[...] it will be observed that the symbolic discriminations run encouragingly parallel to the objective ones based on phonetic considerations. This may mean that the chances of the responses being to a high degree determined by actual word associations of the language of the subject are slim, the meanings of words not being distributed [...] according to any principle of sound value as such; and [...], that we are really dealing with a measurably independent psychological factor that for a want of a better term may be called 'phonetic symbolism' [...] The reason for this unconscious symbolism, the factor of linguistic interference being set aside for the present, may be acoustic or kinesthetic or a combination of both.
- Sapir (1929), Pag. 68