Общество "Россия-Япония":
http://ru-jp.org/

Evgeni Mayevski
Doctor of Cultural Studies, Head of Japanese Language and Literature Chair, Institute of Asian and African Studies, Moscow State University

IDEOGRAPHY IN THE JAPANESE LANGUAGE

The Japanese writing system is traditionally interpreted as a symbiosis of two different sets of characters: kana, usually described as phonographic (or phonetic) symbols (in Japanese, hyoon moji), and kanji (Chinese characters), usually labeled ideographic (hyoi moji) or, in more recent times, logographic symbols (hyogo moji). This distinction is universal and can be also applied (although in somewhat restricted sense) to other scripts and languages, e.g., to the set of symbols on the standard alphanumeric keyboard: letters are phonograms, while figures, ampersand, yen/pound/dollar signs and some other symbols can be regarded as ideograms or logograms. In the theory of writing, alphabets and syllabaries are on the whole classified as phonography, the set of Chinese characters (together with such now extinct sets of symbols as Egyptian or Sumerian) as ideography or logography. The dichotomy is considered fundamental because it has to do with a most basic structural feature of human language-so-called double articulation, or the contrast between meaningless and meaningful units, sometimes also called cenemes (empty units) and pleremes (full units). In Japanese linguistics, this traditional "bimodal" (Unger 1996: 10 - 12) view is predominant.

The term "logography" is said to have been coined more than century and a half ago by Du Ponceau but has come into common use comparatively only recently. It was revived by P. Boodberg (1937) and popularized by I. Gelb (1952) as a handy alternative to the term "ideography" which is regarded by many scholars as misleading. Fighting the Ideographic Myth has long become a respectable business with Western Sinologists and Japanologists. Many are convinced that now, in the new millennium, no hope remains for the Ideographic Myth; it has been expelled in disgrace and dispensed with forever. Perhaps on the whole this is true. In any case, one rarely finds the term "ideography" in major works on writing systems published in the last ten or fifteen years; instead, the term "logography" (as the opposite of "phonography") has become standard or at least preeminent.

Recently, J. DeFrancis and J. Marshall Unger (1994) came up with a "unitary" (as opposed to "bimodal") view of writing system typology. They have shown that no actual writing system usually described as phonographic is purely phonographic, and likewise no writing system classed in textbooks as logographic fully deserves that name. They have made it evident that there is no gap between phonography and logography and that actual writing systems form a kind of continuum (Unger 1996: 12 - 13). I support these opinions and will develop them. But in these works too, as one could well expect, the notion of ideography is treated with something close to contempt.

I am inclined to think that this notion is not completely useless. I believe that Japanese studies are the most probable area to benefit if the notion of ideography is applied, although in some limited sense it could be beneficial also for the study of other systems of writing. I do not think that the Ideographic Myth deserves being reinstated in full glory, but I frankly admit that I would not mind slipping this loathed monster in through the back door.

Ideographs, or ideograms, can be defined as written characters that "symbolize distinct thoughts or ideas irrespective of how those ideas are expressed in any particular language" (Unger 1996: 4 - 5). This does not mean that the notion of ideography presupposes that thoughts somehow can be expressed without any language at all: it is universally recognized that they cannot. To modern linguists, the trouble with ideography is in another presupposition-namely that thoughts can be expressed in writing without mediatorship of speech. In Japan, both laymen and scholars often think and say that each kanji has its own meaning and that this meaning is somehow different from the meaning(s) of the corresponding spoken word(s) or the same word(s) written in kana or romaji. This remarkable quality of Chinese characters (no matter whether real or imaginary) is just what made linguists call them ideographs. But if the meanings of characters do not coincide with the meaning of the spoken units they stand for, then characters make a language in its own right, in spite of the fact that it is only written and not spoken.

Now to some people this is an abhorrent idea. It has been repeatedly and insistently declared that "strictly speaking, language is-strictly-SPEAKING" (Unger 1996: 9), that "although some scholars speak of 'written language' as an autonomous system that largely overlaps with but is distinct from 'spoken language,' it is more realistic to see writing not as language but as a technology for recording language"(Unger 1987: 196), and that the claim of such linguists as Suzuki Takao that the Japanese writing system is an integral part of the Japanese language "is no more compatible with the findings of linguistic science than creationism is with modern biology" (Ibid.: 201).

Every linguist knows that in some respects speech is undeniably primary and writing secondary. In some other respects, however, just the opposite may well be true. Those who deny this fact should look into the problem of mathematical notation. Are we willing to admit that a mathematical formula does not mean anything by itself but obtains its meaning only from the words that stand for it in speech? Anyone with even an initial command of mathematics knows that except for the simplest everyday use of figures this is not the case. If formulae are vocalized at all, most of them can be read in several different ways. But they are very seldom vocalized because vocalization as a rule renders them unnatural, uneconomical and simply incomprehensible. For mathematical notation, the written form is the original and almost the only possible form of existence.

The problem of the primacy of speech remains open for discussion. But I cannot help feeling that at least for the needs of linguistic description the approach to language as a union of two parallel, partially overlapping sublanguages, spoken and written, is more convenient than the view of writing as something subsidiary to language. It just makes description simpler. A few Japanese examples (undoubtedly not novel to the adversaries of the Ideographic Myth) could illustrate that.

Current Japanese orthography still cannot do without ateji and jukujikun, i.e. kanji spellings assigned to spoken words on a non-morphemic basis, according either to the sound or to the meaning of the whole lexical unit, with no respect to its morphemic structure. One of those spellings is 火傷 for yakedo 'burn (injury)'. The two kanji used here are respectively 'fire' and 'wound'. Neither of them corresponds to any part of the spoken word yakedo, though the latter also has two meaningful components, the first of which, yake, can be quite positively interpreted as 'burning'; the meaning of the second one, do / to, is less clear but some lexicographers believe it is related to tokoro 'place' (Kokugo Daijiten 1989: 2366). Needless to say, 'burning' is not exactly 'fire', and 'place' is not 'wound'.

If kanji do not have meanings of their own, are we supposed to think that the kanji and in the combination are meaningless units, mere cenemes? This view is hardly tenable because the word in question is a jukujikun, not an ateji. Kanji may be seen as devoid of meaning in such words as Kubo '(surname) Kubo' or Nara'(place name) Nara', where they are used as syllabic signs, in contrast with their logographic function in kubo 'hollow' or nara 'oak', though even in such obvious cases of phonetic use a way is open for a secondary semantic interpretation. But in the kanji obviously are picked up for their meaning rather than for anything else.

Are we, then, to believe that the meanings 'fire' and 'wound' do not belong to the characters as such but are somehow borrowed from the spoken morphemes ka / hi (both mean 'fire') and sho / kizu (both mean 'wound')? This seems quite plausible, but the problem is that in this particular case (as also in other examples of jukujikun) the kanji in question do not read as either ka or hi, sho or kizu, although they have those readings in many other contexts. So if their meanings are borrowed, they still show an amazing ability to retain what has been borrowed without any immediate connection (or at least with one link less) between the written character and the spoken morpheme. You cannot even tell which morphemes were the source of the borrowed meanings-ka and sho or hi and kizu-since the on and kun readings are in this case (like in most cases) more or less synonymous. Doesn't it amount to admitting that kanji do express ideas with some degree of independence from spoken words?

The same effect can be observed when a spoken lexical unit is represented not by a combination of kanji but by a single compound kanji. This is very common both in Japanese and Chinese; some good examples are provided by what is usually called kokuji, i.e. kanji of Japanese coinage, such as the character 峠 for toge 'ridge, mountain pass'. From the synchronic point of view, the spoken word toge must be regarded as consisting of a single morpheme, though historically it is a modification of a two-morpheme unit, tamuke, where ta / te is 'hand' and muke is 'direction' (Kokugo Daijiten 1989: 1748). Not so the kanji-it consists of three clearly meaningful parts: 'mountain', 'up', and 'down'. Another kokuji of similar structure comes to mind at once: the character that stands for kamishimo 'kamishimo (an item of traditional clothing)', but this example misses the point because this kanji does reflect the morphological structure of the spoken word (kami means 'upper', shimo 'lower'), while the kanji for toge does not. However, there is one more character built after the same pattern; it is nonstandard and should be looked upon as a case of wordplay rather than a regular unit of the kanji inventory, but in Japan almost everybody knows it. The right part of that character is the same that in the previous two: 'up' and 'down', but the left part is different: 'woman.' To my knowledge, several different readings have been ascribed to this character, of which erebeta garu 'elevator girl' is the most innocent. True, this is a very unusual "reading"-loanwords are not, as a rule, given to kanji as kun readings, but then both this kanji as a graphic unit and the way it is supposed to be read are playful, not serious. Does this mean that examples like this should be ignored? On the contrary: wordplay is an area where language lives and develops, so to speak, more intensely than in communication proper.

However, one does not have to resort to such exotic or peripheral phenomena as jukujikun or kokuji to substantiate the claim that Chinese characters express ideas with a certain degree of independence of spoken words, or not quite the way spoken words do. It would be sufficient to point out that in Japanese, kanji have multiple readings. True that in kaji 'fire' the first character has a very definite reading, in hibana 'spark' a different but no less definite one, and we cannot say that it is not associated with any pronunciation even in the limited sense in which is has no reading in 'burn'. However, a character representing a chain of synonyms is not the same thing as a character representing one and only one spoken unit. Such eloquent (although perhaps exceptional) cases as the character with its 20 or so Japanese readings, all of them close or related semantically, give one a strong feeling that the link between writing and thought is much more palpable here than the link between writing and speech.

In fact, it is not even necessary to bring forward Japanese data. The millennia-old Chinese practice of discriminating homonyms (and different meanings of polysemous words) in writing, fully retained and further elaborated in Japanese, is a good proof of (or at least a clear hint at) the fact that writing can be semantically different from speech and richer than speech. Writing, as the Chinese literati always understood it, makes meaning clearer, provides finer semantic distinctions. It is true that the use of kanji in Japanese displays a greater independence of the written sign than anything observed in Chinese; however, the Japanese merely elaborated on what the Chinese had invented long before them.

But I am also willing to admit that Chinese "ideographs" (in their Chinese, or Japanese, or Korean usage) are not the only written symbols to display such independence. Many similar examples can be given that involve kana, Latin letters and other phonetic symbols, and such examples can be found in any language. Nothing prevents phonetic symbols from having meaning in combination or even as isolated units, just as nothing prevents phonemes and syllables from constituting morphemes and words (that is just what they are made for). And in all languages, including those normally written in syllabic and alphabetic symbols, the meaning expressed in a written utterance sometimes may differ in one way or another from the meaning that can be conveyed through or extracted from the exactly corresponding spoken utterance. Japanese is by no means unique (though perhaps more advanced than other languages) in this respect.

Let us bring into discussion such cases as "special" spellings of many Japanese words based on the use of hiragana and katakana as opposed to kanji. Some observant students of Japanese have shown that a word tends to be spelt in kanji when it is stylistically neutral and in hiragana when it has colloquial connotations (Backhouse 1983). It has been noticed that the Japanese word onna 'woman' (normally written with the kanji ) is often spelt in hiragana, i.e., when it is used in the sense of "sexual object", and like any other word it can also be spelt in katakana when there is need to mark emphasis, some special meaning or a foreign accent. It would be very strange if the title of Gilbert and Sullivan's operetta Mikado was rendered in Japanese translations from European languages otherwise than in katakana, i.e. ミカド. A more Japanese-like and serious-looking spelling of this title-for instance, with the kanji - would in this case be quite inappropriate, and I am sure it would be a great shock to Japanese mentality if the Japanese culture were suddenly deprived (for instance, by the abandonment of kanji-kana-majiri-bun in favor of Latin script) of its numerous tools for graphic discrimination of sacred and profane things. Orthographic subtleties of this kind are widely exploited in present-day Japanese fiction, advertizing, comic books and informal everyday writing.

All this proves that writing, in a certain sense and to a certain extent, often does express ideas in a different way than speech does, and that the process of substituting written units for spoken ones (and, nota bene, vice versa) is very close to the process of translation (cf. Haas 1970: 16-23).

In the history of the Japanese language (including writing), there were so many remarkably different styles and orthographies that in the Heian and later periods "people often did not really know what language they were writing in, Chinese or Japanese; and we are often in no better position to make a judgement on the question when we study some of the documents they produced" (Miller 1967: 131). Even today, it would be a gross distortion of truth to say that the Japanese language as it is represented by kanji-kana-majiri-bun is strictly Japanese. It is of course Japanese in its morphology and syntax. But as to its vocabulary-well, one has doubts. Isn't this Japanese language rather similar to Classical Chinese? I am not referring to such obsolete phenomena as kanbun or bungotai. I am not even speaking of kango. Let us limit the field of investigation to wago. When instead of one word miru 'look, see' you have five or ten different miru with the root mi represented by different kanji depending on the shade of meaning because five or ten different morphemes used to express the corresponding ideas in the Chinese language of ten or fifteen centuries ago, then you can't help feeling that there are, in a way, two non-identical languages, both called Japanese due to some bizarre convention.

It should be emphasized that at least from the point of view of synchronic description the relation between those two languages (as well as between speech and writing in general) is symmetrical. This reminds us of the much-quoted utterance attributed to Warren Weaver, one of the fathers of machine translation: "When I look at an article in Russian, I say, 'This is really written in English, but it has been coded in some strange symbols. I will now proceed to decode.'" This formula is not entirely wrong, though of course it is an oversimplification, because translation from Russian into English (or vice versa) is not just a matter of one-level unit-for-unit substitution. What makes one feel that something is not quite right here is the stance of the speaker: it seems to be somehow presupposed that normally any text must be in English. The Russian words we see are taken as the result of some replacements that have been made and now must be undone. But to native speakers of Russian, it is English text that is "coded" or "encrypted." The English way of expressing ideas is no more natural or primary than the Russian. Likewise, in a certain sense the manner in which speech expresses ideas is no more natural or primary than the way writing does.

There are therefore few reasons to assert that "... the function of kanji in the current Japanese writing system is to replace strings of kana" and that "[t]he historically accumulated heuristics underlying the conventional rules that prescribe how and when such replacements should be made are neither a part of the Japanese language nor a necessary part of the knowledge that enables literate Japanese to read and write" (Unger 1996: 23). Kanji are not an extraneous embellishment that must be added to perfectly complete utterances God knows why and can be thrown away without any consequences. Choosing one kanji instead of another with the same reading makes a difference. Dropping a kanji in favor of a kana string makes a difference. The Japanese think (to a large extent) in kanji.

There were and still are many cases when the kanji is clearly the linguistic sign and the corresponding spoken word (morpheme) a mere shadow. When new kango were being coined by the hundreds in the Meiji period, almost no one gave any thought to their sound. The communication on learned topics in educated circles was held mostly in printed or at least written form. The important thing was what is seen, not what is heard. That is the main reason why there are so many homophones in modern Japanese. In due course, with the advent of radio and compulsory mass literacy, it was discovered that large parts of Japanese vocabulary, both traditional and post-Meiji, cause trouble, being poorly discernible by ear. By the end of the twentieth century many such words of "shadowy" nature quietly sank into oblivion, replaced to a certain degree by Western loanwords of the same or close meaning. The abandonment of the characters not included in the Toyo Kanji list was one of the powerful factors in that process. Can we say that the disuse of those characters did not influence the Japanese language? Can anybody really prove that the language remained exactly the same, or that if it did change the absence of the excluded characters played no part in that change altogether? But if writing in general and kanji in particular are that important, shouldn't we include them into the language on a par with speech and spoken entities after all?

Incidentally, all this leaves very little ground for the enthusiasts of script reform in Japan. The Japanese, en masse, are not interested in a script reform because they are prosperous (at any rate, more prosperous than a great many other nations). They can give their children a good education. They are rapidly advancing word processing technologies.They can afford to keep a costly writing system, and they want to keep it for the simple reason that mountains should not be moved if you can do without moving them.

The difference between the spoken and the written forms of a language cannot be reduced to the simple scheme where a form unit (upper-level unit) is embodied in different substance units (lower-level units): one set of units in the spoken language, another in the written. As a rule, there are also many instances when one spoken unit corresponds paradigmatically to two different written units, as in [teil]-tale, tail, and vice versa. This is where the notion of ideography comes in handy. Such typically Japanese cases as the character 'life' with its many readings cannot be described as logography because there is no one-to-one correspondence between the character and the spoken morpheme or word. I think it is quite possible to say that this character conveys the idea of life with no or little respect to how this idea is sounded. This particular case is of course exceptional, even in Japanese. There is probably no such thing as pure ideography (save the very special case of mathematical notation and the like). But Japanese writing is certainly more ideographic than Chinese. It has advanced rather far towards the ideographic extreme, although there are doubts as to whether there is any such extreme.

BIBLIOGRAPHY

Backhouse A.E. 1983. "The Expressive Stratum in Modern Japanese." Gengo kenkyu, no. 83: 61-78.
Boodberg, Peter A. 1937. "Some Proleptical Remarks on the Evolution of Archaic Chinese." Harvard Journal of Asiatic Studies, no. 2: 329-72.
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.
DeFrancis, John, and J. Marshall Unger. 1994. Rejoinder to Geoffrey Sampson, "Chinese Script and the Diversity of Writing Systems." Linguistics, vol. 32, no. 3: 549-54.
Du Ponceau, Peter Stephen. 1838. Dissertation on the Nature and Character of the Chinese System of Writing: In a Letter to John Vaughan. Transactions of the Historical and Literary Committee of the American Philosophical Society, vol. 2. Philadelphia: M'Carty & Davis.
Gelb, Ignace J. 1952. A study of writing. The foundations of grammatology. London: Routledge & Kegan Paul.
Haas, William. 1970. Phono-Graphic Translation. Manchester: Manchester University Press.
Kokugo Daijiten. 1989. Tokyo: Shogakukan.
Miller, Roy Andrew. 1967. The Japanese language. Chicago and London: University of Chicago Press.
Unger, J. Marshall. 1987. The Fifth Generation Fallacy. Why Japan Is Betting Its Future on Artificial Intelligence. New York-Oxford: Oxford University Press.
Unger, J. Marshall. 1990. "The Very Idea: The Notion of Ideogram in China and Japan." Monumenta Nipponica, vol. 45, no. 4: 410-411.
Unger, J. Marshall. 1996. Literacy and Script Reform in Occupation Japan. New York-Oxford: Oxford University Press.