"Thomason wants us indeed to believe that the CM has been so successful
that it "has been altered only in relatively minor ways since" the
1870s and "is envied by many other historical scientists" (2002, p.
102). To
my knowledge, there are quite a few historical linguists, apparently
"good"
(Thomason's term) and respected ones, who have questioned the "famous"
CM. Almost all the contributions to Aikhenwald & Dixon (2001) question its
reliability and "the regularity hypothesis of sound change" (Thomason,
2002,
p. 102). They argue that synchronic structural similarities among
languages,
on which the CM relies, may be due to common inheritance, mere typological
coincidence, or diffusion. Likewise, Campbell (2002, pp. 146147) includes
an informative caveat about the CM. Overall, one must independently know
the history of population movements and contacts among speakers of the
relevant languages in order to sort things out. The position is echoed by
Laks (2002). As Thomason (2001) herself acknowledges, contact has played
an important part in the histories of all languages. Some of the
chapters in
Aikhenwald & Dixon (including those by Calvert Watkins, Randy Lapolla,
James Matisoff, and Bernd Heine & Tania Kuteva) clearly show that the
Stammbaums suggested by the traditional application of the CM do not do
justice to the complex ways in which languages are genetically
related. (This
is precisely the position developed by Laks, 2002.)"
http://humanities.uchicago.edu/faculty/mufwene/RESPONSE%20TO%20THOMASON.pdf
Thefore, "one must independently know
the history of population movements and contacts among speakers of the
relevant languages" before comparative method can be applied. The
comparative method cannot be used as evidence for population movements.
M. Kelkar