[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: VMs: Babelfish translation [plus question for Dana]

Hi Jeff,

At 22:23 13/11/2003 +0000, Jeff wrote:
PS I did retrieve a phrase that babelfish translated as "I am devoted to the
beautiful one." Any ideas what this might possibly mean?

Sadly, it probably means that babelfish is often rubbish. :-(

Jeff - please understand that there is a *big* difference between EVA and the glyphs you see in the scans - not only is <ch> almost certainly a single glyph, but so is <sh> (though consistent, EVA's reuse of the <s> character within <sh> has been [IMO rightly] criticised).

While I honestly have no idea whether you are trying the right kind of approach or not, I'm quite sure that while you base it on an EVA transcription, you won't find out. Look: <s> in EVA is manifestly a different entity in <s> and <sh>, so your tables really shouldn't be treating them as one thing (but they currently are).

A reasonable (yet fairly minimal) extended glyph set might well be based around:-

Similarly, many read the <in>/<iin>/<iiin> strokes as a single glyph, so these too might well be fruitfully considered as single glyphs.

IMO, there's also a strong case for treating these pairs as if they formed individual letters:-

Also: you may want to consider using a different line of text as your testbed. Philip Neal and others (myself included) have pointed out numerous times that the two single-leg gallows (<p> and <f>) are often found close to incongruous-looking pieces of text, especially on the top line of pages. Also: the transcription on f1r isn't really as clear as it could be - the page itself appears worn out, probably from having been on the outside of the Ms for five centuries. :-)

So... I'd suggest finding some clearer text to work from (and avoiding the topmost line if possible), perhaps on a herbal page where there is a reasonably good match with a known plant. However, these two criteria overlap incompletely - for example, f11r has extraordinarily clear text, but f2v has an excellent plant identification ("nanufar", Nymphaea alba).

Dana - as a general point, which herbal page would you suggest offers the optimal combination of textual clarity plus reliable plant identification?

Finally: Jeff, you're getting to the stage where doing this kind of thing by hand will consume all your time (and more). Please install Python on your machine and adapt Seth's script to do the things you want - really, you'll save yourself a *lot* of hassle.

Cheers, .....Nick Pelling.....

______________________________________________________________________ To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying: unsubscribe vms-list