Dr Melissa Terras has possed a very interesting question.
I do not know of any projects that encode the traces left by strokes. I am
intrigued by the example that was offered. The "L" has been stroked a lot
in my lifetime of penmanship and I am aware of the chance accidents of
time. I suspect it is possible to encode both the traces of strokes from a
pen and the keyboard strokes of an electronically produced text (such as
backspacing and overwriting). I do believe that the feature system could
be of some assistance. It would take me a while to work up an example.
Perhaps some quicker hands can forge ahead. In any case, there is the
begining of such "double encoding" (graphic markers with multiple phonetic
renderings) available from a talk given at the University of Alberta in
early 2001. See
Glad to discuss further on and off list.
Boy-scout-scholar for a day :)
> Hello All.
> I have a question: Does anybody know of any work that
> is being done on the markup of texts, not on a
> character level, but an individual stroke level?
> For example, the letter "L" has two strokes, each one
> which has a variety of properties that could be
> In case you think I'm as mad as a box of frogs, the
> reason for doing this is to provide the data to train
> a stochastic AI architecture to effectively "read"
> textual data. As a newbie to TEI I would like to ask
> if anyone (else) has come up with an encoding scheme
> to markup text in this manner, on a stroke by stroke
> Also, is there room for such an encoding scheme in the
> TEI? Where does "text" stop - on the character level,
> or below, on the stroke level?
> I'd be really interested in hearing your reponses.
> Dr Melissa Terras
> Royal Academy of Engineering
> 29 Great Peter Street
> SW1P 3LW
> Do You Yahoo!?
> Everything you'll ever need on one web page
> from News and Sport to Email and Music Charts
Francois Lachance, Scholar-at-large,
knows no "no exit" in a hypertext
every cul-de-sac is an invitation to turn