Re: Request For Status: Language vs. Cipher

[Mark Perakh <perakh@xxxxxxxxxxx>]

Thanks to Rene and Frogguy for the response. In regard to Jacques comment, the texts
subjected to LSC test (the total of 69 meaningful texts) included first the Book of
Genesis in 12 languages, then Moby Dick, War and Peace (in English and Hebrew),
collections of short stories in English and Russian, the UN convention on sea trade, a
full  text of a Russian newspaper, etc.  Meaningless texts included an artificially
created gibberish, artificially created "almost-zero entropy"  texts, as well as texts
obtained by permutations of either letters, or verses, or words of meaningful texts.
The LSC measurement produced, as the direct outcome, curves of the so-called LSC sum,
which displayed characteristic minima and other features, quite clearly differing for
meaningful texts compared to all other types of texts.  Additional information was
extracted by manipulating the LSC sum curves.  The curves for VMS (obtained separately
for A and B and also for the full version) looked exactly like those for all
meaningful texts, and quite differently compared to meaningless conglomerates of
letters. Anybody who feels uncomfortable with math, can omit the first paper on LSC
(or read it diagonally, without delving deeply into formulas derivation) and look at
the experimental material (again not necessaily chewing every detail) in the second,
third, and fourth articles.  After that the two articles on VMS should become rather
comprehensible (I hope).


Jacques Guy wrote:

>
>
> > Stolfi's and Mark's responses make it necessary that I review my
> > earlier statements. True, while we have always been used to seeing
> > the statistics of the Voynich MS text being rather different from
> > normal languages
>
> Count me out on that. A browse through the archives will reveal
> all my rantings.
>
> > About Mark's LSC papers
>
> Oh... I downloaded them, even complaining to Mark that he
> should have split them into frogguy-size bites, started
> reading, realized  that it would be tough reading (I am
> no mathematician). As it was late, rather, early in the
> morning, I set it aside... and populating that  Web
> site on Easter Island side-tracked me. I had completely
> forgotten that Mark's papers were there, in o:\perakh
> right next to the Easter Island stuff (o:\rr) <-- no,
> that was not an emoticon!
>
>
> > Anyway, to cut an already too long story short, I meant to read on
> > but never got round to it.
>
> Join the club.
>
> > I still think there may be important
> > clues in there, so I recommend all to look at it.
>
> We don't know anything about language. And "we" includes
> me as a linguist. I received a book to review "The Origins
> of Complex Language," OUP. It's all wrong. Even the *data*
> are wrong! We don't know how language works. Imagine
> Newton blind and without the accumulated observations of
> earlier astronomers (Tycho Brahe). You have a mathematician
> tackling language.
>
>
> > What is important is to understand precisely how these non-meaningful
> > texts were generated, and how and why the method identifies them
> > as such.
>
> Remember that many texts are really meaningless. Nursery rhymes
> for instance. And, behind its turgid, convoluted argumentation
> "The Origin of Complex Languages" is meaningless. The author
> is hypnotized with his own words and does not realize that they
> make no sense. Consider now meaningful texts which look meaningless:
> a bill of lading, even... yes, I think even cookery recipes
> would look pretty meaningless, and very repetitive if we
> did not know what they were. I have here a cookery book,
> rather, a menu book dating from the last... no! second last
> century (1800's). For every day of the year, a menu. In
> appendix, the recipes for some of the dishes. Think! Let
> it be written in Voynichese (if there is such a thing).
> I'll OCR a few pages and post them here. (It's in English,
> translated from French)
>
> I expect that menus and bills of lading would share many
> statistical properties. Novels, diaries, very different
> from menus and bills of lading. Cookery recipes (or
> alchemical recipes) different again. I posted here quite
> some time ago short excerpts of classical Aztec in translation.
> It is very different from any European literature.
>
> [Mark:]
> > > The application of LSC to VMS resulted in the data
> > > exactly like those obtained for meaningful texts in 12 natural languages.
>
> Hmmm.... the test ought to be applied to "non-meaningful"
> texts. We can't use a monkey-like text generator, because
> humans are very bad at generating random stuff. What is
> a meaningless text? Dennis I think it was directed us to
> a site where there was a corpus of texts from schizophrenics.
> Perhaps that would do. But the litanies of the Holy Virgin
> too, would have "schizophrenic" statistical properties, I
> expect (Ave Maria, stella maris; Ave Maria, gratia plena;
> Ave...)
>
> > > The LSC tests also revealed a considerable difference between VMS-A and VMS-B,
> > > leading to a conclusion (arguable, of course) that both were written in the
> > > same language but A using much more abbreviations than B.
>
> > I remember being somewhat uneasy about that last conclusion, but I will
> > hold back until I really understand the method.
>
> Same here. But I haven't really attempted.
>
> > The silence probably means lack of understanding rather than
> > disagreement.
>
> I haven't even started to try to understand. I really have to
> get off this Easter Island stuff for a while. It is fascinating
> stuff, though, and I can be sure that it is meaningful...
>
> ... well... NO!  I have become persuaded that some of the
> tablets (the London tablet, the Stephen-Chauvet fragment)
> are fakes. So, a part of the corpus is certainly meaningful
> (the lunar calendar), but another is probably meaningless.
>
>
> > Cheers, Rene
>
> You're the optimistic one! I am far from cheerful about all  that :-(
>
> Frogguy