Richard Cook wrote:
>
> Well, not all components are Kang Xi radicals.
>
Right, some are variants and others are characters in their own
right. Practically all of the components are classified to a specific
radical, though.
> What you're talking about is not a Kang Xi index, but a complete
> component index, and this is not quite the technical feat you imagine.
>
This type of index is handy from a font development standpoint.
I've tried to set up a database for Plane Two characters from
scratch. It's very time consuming and tedious. (Almost done
through four residual strokes after a considerable amount of
time and effort, and am only counting residual strokes based
on the significant radical.)
> E.g., I have complete component and rad/str data that is lexicon
> specific (Shuowen), and somewhat less complete general data.
>
> The most comprehensive collection of such data is from
> http://www.wenlin.com . When compiling wish lists, watch Wenlin's development.
>
Wenlin is a great resource. For me, it's good practice to try
to find a character first, but have been stumped several times and
gone to Wenlin.
> >
> > The Han Radical Index is set up for people familiar with CJK,
> > the rest of us will just have to guess (and learn something during
> > each look-up process, I'd suspect.)
> >
>
> Even for experts,there are cases in which the choice of a single
> classifier is completely arbitrary, or at least apprently so to the
> casual user. In these cases, putting the character under both is a good
> idea.
>
An ideal index for the casual or non-CJK user might be quite
different in approach. Perhaps the first component drawn in
any character would be a good basis for indexing rather than
the significant radical. But, as you've pointed out, not all
components are immediately recognizable as pertaining to a
particular radical, especially to the casual user.
Since the Han Radical Index in the Unicode book functions the
same as a radical index in a Chinese dictionary, I balked at the
idea of having any character appearing under a non-significant
radical, at first. But after more thought, I don't suppose any
laws would be violated if characters appeared in more than one
location. It sure would make look-ups easier. Meanwhile, I'll
take your advice and keep an eye on Wenlin.
Best regards,
James Kass.
