A fourth difficulty is that by the doctrine of direct imitation it should not be very much more than two or three times as hard to repeat a two- or three-syllable series as to repeat a single syllable. It is, in fact, enormously harder. This is, of course, just what is to be expected if learning a sound means the selection from random babbling plus previous habits. If, for instance, a child makes thirty monosyllabic sounds like pa, ga, ta, ma, pi, gi, li, mi, etc., there is, by chance, one chance in thirty that in response to a word or phrase he will make that one-syllable sound of his repertory which is most like it, but there is only one chance in nine hundred that he will make that _two-syllable_ combination of his repertory which is most like it. On the other hand, two objections will be made to the opposite view that the word spoken acts only as a model to select from responses otherwise caused, or as a stimulus to habits already existing. First it will be said that clear, indubitable repetitions of words never practiced by the child, either as totals or in their syllables separately, _do_ occur,—that children do respond by repeating a word in cases where full knowledge of all their previous habits would give no reason to expect them to make such a connection. To this the only retort is that such observations should be based on a very delicate and very elaborate record of a child’s linguistic history, and that until they are so made, it is wise to withhold acceptance. The second objection is that the rapid acquisition of a vocabulary such as occurs in the second and third year is too great a task to be accomplished by the laws of exercise and effect alone. This objection is based on an overestimation of the variety of sounds which children of the ages in question make. For example, a child who says 250 words, including say 400 syllables, comprising say 300 syllables which, when properly pronounced, are distinguishable, may actually use less than 50 distinguishable syllables. _Ba_, may stand for the first syllable of father, water, barn, park and the like. _Ki_ may stand for cry, climb, and even carry. For a child to say a word commonly means that he makes a sound which his intimate companions can recognize as his version of that word. A child who can produce something like each one of a thousand words upon hearing them, may do so from actual control over less than a hundred syllables. If we suppose him to have acquired the habits, first, of saying _something_ in such a case, second, of responding to a certain hundred sounds when perceived or remembered by making, in each case, a similar sound, and, third, of responding to any other sound when perceived or remembered, by making that sound of his own repertory which is most like it,[42] we can account for a thousand ‘imitations,’ and still not have made a large demand upon childish powers of learning.