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

Re: The letters <p> and <f>, again



[Oops, I had sent this reply to Rene only instead of the whole list. Sorry...]

    > [Karl Kluge:] D'Imperio points to f57r where there is a repeated
    > sequence of characters, except in one sequence V occurs for B.
    > On that basis, she suggests that perhaps B = V (and therefore,
    > possibly F = P) (i.e., p = f and k = t rather than p = t and f =
    > k).

Yes, you will find a formatted copy of that sequence in 
http://www.dcc.unicamp.br/~stolfi/voynich/Notes/034/Note-034.html
Here it is :

      | 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17
    - + -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
    1 | o  l  j  r  v  x  k  m  f  &L t  r  &H &G y  &I &Y
    2 | o  l  d  r  v  x  k  m  f  &L t  r  &H &G y  c  &Y
    3 | o  l  d  r  v  x  k  m  p  &L t  r  &H &G y  c  &Y
    4 | o  l  d  r  v  x  k  m  p  &L t  r  &H &G y  c  &Y 
    - + -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --

See that page for the meaning of &L, &H, &G, &I, &Y.
Here it is converted to Currier (except for the weirdos):

      | 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17
    - + -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
    1 | O  E  7  R  v  x  F  J  V  &L P  R  &H &G 9  &I &Y
    2 | O  E  8  R  v  x  F  J  V  &L P  R  &H &G 9  c  &Y
    3 | O  E  8  R  v  x  F  J  B  &L P  R  &H &G 9  c  &Y
    4 | O  E  8  R  v  x  F  J  B  &L P  R  &H &G 9  c  &Y 
    - + -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --


Indeed the occurrence of <f> and <p> in homologous positions (column
09) could mean that they are equivalent. But, in that case, then we
should also conclude that <r> = <s> --- which is not impossible, but
very unlikely. On the other hand, the consistent use of <k> on column
07 and <t> on column 12 should also be taken as proof that these two
letters are distinct.

However, position 09 may be a special one. If the period boundaries
are indeed between positions 17 and 01 -- as suggested by the break at
10:30 and the <dairar> label -- then position 09 is precisely the
center of the period. Or, if the &L weirdo (which looks a lot like the
number "17" in old German digits) marks one end of the period, then
the other end would be either column 11 or column 09. Note also that
we have <f> in two consecutive periods, and <p> in the other two. 
So perhaps <f> and <p> are indeed distinct, and the switch is
a meaningful part of the pattern.

In any case the, letters on that ring (and some of those on other
rings, too) seem to be symbols -- "keys" of some sort -- rather than
letters with phonetic value. So their possible equivalence *as keys*
may not have much relevance for other text. E.g. the English letters
"A" and "a" are phonetically equivalent, but are distinct symbols in
math. Conversely "u" and "v" are phonetically distinct, but 
can be interchanged in Latin words or Roman numerals.

Besides, the handwriting on this page seems to be uglier than average;
the two <f>s in particular look quite different from each other, and
several of the symbols don't occur anywhere else. So perhaps *all* 17
symbols were meant to be weirdos, distinct from ordeinary letters; and
what we see is the result of an ignorant scribe interpreting some of
those weirdos as variants of the ordinary letters which he had
previously learned to copy.

    > [Rene:] Stolfi's table shows that one can usually exchange <k>
    > for <t> (or v.v.) and come up with a valid Voynich word.
    
Note that the table did not list complete words, but only word
fragments.

Indeed, swapping <k> for <t> in a whole word generally produces
another valid word.  Ditto for swapping <p> with <f>.  *But these
swaps change the word frequencies substantially*, and there 
doesn't seem to be a clear pattern.   

So I would rather believe that <k> and <t> are "phonetically" similar,
but semantically distinct (like "t" and "d" in Spanish or Italian,
say), and that the vocabulary is so "dense" that almost any
phonetically valid string is a common word.

    > This is also true for the pair <f> and <p>. At the same time, it
    > shows that you cannot exchange {<k> or <t>} for {<f> or <p>}
    > (although there may be exceptions).

True, but the point is that the most obvious differences betwen <k/t>
and <p/f> seem to disappear if we replace *some* of the <p>/<f> by
<pe>/<fe>, some <cfh>/<cph> by <cphe>/<cfhe>, and omit the ligatures
in some of the <ch> that follow those letters.

    > The appearance of <f> and <p> at top lines of paragraphs only
    > (virtually), should remind us of the gallows in the letter shown
    > in Capelli, where these are purely ornamental additions to
    > existing letters at the top and bottom lines only.
    
Surely <k> and <t> are not ornamental; their distribution is too 
consistent for that.

    > Do we get valid Voynich words if the f's and p's are simply
    > removed?
    
Good question.

Here is a tentative answer. Below are the words with <p>/<f> gallows,
where both variants together have at least 5 occurrences in the book,
for which omitting the gallows produces a word with less than 5
occurrences. (The first line says that the words <qotchdy>, <qokchdy>
<qopchdy> and <qofchdy> occur 19, 44, 15, and 1 times, respectively, 
while <qochdy> does not occur at all.)

         gallows letter
      ---------------------
        t   k    p   f    -  rest of word
      --- ---  --- ---  ---  -------------
       19  44   15   1    0  qo-chdy
       21  38   30   8    2  qo-chedy
       21  16   16   5    1  o-chdy
      123 137    8   3    2  o-al
       61  65    9   1    2  qo-chy
       75 138    9   1    2  chc-hy


And here are those where the <p> and <f> variants occur 
at least 5 times, and the gallows-less variant occurs
more than 10 times:

         gallows letter
      ---------------------
        t   k    p   f    -  rest of word
      --- ---  --- ---  ---  -------------
      128 120   10   3   13  o-ar
      142 202   13   4   23  o-aiin
       10   7    4   2   15  y-chey
        3   0   12   0   31  -chedar
        9  10    4   2   24  qo-
        1   5    4   4   35  ol-chedy
       73 252    4   1   23  qo-aiin
        0   1    7   0   35  -olchedy
        0   0    7   1   44  -olaiin
       12   3    8   0   44  c-haiin
        3   0    6   1   41  -chodaiin
      110  40   11   6  148  c-hy
       15  20   10   4  136  -chdy
        6   4    6   2   83  -cheody
        5   6    3   2   63  -cho
        8   3    6   1   86  -chody
       29  20   33  10  496  -chedy
       15   1    3   2   68  c-har
        6   6   12   1  165  -cheol
        0  13    6   2  117  l-chedy
       19  20   12   2  199  -chor
        3   4    5   0   92  -cheor
       23  28    3   4  148  -chy
        6   6    4   4  167  -cheey
       57  22   16   3  384  c-hol
       47  36   22   3  518  -ol
       20  20   10   2  334  -chey
       47  10    6   0  199  c-hor
       14  20    9   3  384  -chol
       53  29    6   4  334  c-hey
       41  47    5   4  336  -ar
       20  21    7   1  335  -or
       42  62    7   0  431  -aiin
       23  61    5   0  334  che-y
        9   5    3   2  423  -shedy
       14   4    5   0  496  c-hedy
    

It seems that the <p> and <f> gallows are deletable when they
occur at the beginning of the word (with or without 
platform. On the other hand, the instances that cannot be 
deleted are generally receded by <o> or <qo>.

For comparison, let's look at the <t>/<k> gallows too. Below are the
words with <t>/<k> gallows, with combined counts >= 10, for which
omitting the gallows produces an invalid word. (The first line says
that the words <qotey> and <qokey> occur 23 and 106 times,
respectively, while <qopey>, <qofey>, and <qoey> do not occur at all.)

         gallows letter
      ---------------------
        t   k    p   f    -  rest of word
      --- ---  --- ---  ---  -------------
       23 106    0   0    0  qo-ey
       56  62    0   0    0  o-ey
       36  60    0   0    0  o-eol
       43  45    0   0    0  -edy
       24  54    0   0    0  y-eey
       19  44   15   1    0  qo-chdy
       30  24    0   0    0  o-eody
       24  24    0   0    0  y-eedy
        6  39    0   0    0  l-eedy
        1  40    0   0    0  l-eey
       21  20    0   0    0  y-edy
        1  38    0   0    0  ol-eey
        7  32    4   2    0  chc-hey
        9  30    0   0    0  qo-eody
       16  23    0   0    0  -eol
       12  26    1   0    0  qo-am
        5  28    0   0    0  l-edy
        4  26    0   0    0  ol-edy
        9  18    0   0    0  o-eeol
        5  21    0   0    0  -eody
       10  16    1   0    0  o-aly
       11  14    0   0    0  o-eeody
       12  13    0   0    0  o-ody
       12  12    1   0    0  y-al
        5  17    0   0    0  qo-aly
        9  13    1   0    0  y-eody
       13   9    3   1    0  qo-chor
        8  11    0   0    0  o-eal
       12   7    0   0    0  o-edar
        3  15    0   0    0  qo-eed
        9   9    0   0    0  qo-ody
        5  12    0   0    0  y-eol
        8   8    0   0    0  y-eeody
        9   7    0   0    0  o-o
        6   9    2   4    0  -
        6   9    4   0    0  chc-hedy
        8   7    0   0    0  y-ain
        2  12    0   0    0  qo-echy
        4  10    2   0    0  cheoc-hy
        7   7    0   0    0  cho-ey
       10   4    0   0    0  y-am
        4   9    0   0    0  o-shy
        5   8    0   0    0  qo-eo
        1  11    0   0    0  y-eeol
        4   8    0   0    0  -eeody
        4   8    0   0    0  -eor
        6   6    1   0    0  o-oldy
        1  10    0   0    0  ol-ey
        3   8    0   0    0  qo-ed
        3   8    0   0    0  qo-edar
        4   7    1   0    0  o-aldy
        5   6    0   0    0  y-eeey
        6   5    0   0    0  choc-hey
        8   3    0   0    0  o-eed
        3   7    0   0    0  -eo
        4   6    0   0    0  chy-y
        4   6    0   0    0  qo-chd
        4   6    0   0    0  y-eor
        5   5    4   0    0  o-ary

Given that about half of the word occurrences have no gallows, the
fact that words like <qoey> don't occur seems quite significant. There
are a dozen more common <t>/<k> words where deleting the gallows
produces a valid but very rare word.

On the other hand, here are some common <t>/<k> words where the
gallows apparently can be omitted:

                        gallows letter
      -------------------------------  
        t   k    p   f    -  rest of word
      --- ---  --- ---  ---  -------------
       40 300    0   0   14  qo-eey
       82 137    4   0   10  qo-y
       72 301    0   0   18  qo-eedy
      128 120   10   3   13  o-ar
       55 132    1   0   10  qo-ar
      142 202   13   4   23  o-aiin
       73 252    4   1   23  qo-aiin
       34  34    3   1   12  cho-y
        1  45    1   0   13  l-aiin
        1  28    0   0   12  ol-ain
        1  17    0   0   13  l-y
        2  11    0   1   11  ol-ar
       10   7    4   2   15  y-chey
       12   5    1   1   15  y-chor
      110  40   11   6  148  c-hy
        6   6    3   0   12  y-chol
        9  10    4   2   24  qo-
        1  31    0   1   44  ol-aiin
        1  10    0   0   17  ch-ain
       11  33    0   0   73  -ain
        1  11    1   0   25  she-
        2  18    1   1   44  ch-aiin
        6  13    2   0   47  ch-al
        1  19    0   1   55  ol-y
       12   3    8   0   44  c-haiin
       23  28    3   4  148  -chy
        8   6    2   0   42  -ody
       16   5    1   0   63  c-ho
       12  12    2   0   72  -air
       17  22    2   0  135  -y
       47  10    6   0  199  c-hor
       18   4    2   0   86  c-hody
       41  47    5   4  336  -ar
       15  20   10   4  136  -chdy
       12   5    2   1   69  o-
        5   7    2   1   50  che-ar
       23  61    5   0  334  che-y
       53  29    6   4  334  c-hey
       15   1    3   2   68  c-har
       42  62    7   0  431  -aiin
        3  12    1   0   68  ch-ar
        6   4    2   0   47  c-hal
       57  22   16   3  384  c-hol
        7   9    0   0   83  -am
       19  20   12   2  199  -chor
       12  16    1   0  148  ch-y
       21  28    2   0  262  -al
        5   6    3   2   63  -cho
        8   2    3   0   58  -odaiin
       47  36   22   3  518  -ol
        7  34    0   0  268  she-y
       14  11    1   1  167  c-heey
        8  13    0   1  143  shee-y
        3  11    0   1   98  sh-y
        6   4    3   0   72  -cheo
        3  21    0   1  167  chee-y
        6   5    3   1   83  c-heody
        8   3    6   1   86  -chody
        6   4    6   2   83  -cheody
       20  21    7   1  335  -or
       20  20   10   2  334  -chey
        0  13    6   2  117  l-chedy
        4   6    1   1   98  -shy
        5   7    0   0  121  -sho
       10   6    2   2  165  c-heol
        3  13    0   0  167  ch-eey
       29  20   33  10  496  -chedy
       14  20    9   3  384  -chol
        6   5    0   1  136  c-hdy
        6   6   12   1  165  -cheol
        6   6    4   4  167  -cheey
        5   6    0   0  167  che-ey
        1   9    0   0  158  qo-l
        6   4    4   0  180  -shol
        8   6    2   0  268  -shey
       14   4    5   0  496  c-hedy
        9   5    3   2  423  -shedy

It seems that the cases of "essential" <t>/<k> gallows are generally those
words with <o>, <qo>, or <y> prefixes; and words where the gallows 
has a platform, or is followed by "e"s.  The "removable" <t>/<k> gallows 
also seem to include a large fraction of initial cases. (I believe that
John Grove has already noticed that.)

Of course, the fact that a given letter can be removed 
from many words does not mean that it is superflous. (Consider 
final "s" or "y" in English, or 

    > Or are they ornate variations of other letters?

On the whole, considering their distribution whithin paragraphs,
I would say that they are ornate versions of *something* --- probably
combinations of letters.

    > On a more frivolous note, having recently been to Prague I find
    > it irresistable not to learn a bit more about the Czech language.
    > (I can already say: "do not enter or leave the train, the doors
    > are about to close") :-)

Let me see, I bet it it sounds something like 
"U concete prosím u vístup a nástup ..."  8-)

    > To the point. Czech has a number of orthographic rules which remind
    > me a bit of some of the observations made by Stolfi (no 'e' after
    > 'f' or 'p').

Unfortunately many other languages have these rules, too.
(Even... you know which one. 8-)

All the best,

--stolfi