Volume 31, Issue 1

In a previous paper, the author had discussed the problematical phonetic existence of the «loci» and had concluded that only the F2-F3 loci of the dental and alveolar consonants might be applied without modification to synthetic speech, owing to their phonetic existence. He now discusses the validity of the theoretical bases of the «loci» as isolated not from real speech, but by means of auditory tests. A possible (although not universal) application of the «loci» of labial and velar consonants to synthetic speech is due to some coincidence between such artificial formantic patterns and some real formantic patterns heard the same way. But the precision of such «loci» does not by any means warrant that they contribute maximally to the perception of a given «place of articulation». The only loci on which an investigator may rely are the F2-F3 loci of dental and alveolar consonants because they are the only ones warranted by a very real articulatory correlation. The other «loci» may be profitably replaced by very simple rules of second formant-transition proportionality. The author minimizes the validity of «statistical verifications of the loci» with natural speech, in which formant frequency can by no means be regarded as a valuable criterion of vocality or of transition if it is not verified by synthesis and auditory tests.

Two studies compare men’s and women’s production of Arabic emphatics (pharyngealized consonants). The results of the first study with native speakers of Cairene Arabic indicate that women show significantly less acoustic differentiation between emphatic and nonemphatic segments. The results of the second study show that while Arab men and women differ significantly in their production of emphatics, American men and women who have been taught Arabic by male speakers are much more similar to each other in the pronunciation of emphatics. Among Arabs, the magnitude difference in formant frequencies between men and women was found to be much greater than anatomically predicted by FANT and the direction of the sex-related formant differences was opposite to FANT’s predictions. The results of these two experiments point toward nonphysiological bases of variation in vowel formants for men and women.

This article, the fifth in a series of reports on the results of a computer-assisted analysis of a surreptitiously recorded corpus representing 50 half-hour conversations with members of the Paris ‘establishment’, deals with the glottal stop. Two types are described: The first, a towel - releasing articulation, occurs in utterance-initial and -internal positions. It serves in both positions as a stress marker. The second is an arresting articulation, occurring internally and in utterance-final position. It serves to call attention to a preceding or following element, to abort an unwanted utterance, or to terminate an utterance as quickly as possible in order to get on with the next. It also terminates words repeated in hesitation. The overall frequency of occurrence of the glottal stop in French varies as a function of sex, age, occupation, speaker’s intent, voice level, type of articulation and utterance length. Its pertinent acoustic features are deduced from measurements of spectrograms and synthetic patterns.