Re: Japanese RTL (was RE: Mongolian (was RE: Syriac and Mongolian

From: Kenneth Whistler (
Date: Tue Jan 11 2000 - 19:43:34 EST

> Kenneth Whistler wrote:
> >
> > Andrea Vine asked:
> >
> > >
> > > I am not a native reader, though. I would be interested to hear from native
> > > Japanese and Chinese character readers as to whether they sense a
> > > directionality, or behave as though there is one. For example, in interpreting
> > > obscure characters, where do they first look for a helpful radical?
> >
> > The position of a radical in the character also likely has nothing to do
> > with text directionality. A majority of the common radicals are on the left side of a
> > character, but there are also many common radicals that occur on the
> > top, bottom, or right side of a character. They are learned by overall
> > pattern recognition that (I believe) is independent of left-to-right
> > or right-to-left directionality.
> >
> > --Ken
> Yes, Ken, but where does a _native_ reader look _first_? How does a native
> reader parse the character? Just because handedness dictated the writing
> directionality (universally, it seems) doesn't mean that it didn't provide an
> inherent reading directionality. These are not mutually exclusive.

Well, first of all, there really is no such thing as a "native reader".
Unlike spoken language, which can be acquired natively, or learned
(laboriously) later, everybody "learns" to read their writing systems by
a rather laborious process *after* they have acquired the spoken skills.
The native speakers just have it easier, since they don't have to
simultaneously learn the language that the writing system is being
used to write.

> And I wait to hear from a native about parsing directionality for an unfamiliar
> character. Familiar characters may well be read as a whole. Or not. But an
> unfamiliar character must be parsed, at least to look it up.

And I'm suggesting that for Han characters this is done by a process of
pattern recognition from a couple hundred memorized common radical templates that
stare every dictionary user in the face. In fact, it is something like
facial recognition. You can just see the grass radical glaring out at
you, like you recognize Uncle Jim or Aunt Sadie. So the radicals themselves
are familiar gestalts to be grasped, even when the character as a whole is
not recognized. I see no particular reason why these gestalts should have
to have a directional parsing characteristic. When the character has a radical that
doesn't fit the obvious patterns, then "native readers" are reduced to
laborious guess and search for which radical the character might be filed
under, just like the rest of us who learn the system from afar, as it

And lest you mistrust my intuitions on this, I just tried some
fieldwork with the college-educated native Chinese speaker (and reader)
who sits across from me:

Task: identify the radical in four reasonably common low-stroke characters whose
traditional radical for lexicographic filing is *not* an obvious pattern:

U+F95E tan1 "cinnabar; red"
U+4E43 nai3 "to be; but"
U+4E8B shi4 "affair, job, undertaking, etc."
U+4E5F ye3 "and; still"

Result: 4 for 4 wrong answers. The guess strategy she employed in each case
was try the first stroke (not always the left-most stroke, by the way),
which leads to an incorrect radical for the first
three, and to a stroke that isn't a radical per se in the 4th case.

In these kinds of cases, what people actually do is count the strokes when
they don't know the character, and then look it up in the stroke count
index of the dictionary, or if they *do* know the character (as for these
4 reasonably or very common characters) look it up by pronunciation.


> Andrea

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:58 EDT