ON THE COMPARISON OF ASIATIC LANGUAGES. 22 
which recall in turn each of the three great Asiatic groups, 
The reason may be that these very ancient languages go 
back beyond the time of the special and separate growth of 
Mongol, Aryan, and Semitic speech. To compare the nouns 
of one language with those of another will generally be 
unconvincing, but when we are able to compare the. roots, 
whence such nouns are formed, and from which the verbs and 
other parts of speech also spring, we are following a method 
safer, and more likely to lead to real conclusions. It is now 
therefore proposed to attempt such a comparison, and to draw 
such general deductions from it as may serve to cast a light 
(however dim) on the earliest conditions of the human race 
in Asia, 
COMPARISON oF Roots, 
The table appended to this paper may perhaps serve to 
call attention to the possibilities of such a method, though it 
cannot claim to be more than a preliminary sketch. It appears 
to me legitimate to suppose that changes in vowel sound, 
such as we find in all dialects, occur also in the roots of the 
three groups, and that the letters which we know to be only 
distinguished with difficulty are not original distinctions, but 
the result of a constant specialisation of sound, due to the 
increasing power of language in distinguishing shades of 
meaning. But it will not be found that any very ingenious 
process is necessary, since the comparisons are much easier 
than would at first be expected. Nor willit be found, I think, 
that I have been misled by foreign words, which have been 
carefully excluded from consideration as affording no evidence 
of the true connection. 
About 170 roots, all connected with the most ordinary 
ideas of action, serve to connect together the various groups 
of Asiatic languages, and of these about 50 are still trace- 
able throughout the entire number, that is to sayin Akkadian, 
i Egyptian, in Aryan, in Semitic, and in Mongolic speech 
alike. It oppears to me that the number alone is suflicient 
to prove that these resemblances are not accidental, and 
especially so, since the more advanced languages —the Aryan 
and Semitic—in a great many cases agree not only in the 
monosyllabic, but also in the derived bisyllabie roots. But 
beyond such a comparison of roots it is difficult, if not 
impossible, to proceed. In grammatical construction, in 
pronouns, and in syntax, the various groups are separated by 
cardinal differences which must not be overlooked. Two 
Q 2 
