On 11/10/2001 09:41:23 AM John Jenkins wrote:

>Meanwhile, it *does* bring up an interesting problem in script analysis

--

>trying to determine what visual features of a writing system are truly

its

>essential characteristics and which are just artifacts of the way the
>sample happened to be written.

Perhaps there's a good use for grunge fonts after all: they can be used to
test how far a glyph can stray from a prototypical form of a letter
(probably an oldstyle font could serve as the prototype) and still be
recognisable. :-)

Of course, one would need to look beyond grunge fonts to the full range of
typographic expression a community may generate. There are potentially
complications galore to deal with, though. For instance, I'm sure we've
all seen examples of Latin typography in which some letters are drawn in a
visual negative space. One has to decide to what extent such mechanisms
are part of what is to be measured as opposed to being factors to be
controlled in the analysis.

Related to this is a question I've thought would make for interesting
research: where does a script carry its information load? For example, if
you masked that top half of every line of Arabic text and asked people to
read it, you'd probably find that more information was lost than if you
repeated the test with the bottom half masked since the connections along
the baseline are largely redundant information.

Related to that, I found it interesting when I was learning Thai and later
when I was learning about Ethiopic to see the information load in fairly
subtle details. For example, in Thai, the presence or absence of notches
and the orientation of small circles (on the inside or the outside of the
stem) is important. Similarly in Ethiopic, the particular type of flag can
be significant. These are pretty subtle details, for which scripts like
Latin have no comparison. One might expect such subtle details to
interfere with reading fluency. But then we're getting into reading
theory, and the relative importance of details of individual characters
versus recognition of character sequences as a whole becomes a factor.
It's all potentially quite interesting, I think, particularly given the
cross-disciplinary nature of the issues involved.


- Peter


---------------------------------------------------------------------------
Peter Constable

Non-Roman Script Initiative, SIL International
7500 W. Camp Wisdom Rd., Dallas, TX 75236, USA
Tel: +1 972 708 7485
E-mail: <peter_constable@...>