1936/01/01 by B. F. Skinner · 3 citations
Computer Science · Psychology · #Language, Metaphor, and Cognition #Natural Language Processing Techniques #Speech and dialogue systems
paper · doi:10.1080/00223980.1936.9917445
openalex publication_date 1936/01/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/03
T h e verbal summator is a device for repeating arbitrary samples of speech obtained by permuting and combining certain elemental speech-sounds.One of its uses is comparable with that of ink-blot and free-association tests.T h e speech-sample does not fully represent any conventional pattern in the behavior of the subject, but it functions as a sort of verbal ink-blot.When it has been repeated a sufficient number of times (the number corresponding to the time of exposure of the blot), the subject makes a conventional verbal response resembling it.For example, a few repetitions of the skeletal sample ah-uh-uh-00-uh may evoke the response stars ooerlooking.This is by no means the only response matched by the sample, and since it is not evoked by any stimulus acting at the moment, its emergence may be said to be due to its own relative strength.By presenting a wide variety of skeletal samples, an investigator is able to obtain a list of the especially strong responses of his subject.This is essentially what is done in ink-blot and free-association tests.Aside from its use as a test, the summator is valuable in the study of other aspects of verbal b e h a ~i o r .~ W h e n we say that a man uses a certain number of expressions-when we refer to his "vocabulary" -we say nothing about the relative importance of each expression in his behavior.A "vocabulary" does not exist in a uniform state 'Received in the Editorial Office on April 20, 1936, and published imme-'Society of Fellows, Harvard University.'It seems advisable to give the greater part of this account in the loosely defined but familiar terms ordinarily used in discussing language.A much more rigorous formulation is available in reflex terminology, but it would be practically inaccessible in that form to many who may find the material useful.I should like to say, however, that the apparatus and procedure were designed to test the principle of the summator after it had been arrived at theoretically in setting up a system to deal with language purely a s behavior, and that a more rigorous presentation of the data will be given in an exposition of that system now in preparation.'Not to be identified with "word."'Cj.Paget ( 1 ) .'1 have to thank Prof. F. C. Packard of the Harvard Voice-recording Laboratory for very generous help in the preparation of these records.'When the optimal intensity is being explored, the value at which the imitative response emerges is still low enough to avoid exposing the nature of the material, provided too many examples are not given.'Examples of pairs of responses reported as coming out at the same time: silver plaything-atmost anything, forlorn of hope-four on the boat.