Band 36, Heft 4-5

Descriptive phonetics and instrumental phonetics have two entirely different empirical domains as their subject matters. On this basis, Zwirner’s central problem of the theoretical relationship between sound category and sound realisation is re-formulated and discussed. It is shown how the traditional descriptive procedures of segmentation and decomposition can be connected with the segments of digital speech signals and the features analysed from them.

The relation between input and output signals in a recognition system is described extensionally, and with these means the fundamental problem of automatic speech recognition is formulated. On this basis, the second part of the paper deals with functional problems of signal classification.

Mandible movements in the vowel-to-consonant transition manifest consistent differences between /t/ and /d/ in syllable final position. Both mandible position at its peak value near the articulatory release of the stop and velocity of its ascending movement in the transition are found to be apparently relevant to the tense/lax distinction, even though the reliability of these criteria for categorizing t/d data somewhat varies depending on the speaker. Some related characteristics of lingual movements and similar mandible characteristics for the k/g distinction are mentioned. The data were obtained by the computer-controlled X-ray microbeam system in cooperation with the University of Tokyo group.

This investigation attempts to determine, via classical reaction-time techniques, whether speakers require longer times to initiate the production of so-called ‘bite-block’ vowels than they do for normal vowels. The results suggest that speakers do spend measurable amounts of time in neurophysiological reorganization for such vowels. These results are discussed in terms of some current models for speech motor control.

The relations between area functions of the vocal tract and the acoustic result are of phonetic interest only if they can be interpreted in articulatory terms. We show which problems may arise if two-dimensional configurations of the vocal tract are to be transformed into area functions, and which relations exist between articulatory variables, area functions, and formant frequencies in an articulatory model.

Methods for estimating articulatory data from acoustic measurements are reviewed. First, relations between the vocal-tract area function and formant or impedance data are pointed out. Then the possibility of determining a (discretized) area function from the speech signal itself is considered. Finally, we look at the estimation of certain articulatory parameters rather than the area function. By using a regression method, such parameters can even be estimated independently of any vocal-tract model. Results for real-speech data are given.

The accuracy with which naive listeners can report sentences presented 12 dB below a background of continuous prose was compared with accuracy in four audio visually supplemented conditions. With monochrome displays of the talker showing (i) the face, (ii) the lips and (iii) four points at the centres of the lips and the corners of the mouth, accuracy improved by 43, 31 and 8%, respectively. No improvement was produced by optical information on syllabic timing. The results suggest that optical concomitants of articulation specify linguistic information to normal listeners. This conclusion was reinforced in a second experiment in which identification functions were obtained for continua of synthetic syllables ranging between [aba], [ada] and [aga], presented both in isolation and in combination with video recordings. Audio-visually, [b] was only perceived when lip closure was specified optically and, if lip closure was specified optically, [b] was generally perceived. Perceivers appear to make use of articulatory constraints upon the combined audio-visual specification of phonetic events, suggesting that optical and acoustical displays are co-perceived in a common metric closely related to that of articulatory dynamics.

The analysis of the production of fortis and lenis plosives in German has shown the importance of the duration ratio vowel/ (vowel + closure) for the distinction. To complement these results a perception test was carried out in which 29 native speakers identified a randomised sequence of 220 stimuli from tape as one of the phrases ‘Diese Gruppe kann ich nicht leid(e)n (leit(e)n)’. The stimuli were obtained from the two naturally produced originals by changing the ratios and the length of voicing in the plosive through computer processing. The test results indicate very conclusively that judgment can be reversed simply by changing the ratio to the appropriate ones found in production. There is a hierarchy of perceptual dimensions: duration ratio > formant transition > voicing.

The great majority of Welsh word-final plosives are devoiced lenes (fully voiced before a following vowel or sonorant), with a long vowel preceding. There are also a certain number of words – nearly all borrowed – ending in a fortis plosive, voiceless in all circumstances, with a short vowel preceding. There are quite a few minimal pairs where a long vowel + lenis plosive is opposed to a short vowel + fortis plosive. Arguments for the primacy of voicing and of vowel length, respectively, are presented. A solution is put forward which specifies plosives with regard to voicing in the lexicon and which provides rules (a) doubling any voiceless plosive after a stressed vowel and (b) lengthening any stressed vowel in the context before up to one consonant. A number of exceptions are discussed. As a linguist, the author prefers a solution which requires the specifying of perhaps 60 or 80 exceptions as exceptional over one which would call for at least one extra feature specification for every word in the lexicon.