[RNLD] handwriting recognition software?

[Bill Poser]

As Claire says, I've never heard of handwriting recognition software that
is likely to be at all reliable for a language it hasn't been trained on. I
also suggest having it transcribed by two different human transcribers. You
then run a comparison of the two versions (after appropriate normalization
so that you don't detect things like different line breaks), which will
give you a large percentage of transcription errors. This is the technique
used by large-scale digitization projects.

On Tue, Jul 10, 2012 at 10:44 AM, Claire Bowern <clairebowern at gmail.com>wrote:

>  Undergraduates.
> Seriously, there's so much contextualisation that goes on when reading
> handwriting, unless there's a machine learning model of the language, I can
> pretty much guarantee that an undergrad or two with some epigraphy training
> and a general idea of possible syllables in the language will do much
> better than any program.
> Claire
>
> On Tuesday, July 10, 2012 at 1:41 PM, James N. Stanford wrote:
>
>   Dear RNLD,
>
>  Does anyone have a suggestion about handwriting recognition software? An
> Abenaki student is trying to take 400 pages of old documents of handwritten
> Abenaki and turn them into text, i.e., from a scanned handwritten PDF to an
> editable Word document. It's written in a Roman alphabet script.
>
>  Any ideas?
>
>  Thanks,
> Jim
>
>
>   ==================
>
> James N. Stanford, Ph.D.
> Assistant Professor of Linguistics & Cognitive Science
> Dartmouth College
> 305 Reed Hall, HB 6220
> Hanover, NH 03755
> (603)646-0099
>
> www.dartmouth.edu/~jstanford <http://www.dartmouth.edu/%7Ejstanford>
>
>
>
> e-mail: James.N.Stanford at Dartmouth.edu
>
> or: stanfo23 at gmail.com
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/resource-network-linguistic-diversity/attachments/20120710/0070c08f/attachment.htm>