TECH.LING: Lojban glottochronology

[Logical Language Group <lojbab@ACCESS.DIGEX.NET>]

I have run Jacques Guy's package GLOTTO on the Lojban etymology files which
were used to construct the Lojban gismu.  These files contain the current
gismu and Lojbanized words from Chinese, English, Hindi, Spanish, Russian,
and Arabic which were used to form the gismu.

GLOTTO requires that the words be classified by whether the user thinks they
are cognate or not.  I used the assumption that a word is cognate to Lojban
if its score was 2 or more, and not cognate to Lojban (or any of the other
languages) if its score was 0.  (Scores of 1 are not possible.)

The following table indicates the percentages of cognate words between
each of the pairs of languages:

        % (cognates per hundred)
        Lojban
        2 Chinese
          99
        3 English
          93  92
        4 Hindi
          85  84  79
        5 Spanish
          74  73  70  63
        6 Russian
          65  64  61  55  49
        7 Arabic
          47  46  44  41  35  32

>From this data, another GLOTTO program was able to produce a family tree.
Note, however, that the family tree is NOT a rooted one: it shows relative
relationships only, and the root of the tree is not to be identified with a
hypothetical proto-language of the seven.  The relationships are mostly
quite correct.


Tree reconstructed from cognate percentages in LOJBAN.PC

 Lojban -1002----:-996-----:-996-----:
Chinese -988-----'         |         |
  Hindi -----852-----------'         |
English -----937-----------:-994-----|
Spanish -----747-----------'         |
Russian -----665-----------:-984-----'
 Arabic -----481-----------'

(Figures on branches are words retained per 1000)

--
John Cowan              sharing account <lojbab@access.digex.net> for now
                e'osai ko sarji la lojban.