VMs: Re: Numbered transcription

[Jacques Guy <jguy@xxxxxxxxxxxxxxxx>]

02/10/02 00:31:02, Jorge Stolfi <stolfi@xxxxxxxxxxxxx> wrote:

>Hi all,
>
>EVA is
>even compatible with some alternative views, e.g. that <a> is the same
>as <ci>, or that <ee> is a single letter (just use bitrans).

You don't _need_ bitrans for that. Imagining <ci> when you see <a>
or <A> when you see <ee> is so trivial that you don't 
even need your brain. The brain of an ant will do quite
adequately. And if, given a corpus in EVA, you want to compute
statistics with <a> reinterpreted as <ci> and <ee> as <A>,
it is, again, the most trivial of things for anyone who can
write C, Pascal, Basic, Fortran, or the language I have 
switched to for development speed (Euphoria--it's interpreted,
but fast).

>So,
>again, I think I will stick to EVA for the time being --- possibly
>until the "code" is cracked, or at least until we feel more confident
>about which finer distinctions are real. Needless to say, to address
>this issue we need at least some partial transcriptions with much
>finer categories than EVA --- such as the one which Glen is building.

What we need is quality photographs. But even in their absence, we
are immensely better off than the poor sods who hack away at
the Easter Island hieroglyphs. They have nothing truly reliable
at all to work from.

>In any case, much of the Voynichese which I have too look at is
>in odd places --- in scripts and programs, in tables, in intermediate
>data files, in debugging messages --- were one cannot expect it to be
>properly XML-formatted and fontified.

How do you produce statistics of a corpus of properly XML-formatted and
fontified piece of text? How sooner would Michael Ventris had cracked
linear B if he'd been working on an XML-formatted corpus of it? Answer:
he'd still be arguing about tags with John Chadwick.