<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=Content-Type content="text/html; charset=iso-8859-1">
<META content="MSHTML 6.00.2900.2853" name=GENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=#ffffff>
<DIV><FONT face=Arial size=2><FONT face="Times New Roman" size=3><FONT
face=Arial size=2>Dear Corpora colleagues, we have computed the phonetic corpora
of some AmerIndian languages. The result of comparing the speech sound
pictures:</FONT> </FONT></FONT></DIV>
<DIV><FONT face=Arial size=2><FONT face="Times New Roman" size=3><FONT
face=Arial size=2>T</FONT><FONT face=Arial size=2>he dissimilarity is based on
the value of the coefficient of variance. The more the variance the less the
similarity. <BR>The dissimilarity between the languages (Piratapuyo,
Siriano, Secoya) of the Tucanoan family is 37.14%, which is greater than that
between the Quechuan (Quechua, Inga) languages – 26.09%. The Chibchan family
(Cofan, Guambiano) is much less similar – 43.90%. The most similar turn to
be Tupi-Guarani languages (Guarani, Kaiwa) – 20.07%.The dissimilarity between
two texts of the Sweet Grass Cree language is 1.30%. The two texts of Guambiano
(Part 3 and Part 4) also show a greater degree of similarity (3.02%). Two
Cofan texts (Part 1 and Part 2) are also similar enough (3.68%). It tells us
that this phonostatistical tool works correctly. If you can advise in what
linguistic journal I may publishe my results, please write to <A
href="mailto:yutamb@mail.ru">yutamb@mail.ru</A> Remain sincerely yours
Yuri Tambovtsev</FONT></FONT></FONT></DIV></BODY></HTML>