually similar)
pronunciations (e.g. qi
‘wonderful’ and ji
‘odd number’); and sometimes it is impossible to know whether alternative
senses associated with a particular graph/pronunciation pairing represent
polysemous evolution of meaning within a single etymon, or represent
historically-unrelated homophones.
But even when a single graph clearly stands for two or more homophonous
morphemes, the set of morphemes written with that graph will virtually always
be a minority of the total set of morphemes sharing that pronunciation. If one were to consider Chinese graphs
as atomic units lacking internal structure, the script would rather clearly be
basically logographic, though with occasional ambiguous assignments of graphs
to alternative morphemes which are phonetically identical or related.[10]
However, most Chinese
graphs are not simple forms but contain internal complexity, and the nature of
this complexity is such that a graph often includes a more or less precise clue
to its pronunciation. This feature
of the script may be (and is, by DeFrancis) used to argue that it is
fundamentally a syllabic phonographic script with logographic accretions.
The internal structure
of Chinese graphs is best explained historically. On the order of a thousand morphemes are represented by
simple graphs which originated as pictures corresponding to their meanings
(though at an early date the iconic quality of the graphs was lost through
stylization of their shapes). A
morpheme X having no simple graph was written by borrowing the graph for some
morpheme Y that already had a written form and that was pronounced similarly to
X (though, commonly, X and Y were not pronounced identically); and usually a
distinguishing element was added to differentiate the written form of X from
that of Y, in the shape of a simple graph for a morpheme related in meaning to
X (‘hand’ for verbs of action, ‘water’ for words connected with liquids,
etc.). There appears to have been
a very early stage at which the system of borrowing graphs by reference to
similarity of pronunciation had not yet come into play, but only through that
principle did Chinese script become a full-scale writing system capable of
representing everything in the spoken language. Thus the bulk of all graphs in the system which emerged are
compounds containing two parts, a phonetic
and a signific, and of the two the
phonetic component is unquestionably more important: there are on the order of 1400 different graphs used as
phonetics within phonetic/signific compounds (some of the 1400 are themselves
phonetic/signific compound graphs), but only about 200 graphs used as
significs, and in the early history of the script there seems to have been some
flexibility in the use of significs ã they might be omitted, or a single
morpheme might be written with alternative significs ã whereas a morpheme was
only rarely represented by alternative phonetics.
I shall use the term
‘compound graph’ for a Chinese graph that can be analysed into phonetic and
signific. (There are also graphs
which might be called compound because they combine two simple graphs to represent
a word whose sense is connected with the senses of both components, but for
present purposes graphs of this sort are not interestingly different from
simple graphs, and my use of the term ‘compound graph’ will exclude them.)
It might, then, be
reasonable to describe the Chinese script during an early phase of its history
as fundamentally phonographic, although phonetically imprecise (a phonetic
element was associated not with a fixed syllabic pronunciation but with a range
of related pronunciations), and having important logographic features (the use
of significs, and the fact that a morpheme lacking a simple graph was written
not with the graph for any
similar-sounding morpheme but with that of one particular conventionally-fixed
near-homophone).[11] To this extent I accept DeFrancis’s view
of the nature of Chinese script.
However, since Saussure it has been accepted among linguists that the
structure of a linguistic system at a given time is a separate issue from the
question of how the system evolved historically. The evolution of Chinese script and spoken language over
some three millennia have brought about changed relationships between script
and speech, which make the script now more logographic and less phonographic
than before.
Thus, sound-changes
have in many cases caused syllables which used to be phonetically similar to
diverge in pronunciation. The
morphemes for ‘two’, ‘grease’ were pronounced in Middle Chinese nzi-, ni- respectively, so it was reasonable for ‘grease’ to
be written as a compound graph combining ‘two’ as phonetic with ‘flesh’ as
signific; but, through the operation of regular sound laws, the modern Mandarin
pronunciations have become respectively er, ni, making the rationale of the ‘grease’ graph quite opaque. Furthermore, the massive loss of
phonemic contrasts which has been a marked feature of the history of Chinese
phonology means that a given absolute degree of difference between the
pronunciation of two syllables constitutes a much larger relative difference,
in the context of the impoverished modern phonological system, than it did in
the richer phonological systems of Middle or Old Chinese; this again reduces
the perceived phonetic homogeneity of a family of morphemes written with the
same phonetic element. It often
happens that a morpheme X is written with a compound graph in which the element
that is historically the phonetic seems quite inappropriate with respect to its
modern Mandarin pronunciation, although there are several graphs standing for
words that are perfect homophones of X which are used as phonetics in other
compound graphs. (For instance, di ‘place’ is written with the phonetic ye ‘also’ and the ‘earth’ signific, but
there are at least two simple graphs pronounced di and used as phonetics in other compound
graphs.) Sometimes a morpheme which
remains current is written with a compound graph, the phonetic element of which
when written independently stands for an obsolete morpheme, so that the
phonetic element plays no part in the average reader’s understanding of the
compound graph (for instance, cha ‘insert’ is written with the phonetic cha ‘pestle’ and the ‘hand’ signific, but cha ‘pestle’ is thoroughly obsolete).
These factors relate
to developments in the spoken language; but developments in the written forms
have also tended to change the nature of the script/spoken-language
relationship in the same direction.
The written representations of morphemes have become more fixed: a minority of morphemes do still have
recognized alternative graphs, but there is no general freedom to omit or vary
signific elements (where variant forms exist, commonly the difference between
the alternatives is limited to different spatial arrangements of the same
phonetic and signific elements).
The shape of a graph used as a phonetic element in a compound graph has
sometimes diverged from the shape used when it occurs as an independent graph,
so that the historical identity of the two is no longer recognizable. (Thus feng ‘envelope’ was originally written with a
compound graph containing the phonetic element feng ‘gracefulness’, but within the ‘envelope’
graph this element was given an extra stroke changing it into the graph gui ‘sceptre’; the graph for cha ‘inspect’ began as a compound containing
the phonetic element qie
‘moreover’ ã the phonetic relationship between these syllables is not very
close,[12]
but this relationship is in any case irrelevant to the modern reader since a
slight alteration in the writing of the phonetic element within the ‘inspect’
graph has made it appear to derive from dan ‘dawn’ rather than qie ‘moreover’.)
The net effect of such
developments is that modern Chinese script, as a writing system for modern
spoken Mandarin Chinese, is one in which many graphs contain recognizable clues
to the pronunciation of the morphemes they represent but many others contain no
such clues, and where there are such clues they are often very vague. From a knowledge of the pronunciation
of a morpheme and the information that its graph is a phonetic/signific compound,
it would only rarely be possible to predict the identity of the phonetic (one
would more often be able to predict the identity of the signific from the
meaning of the morpheme). From the
point of view of a present-day Chinese-speaker learning to read and write, the
phonetic element of a compound graph is often just an arbitrary part of the
graph’s overall shape to be learned, and its historical role as phonetic is as
opaque and irrelevant to his task as is the derivation from Greek ana and luein to an English-speaker learning to use the word analyse.
Whether one regards
such a system as essentially logographic with elements of a phonographic
principle, essentially phonographic with elements of a logographic principle,
or as too mixed to assign to either category, must depend on a subjective judgment
as to how close and regular the relationship between pronunciations and written
forms needs to be before one treats that relationship as the central organizing
principle of a script. I tried to
give readers of Sampson (1985: 157-8) an impression of the situation by
discussing the analysis of ten graphs chosen at random, and I used this
discussion to support my own judgment (shared with many other commentators)
that modern Chinese script is essentially logographic though with limited
phonographic features. DeFrancis
makes the opposite judgment, and supports it with numerical statistics
(DeFrancis 1989: 99-113). But
DeFrancis reaches very high figures for the degree of phoneticity of Chinese
graphs by counting a graph as phonetically motivated if there is any
resemblance at all between the pronunciation of the graph and that of its
phonetic element, and even (in the case of the highest figure quoted, on p.
113) by counting all graphs that were historically phonetic/signific compounds
as phonetically motivated irrespective of whether sound-changes have destroyed
that motivation. While in general
I favour the use of precise numerical methods rather than impressionistic
approaches in linguistics where possible (cf. Sampson 1992), I am not persuaded
that the question of how far Chinese script is phonetically motivated for its
present-day users is one that can be addressed statistically: it is not clear what we ought to count. We have no accepted measure, for
instance, of phonetic distance between linguistic forms that would allow us to
specify how appropriate on average the phonetic element of a compound graph is
to the graph’s current pronunciation.[13]
I continue, then, to
regard modern Chinese writing as essentially logographic; although subjective
judgments cannot ultimately be disputed, it is surprising to me that someone
would make the opposite judgment about the script as it now exists.[14] However, the script was certainly more
phonographic at earlier stages of its history. There would be nothing surprising in a judgment that Chinese
script at the period when it first developed into a full writing system capable
of representing the whole spoken vocabulary was an essentially phonographic
system (though many would prefer to classify it as too mixed to assign to
either type, and, as DeFrancis rightly points out (1989: 99ff.), the facts
as well as their appropriate interpretation are quite debatable in this
area). Part of DeFrancis’s thesis
may be a diachronic claim, that full-scale writing systems capable of
expressing everything expressible in speech must always develop through heavy
use of a phonographic principle, whether or not their phonographic status is
compromised by subsequent evolution.
This is a conjecture which seems likely to be true: it is hard to imagine a script with
separate symbols for all the thousands of elements in the lexicon of a natural
language being created without any exploitation of phonetic similarities for
the task of generating the symbols.
It can be no more than a conjecture; to date too few writing systems
have been created independently in the history of Mankind to test it
adequately, and the cultural integration of the world means that future
independent creations are now unlikely.
5. Can a script be less phonographic than Chinese writing is?
DeFrancis uses his
interpretation of the Chinese writing system as an essentially phonographic
system in order to support the more general thesis that no script can be
logographic. But even if one
accepted DeFrancis’s interpretation of Chinese script, the more general point
would not follow.
There is in fact
another example of writing which is much more clearly logographic than Chinese
writing: namely the use of Chinese
graphs to write Japanese words. Japanese
uses a mixed script in which grammatical suffixes and particles are written
phonographically while stems of lexical items are written with Chinese
graphs. Many modern Japanese
lexical items are Chinese loanwords, and in these cases the function of the
phonetic element within the Chinese graph (if it is a compound graph) remains
what it is with respect to the Chinese language ã except that the Japanization
of morpheme pronunciations introduces a further layer of distortion into the
relationship between graphs and their spoken forms. But Japanese also retains a stock of native lexical items,
which are written with graphs for Chinese words having the same or similar
meanings. Since Japanese and
Chinese are genetically unrelated, this means that the synchronic relationship
between the pronunciation of a native Japanese stem and the phonetic element
within the Chinese graph used to write it, if this is a compound graph, is
totally random.
The very complex
Japanese writing system contains more than the writing of native Japanese
lexical stems, but this is an important subpart of the writing system as a
whole. In the 1946 codification of
officially-sanctioned current graph-uses, graphs used for native stems
accounted for 1116 out of a total of 3122 morphemes written with Chinese
graphs. This establishes that a
method of writing which does not exploit phonetic relationships is possible,
whatever one’s view of the status of modern Chinese script.
6. Conclusion.
I conclude that the
diversity of the world’s writing systems, viewed as synchronic systems, is
greater than several recent writers have supposed, even though practical
considerations may lead to somewhat less diversity than is hypothetically
possible, and scripts may be less diverse with respect to their historical origins
than with respect to their current functioning.
REFERENCES
Bright, William, ed. (1992). International
Encyclopedia of Linguistics (4 vols.). New York:
Oxford University Press.
Burling, Robbins (1992). Patterns of Language:
Structure, Variation, Change.
San Diego, Calif.: Academic
Press.
Carello, Claudia, et al. (1992). ‘Can theories of word recognition
remain stubbornly nonphonological?’
In Frost & Katz (1992).
Coe, Michael (1992). Breaking the Maya Code. n.p.: Thames and Hudson.
Coulmas, Florian (1992). Entry on ‘Writing Systems’. In Bright (1992), vol. 4, pp. 253-7.
Daniels, Peter (1991). Review of Wayne Senner, The
Origins of Writing. Language 67.835-42.
DeFrancis, John (1984). The Chinese Language: Fact
and Fantasy. Honolulu: University of Hawaii Press.
DeFrancis, John (1989). Visible Speech: the Diverse
Oneness of Writing Systems.
Honolulu: University of
Hawaii Press.
Diringer, David (n.d.) [published 1948]. The
Alphabet: a Key to the History of Mankind. London: Hutchinson.
Frost, R. & L. Katz, eds. (1992). Orthography,
Phonology, Morphology, and Meaning.
(Advances in Psychology, 94.)
Amsterdam: North-Holland.
Gelb, I.J. (1963). A Study of Writing, 2nd
ed. Chicago: University of Chicago Press.
Hockett, Charles (1955). A Manual of Phonology. Memoir 11 of the International Journal of American Linguistics. Baltimore, Md.: Waverly Press.
Hockett, Charles (1973). Man’s Place in Nature. New York: McGraw-Hill.
Karlgren, Bernhard (1957). Grammata Serica Recensa. Bulletin no. 29, Museum of Far Eastern
Antiquities, Stockholm.
King, Brian (1991). Review of DeFrancis (1989). Language 67.377-9.
Knowlson, James (1975). Universal Language Schemes
in England and France 1600-1800.
Toronto: University of
Toronto Press.
Krippes, Karl (1990). Review of DeFrancis (1989). General Linguistics
30.126-9.
Liberman, Alvin (1992). ‘The relation of speech to reading and writing’. In Frost & Katz (1992).
Martinet, André (1949). La double articulation linguistique. Travaux
du Cercle Linguistique de Copenhague 5.30-7.
Mattingly, Ignatius (1992). ‘Linguistic awareness and orthographic form’. In Frost & Katz (1992).
Sampson, Geoffrey (1985).
Writing Systems.
2nd
ed., London: Hutchinson, 1987.
Sampson, Geoffrey (1991). Review of Reading the Past,
ed. by C.B.F. Walker et al. Times Literary Supplement, 6 September 1991,
p. 27.
Sampson, Geoffrey (1992).
Entry on ‘Statistical Linguistics’.
In Bright (1992), vol. 4, pp. 74‑6.
Sasanuma, S. & O. Fujimura (1972). An analysis of writing errors in
Japanese aphasic patients: Kanji vs. Kana words. Cortex 8.265-82.
Tzeng, Ovid (1991). Review of DeFrancis (1989). Contemporary Psychology
36.982.
Wrenn, James (1990). Review of DeFrancis (1989). Journal of Asian Studies
49.102-3.
[1]I would prefer to cite
the 1987 edition of this book as its standard version, since the original 1985
edition (together with all American printings) contained a number of errors due
to the fact that I was not shown complete proofs before publication. But I believe none of the errors
corrected in the 1987 edition, which were concentrated in chapters 5 and 6, are
relevant to DeFrancis’s disagreements with me.
[2]DeFrancis urges
(1989: 57) that a discussion of semasiography is as out of place in a
survey of writing systems as would be an introductory section on ‘Oxcarts of
the World’ in a history of the motor car, suggesting that it is self-evident
that these two things have nothing to do with one another. But, although the ox-cart in particular
was not a direct forerunner of the car, vehicles drawn by animals certainly
were: one could not hope to
understand the early evolution of motor-car design without knowing about the
structure of horse-drawn carriages from which cars were developed. Such questions about relevance must be
settled empirically rather than aprioristically.
[3]In discussing my views,
DeFrancis is selective in his use of quotation. He describes semasiography (DeFrancis 1989: 34-5) as
‘writing that Sampson would have us believe might be capable of evolving into
“a full-fledged semasiographic language rivalling English, French, and German
in expressive potential” (Sampson 1985: 32)’. What I wrote at the place cited was ‘Doubtless it is hardly
likely to lead to the evolution of a fully-fledged semasiographic language
rivalling [etc.] ...’.
[5]Since the matter has
become unexpectedly contentious, I should add that the document does include
some writing of the ordinary kind.
The whole message is headed ‘TO START’, and the two rows of pictures are
respectively captioned ‘COLD’ and ‘WARM’ (in ten languages in each case). But these fragments of glottographic
script add up to far less than is conveyed semasiographically.
[6]This point follows
straightforwardly from the standard account of the history of Chinese
phonology, as represented in Karlgren (1957); and, with respect to the
differences between modern pronunciations and those of the Middle Chinese
(Karlgren’s ‘Ancient Chinese’) of ca A.D. 600, this account is solidly based on
several different categories of evidence.
With respect to the reconstruction of the Old Chinese or ‘Archaic
Chinese’ pronunciations of ca 1000 B.C. the evidence is less solid, and there
is a risk of circularity in relying on the standard reconstructions in
connexion with a discussion of the nature of Chinese script, since the forms of
the written graphs are a chief category of evidence used in reconstructing Old
Chinese pronunciations. But the
point above about relative lack of homophony in earlier forms of spoken Chinese
requires only acceptance of the Middle Chinese reconstructions.
[7]This second point is
harder to establish irrefutably, since we inevitably have little direct
evidence of linguistic patterns in the speech of learned Chinese at early
periods. But consider for instance
the gu wen (‘old style’) literary movement initiated by Han Yu ca A.D.
800: this represented a revolt
against a prevailing literary style that seems to have been perceived as
artificial because it lacked the organic quality of writing which followed
spoken norms.
[8]Modern spoken Chinese
uses a smaller stock of morphemes than literary Chinese, its vocabulary has
developed in such a way that it uses a longer sequence of morphemes to express
a given proposition, and it is less flexible than literary Chinese with respect
to permissible ways of combining morphemes: these factors between them explain why spoken Chinese is
comprehensible while literary Chinese read aloud is not.
[9]My use of the term
‘double articulation’ leads DeFrancis (1989: 253ff.) into a discussion of
André Martinet’s individual theories about linguistic structure. But the concept is, so far as I know,
shared by linguists of all theoretical persuasions. I used Martinet’s term in Sampson (1985) not to imply
allegiance to one individual’s theories but because it is the classic label for
an uncontroversial idea. An
American term for the same idea is ‘duality of patterning’ (Hockett
1973: 106).
[10]Chinese dictionaries give
long lists of subsenses for many graphs, without making any differentiation
between cases where alternative subsenses are polysemous extensions of single
etymological senses, and cases where historically unrelated homophonous etymons
have been assigned the same graph.
The distinction between polysemy and homophony is not a clear one in the
Chinese philological tradition, because the written graph is felt to be the
essence of a morpheme ã if two homophones are written with the same graph then
they are regarded as the same word.
In English, ear of corn is popularly taken to be a meaning-extension of ear as organ of hearing;
experts can use comparison with other Germanic languages to go behind the
spelling and establish that the two senses are in fact a case of accidental
homophony, but comparable evidence is almost entirely lacking for Chinese. One strategy available to someone wishing
to interpret Chinese writing as predominantly phonographic might be to argue
that more of the subsenses listed for individual graphs are accidental
homophones and fewer are cases of polysemy than is commonly recognized (so that
graphs would not be in a one-to-one but in a one-to-many relationship with
morphemes). However, although this
may well be true, it would not do much to establish that the present-day
synchronic functioning of the system is phonographic, since typically when a
graph has a long list of subsenses in a large dictionary most of these are
obsolete. And I would add that in
any case unanswerable questions about the prehistory of etymons 3000 years or
more ago must surely be irrelevant to judgments about the synchronic properties
of a linguistic system: if Chinese
speakers in the historical period have regarded the various senses associated
with a single graph as subsenses of the same morpheme, then that fact alone
perhaps requires us to say that (provided there is no pronunciation difference
as in the ‘wonderful’/‘odd number’ case above) they are subsenses of one morpheme.
[13]Robbins
Burling has pointed out to me that any method for quantifying the degree to
which a script is phonographic would need to give separate figures from
writer’s and from reader’s points of view. Suppose the phonemes of a spoken language were in a
one-to-many relationship with the letters of an alphabet, in such a way that
the spelling of any word was a perfect predictor of its pronunciation, but the
choice among alternative letters to represent a given phoneme had to be learned
for each word individually. Such a
script would be 100% phonographic for the reader, but only somewhat
phonographic for the writer.
Intuitively it seems true that Chinese script is more phonographic for
reader than for writer; and it may be that scholarly discussions of
script-types have tended to give undue emphasis to writer’s rather than
reader’s point of view. Even from
the reader’s viewpoint, though, I would judge Chinese script to be ‘not very
phonographic’.
Aucun commentaire:
Enregistrer un commentaire