[This is a follow-up to Jon's 2000.10.19 post on the unicode mailing
list, which is also http://www.egroups.com/message/unicode/4287 . -TC]

On Thu, 19 Oct 2000, Jon Babcock wrote:

[MC]
> > Jon Babcock is satisfied to stop here, and indeed two
> > "holograms" can greatly reduce the number of characters needed.

[JB]
> Not two holograms usually, but two * hemigrams *, one of which is a
> 'hologram' (wen2) and other may be a 'hologram', but most likely it
> will be another digraph (zi4), capable of being split into two.

[TC]
I think the most examples of wen + zi cases, or zi + zi cases, may be
found in Vietnamese chu+~ no^m characters, which are mostly composed on
the signific-phonetic xingsheng principle, where each half is usually a
pre-existing character (in Chinese usage). You can see some in Ngo^
Thanh Nha`n's 1998 article "A Review of Dictionary Indexing and Lookup
Methods for Ideographic Scripts in Computer"
(http://www.cs.nyu.edu/~nhan/vsic98.pdf).


[JB]
> (But if someone can tell me why it matters, except in a very few cases
> (like mang 5fd9 and wang 5fd8), how the hemigrams are positioned with
> respect to each other, I would be eager to hear. The Chinese character
> can be identified regardless of the relative positioning, can it not?
> Or are we talking strictly about creating fonts?)

[TC]
I don't have an answer to this, although I can think of more examples of
exceptions, such as jie2, composed of ba1 U+5DF4 over shan1 U+5C71, Hanyu
Da Zidian 10765.11, which is 'the place where a mountain turns', among
three other meanings; vs. ba1, composed of shan1 U+5C71 over ba1 U+5DF4,
Hanyu Da Zidian 10765.12, which is part of a placename.

[TC]
However, let me quote an interesting passage I read today from CHIANG
Yee's _Chinese Calligraphy_ (London: Methuen, 1938), that seems relevant
to what Jon is asking:

(10) Chieh-Huan (U+501F U+63DB), Exchange. This principle, unlike
those prescribed so far, is not for general use. The sanction of
precedent--the precedent of some great scholar or calligrapher--is
almost obligatory. It consists of the transposition of elements in
a character for aesthetic reasons; and such a practice can obviously
only be followed with caution and great skill if confusion is to
be avoided.

Examples of the successful implementation of the principle are Su,
Chiu, and E^. Su is generally written, as in Fig. 130, A, with the
U+9B5A sign on the left, but the difficulty of arranging the right-
hand side, with its rather short horizontal strokes and long
sweeping rightward stroke, makes this composition awkward. So the
two sides are transposed and, necessarily, the sweeing rightward
stroke curtailed (Fig. 130, B). The same applies to Chiu, which is
commonly written as in Fig. 130, C, with the right-hand side
containing fewer strokes than the left and the character tending to
sprawl. If the two sides are transposed, the long rightward-sweeping
stroke of symbol U+79BE abbreviated, and that of the U+70AC symbol
lengthened (Fig. 130, D), the effect is more closely knit. E^ has
three variants. Fig. 131, A, is the regular form, but it is obvious
that the long hook swinging rightward in the left-hand part of the
character will tend to collide with the right-hand part and give a
tight appearance. Both the alternative compositions, Fig. 131, B
and C, obviate this contraint. Personally I prefer C, but there is
a danger in it of over-elongation.

Not much licence is permissible with the Chieh-Huan principle. We
prefer a character to be written well in a normal form to one in
which a new or eccentric composition is chosen in order to make the
balance easier. (181-182)

[TC]
The figures that Chiang refers to are:

fig 130a is U+8607
fig 130b is the same, but the two pieces in the bottom half are swapped

fig 130c is U+79CB
fig 130d is the same, but with the two halves swapped

fig 131a is U+9D5D
fig 131b is the same, but with the two halves swapped
fig 131c is like fig 131a, but arranged vertically rather than
horizontally (so that 'I' is above 'bird')


Thomas Chan
tc31@...