[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: VMs: 1st Impressions and baby steps
> > Barbara Blithered;
> > I hope that a Kober analysis will identify glyphs when they are
> > encyphering vowels, consonants, and the rules under when their phonomic
> > values change. This isn't decipherment as such, because the CV values
> > will be assigned in the abstract. It'll produce plausable CV
> > constuctions of the VMS text just as this paragraph's first dozen words
> > can be can be represented thus;
> > VV CVVC CVC V CVVCV VCVCVCVC CVC VVCVCCVCVV CCVCC CVC CV VC...
> Dennis Dotted;
> We've used the Sukhotin vowel recognition algorithm,
> which Bruce mentioned, to identify the vowels in the text. I can't
> find it, but I believe Gabriel Landini did some recent work on
> this; ask him. It would make a big difference for your work! Of
> course, it does depend a lot on which transcription system you use!
> Gabriel looked at that too, which is why I mentioned it.
The *logic* of the algorithm may be useful but as I plan to work
directly from the VMS text without transcription systems (or invent my
own) I doubt it'd be very useful. The logics for assigning C & V values
are not too difficult.
A stand alone letter (lets say X) is most likely a vowel as single
consonants are very difficult to pronounce. Thus when you come across a
word with a construction of XYXY it can not be VVVV or CCCC, it has to
be either CVCV or VCVC, this makes the most likely construction (given
X's stand-alone usage) to be VCVC.
Of course, the above is only true if several other conditions are met
also, and there are obtusifications that could have been used that would
invalidate that deduction (eg X could be a consonant used as an
abbreviation for a common word like using isolated "t" for "the" in
english). However if we are looking for something that has very simple
rules (so that it can be remembered with ease and the VMS can be read
off in a single pass - a notion for which I've already posted my reasons
for believing) then the deduction has a high likelihood of being valid.
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying: