1. The science of speech-sounds has been so observant of 'objectivism' as to neglect the objective value of the 'subjective' response on the part of the speaker; it has had too much of: dogmatic fiction deriving from the observers' preconceived 'systems'. A ture 'descriptive' science should have no 'system'. 2. To prove this, I instituted, in 1950, a questionnaire (III) at Baton Rouge, Louisiana, to 48 Americans, most of whom were college students. 3. Question [2] gives us the following results: a. About 2/3 of the answerers recognize a double consonant after a strong short vowel before a weak syllable, unless influenced by word-formation, syllable structure, or, most seriously, spelling. (VII, VIII, and IX) But no consistency prevails. b. The responses to the questions in XI and XII prove that spelling, can hardly influence phonology without due grounds. The so- called 'spelling pronunciation' means, then, the healthy tendency among the speakers to try to conform sound to spelling, not any reckless distortion of their sound-images caused by spelling. c. [t〓], [d〓], [ts] and [tθ] seem to be fairly well-established single phonemes in some. (XII, XIII and XIV) d. In about 1/4 of the speakers, '-ng' is not [〓], but [〓g] or [ng] (XVI) About the same number have [-〓〓r], not [-〓g〓r], for longer and finger. e. Most important: For not a few, '-ng ' can be either or' [〓g, ng] according to circumstances. No consistency marks the English language today. f. For many, [ju:] is a single vowel both medially and initially. (XVII) g. For most of the speakers, Marx is not phonologically equal to marks. (XVIII) [-ks] is often 'flexible'. h. 14 out of the total 48 have [wr]. (XIX) 4. The above-mentioned conclusions were consolidated by the data from the questions other than [2].
