ther the roots of language originated onomatopoetically or
interjectionally, but simply whether they originated through imitation
or not. For interjections, all of them, could in no way come to be
joined together so as to be means of mutual understanding, i. e.,
words, unless one person imitated those of another. Now if the alalic
child be tested as to whether he forms new words in any other way than
by imitation and transformation of what he imitates, i. e., whether he
forms them solely of his own ability, be it by the combination of
impulsive sounds of his own or of sounds accidentally arising in loud
expiration, we find no sure case of it. Sound combinations,
syllables--and those not in the least imitated--there are in abundance,
but that even a single one is, without the intervention of the persons
about the child, constantly associated with one and the same idea
(before other ideas have received their verbal designation--likewise by
means of the members of the family--and have been made intelligible to
the child), can not be shown to be probable. My observations concerning
the word _atta_ (p. 122 _et al._) would tend in that direction, were it
not that the _atta_, uttered in the beginning without meaning, had first
got the meaning of "away," through the fact that _atta_ was once said by
somebody at going away.

So long as proof is wanting, we can not believe that each individual
child discovers anew the fundamental fact of the expression of ideas by
movements of the tongue; but we have to admit that he has inherited the
faculty for such expression, and simply manifests it when he finds
occasion for imitations.

The first person that has attempted to fix the _number_ of all the words
used by the child, independently, before the beginning of the third year
of life (and these only), is an astronomer, E. S. Holden, director of
the Observatory of the University at Madison, Wisconsin. His results in
the case of three children have been recently published (in the
"Transactions of the American Philological Association," 1887, pp.
58-68).

Holden found, by help of Webster's "Unabridged Dictionary," his own
vocabulary to consist of 33,456 words, with a probable error of one per
cent. Allowing a probable error of two per cent, his vocabulary would be
comprised between the limits of 34,125 words and 32,787 words. A
vocabulary of 25,000 words and over is, according to the researches of
himself and his friends, by no means an