Band 36, Heft 1

An attempt was made to determine the prosodic features that distinguish modalities and emotive attitudes through controlled variation of fundamental frequency and overall sound pressure in a sample of four synthesized French sentences. The following tendencies were observed: (a) judgements ‘interrogative’ in disjunctive sentences were essentially more frequent with a steeper rise in fundamental frequency (interval/time) in the first stressed syllable; (b) in the case of certain melodic cliches, assertive and interrogative sentences were distinguished by characteristic intrasyllabic (constant vs. rising) fundamental frequency changes; (c) basic emotions such as anger, joy could not be distinguished on the basis of mere changes in the fundamental frequency; more complex social attitudes, however, can be differenciated in this way, and (d) if two fundamental frequency curves differed only in their angularity (in suddenness of change of direction) then sentences with a more angular fundamental frequency curve were rated as being significantly more aggressive than their smoother counterparts.

Cinefluorographic tracking of pellets on the tongue body, tongue tip, jaw, and lips of three subjects indicates that there are certain basic strategies used for initial /sp/, /st/ and /sk/ clusters. The data reveal an economy of effort in tongue movement, and support the concept of the tongue as a multiple articulator. In addition, the shorter fricative duration of /s/ before /p/ is seen to be a result of early lip closure relative to the occlusions for /st/ and /sk/, which are delayed due to tongue involvement with the /s/.

Durational characteristics of [s] were estimated as a function of two experimental tasks, i.e. reading and delayed imitation. In the two tasks, each of six subjects produced 25 tokens of /s/; the s segments were embedded in sentences, and were controlled for location of lexical stress and syllable and morpheme boundaries. In addition, phonetic context effects on [s] were controlled. Results indicate that the reading task introduces exaggerated duration effects relative to those observed in delayed imitation. For example, the hypothesis that an underlying double s may account for longer [s] in the first member of stimulus pairs such as misstep-mistake was shown to be true under the reading condition but not the delayed imitation condition.

Is there a steady transition between unrounded, ‘outrounded’, and ‘inrounded’ Swedish vowels, or are these distinct categories? By which articulatory parameters can these distinctions be described? These questions, actualised through confusion data in speech reading, are answered by means of articulatory measurements of lip-opening, lip protrusion, and jaw-lowering in 4 subjects. It is found, i.a., that rounded and unrounded vowels form clearly separated categories, and that unrounded vowels, including /o/, are distinguished from outrounded ones by their smaller lip-opening and, if long, by increased lowering of the jaw.

This paper describes a procedure with which complex raw fundamental-frequency (F₀) tracings of Danish can be split into their components, namely (1) coarticulatory F₀ variations; (2) intrinsic F₀ level differences between segments (vowels); (3) F₀ movements pertaining to individual segments (vowels); (4) F₀ patterns associated with stress groups, and (5) the intonation contour of the sentence. ‘Stød’ is not considered here. The problems inherent in such a procedure are discussed as are the hypotheses concerning physiological production models of the speech signal and its perceptual correlates.