Sorry about the HTML format
John Babcock-ek idatzi du:
>> Markus Kuhn wrote:
>> Only in
>> those rare situations where we encounter a word (such as a name) that we
>> haven't seen in a long time, we fall back into a character-by-character
>> reading mode where we actually process the script as a phonetic
>> representation (something the Chinese can't do in their writing system).
>In the case of the Chinese script too, it is possible to fall
>back into a grapheme-by-grapheme reading mode for a large
>majority of individual graphs -- perhaps more than 70% of the
>more than 50,000 graphs -- where the phonetic hemigram may be
>distinguished from the other, purely semantic hemigram, and
>which conveys at least roughly, the sound, and thus the
Not true on the first account and not quite on the second. I can only speak for Cantonese, but since cantonese has a smaller number of homophones, I doubt that the situation could be any better in Mandarin ...
These days most Cantonese/Mandarin dictionaries use a romanization system for giving pronounciation (Pinyin, Yale, Yùht Pèhng etc.) but there is/was another way which occasionally still can be seen in modern dictionaries and definitely in most old Cantonese/Mandarin character only dictionaries. It's called Chit Yām, 切音,"cut tone" and employs the C. classic notion of homophones and dividing charcaters up into initials and finals. If no direct homophone exists, two characters are employed, one to supply the initial (consonant) and the other to supply the tone and final. I have never read anything on this method and only know it from usage, but the number of charcaters used seems around maybe 100 - 200.
To give an example for all those who have a Cantonese/Mandarin unicode font, the homophones/characters pronounced 'yuht' would be transcribed as follows
wheras words containing the initial k- or the final -ūk would be transcribed as follows (the tone is supplied by the second character)
曲[卡屋切] k(a) + (ng)uk = kuk
咳 [卡乞切] k(a) + (h)at = kat
咯 [卡革切] k(a) + (g)aak = kaak
哭 [哈屋切] h(a) + (ng)uk = huk
and if nothing else helps, a homophone which ignores the tone is given and the needed tone supplied separately:
市[時低上] sih + low rising = síh
which is pretty phonetic I suppose and all in our own writing system. If you want to look it up, try the 新雅中文字典 for a modern one, I don't have my old one here at the moment, but I can look it up.
On the second point, I doubt that the number of characters whose pronounciation (at least in cantonese) can be derived from the phonetic component is anywhere near 70% ... I was taught the rule 有邊讀邊 "if you have a phonetic element, then read it" rule too. but that only applies for characters which are ... not too common and only as a last resort, because this methid is highly approximative.
>underlying word. It is only a tiny minority of graphs that can
>be properly called ideographs, perhaps less than 100 (I'm
>thinking of the 'zhishi' graphs, the 'dactyliograms') and not
>that many more that are purely pictographs or, 'zograms'
>(drawings from nature). Most Chinese graphs include a
that number varies greatly because if ALL characters that are theoretically in use are included that number rises and if only those in common use, the number diminishes, but let's not go into that.
>phonetic hemigram that conveys, if in a rather 'molecular'
>fashion, the sound. But, as you mention, this fall back is
>only used as a sort of last resort when one doesn't instantly
>recognize the word, or, in Chinese, the graph.
true, see above
>Incidentally, if could be pointed out that precisely this
>phonetic 'imprecision' had allowed the transmission of the
>Chinese script to proceed throughout vast areas of space and
>through many centuries of time relatively unscathed and so
>one can speak of such a thing as 'kanji culture'. The degree
>to which the sound of a word is accurately recorded in the
>script is the degree to which it is bound to a particular
>pronunciation and tied to a specific time and place. The
>hodgepodge that is English spelling may indeed be one, only
>one, of the reasons it is used so extensively throughout the
>world, i.e. one quickly learns to ignore the particular
>phonetic information that might be conveyed through the
>letters and to read it word by word, chunk by chunk, just as
>one reads Chinese, pronouncing it any way one wants.
>If anyone has any doubts about Markus Kuhn's analysis, I
>suggest they try to learn how to read katakana loan words in
>Japanese. The majority of these (and there are perhaps two
>thousand in daily use) come from English, but it is by no
>means an easy task for a native English speaker to become
>comfortable reading these 'English' words even though the
>katakana spelling is roughly accurate and more or less
>consistent. Why? Because the do not _look_ like English
true, it's highly entertaining to have to guess the meaning of Katakana words in class; my personal guess many years ago was that 'sandoichi' meaning 'sandwhich' meant 'three germans' since the word before has been 'kaate' meaning a sick card from the German 'Karte' .....
>See 'Chinese and Latin Scripts Compared', about half-way down
>the page on
><http://www.kanji.com/kc/cha/introduction-to-cha.html> where I
>have tried to make an interesting comparison between the two
>Jon Babcock <email@example.com>
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:51 EDT