Volume 52, Issue 3

A relatively wide range of cross-language similarities in prosodic patterns, i.e. not only in the final fall of statements and the like, but also the treatment of compound words, the marking of phonetic integrity of underlying semantic blocks and their hierarchy may be explained if the typical prosodic configurations are hypothesised to arise from a common archetypal, biologically related (rise-fall) contour, and from a few derived contrastive (rise-nonfall) patterns, whose characteristics may be motivated on psychological and/or ethological grounds. The rise-fall contour seems to be recursively implemented on different sizes of constituents; a rise appears to be mainly associated with the notion of beginning and uncompleteness, a fall with the notion of end, and the valley (fall-rise) with the notion of disjuncture.

Four experimental approaches frequently used in speech perception research are discussed with respect to their impact on word recognition models and their implicit assumptions on the mental representation of speech. These approaches are (1) reaction time experiments; (2) the procedure of click localisation; (3) the method of selective adaptation, and (4) the assessment of word similarities. The results of the studies vary as a function of the experimental procedure chosen. Phonetic features, single sounds, syllables and words as entities are alternatively favoured as primary perceptual units. A critical evaluation and an attempt at integrating the data lead to the assumption that the adult speaker/listener has different kinds of mental representation of speech at his/her disposal. Depending on the focus of perception, units of different sizes are primarily focused in the recognition process. This implies that the listener is able to modify his/her temporal analysis window to a certain extent. Nonetheless, as a default case, the syllable serves as the primary perceptual unit.

The aim of this paper is to answer the two questions ‘What is phonetics?’ and ‘Is it a discipline in its own right?’ Because there seem to exist different answers, these questions are of some interest. Sometimes, phonetics is seen as less than a fully fledged discipline, i.e. merely a part of other disciplines. But on the other hand, it seems equally true that phonetics can be said to be more than just one single discipline, in which case we would get a whole set of ‘phonetic sciences’. Neither of those two paradoxical, or even contradictory, alternatives presents a satisfying solution. In the following we would like to show that phonetics can be called a well-defined discipline in exactly two different senses. Phonetics can be seen as one of two very specific versions of scientifically approaching what will be called the phonetic facts of spoken language . Such facts are a necessary condition for every natural speech act, because there is no speech act without observable phonetic facts. Together, these two versions of scientifically dealing with phonetic facts, a smaller and a larger one, form phonetics as a whole discipline in its own right. This contribution to the Festschrift is written in German, mainly because its title cannot be easily translated into English; for further explanations, please refer to note on p. 158.

In the history of various European languages (Latin, Spanish, English, Swedish, and various dialects of French) there are instances of the cluster mn appearing as mpn . There are philological controversies as to whether this ‘epenthetic’ p was actually pronounced or was just a learned hypercorrect spelling. I offer here a novel phonetic scenario supporting the claim that the p was pronounced and arose in a phonetically natural way: I posit that in the mn cluster there was temporal overlap of the m and n closures. The simultaneous labial and apical closure would create a pocket of air between them which, when the labial closure was released, would undergo a momentary rarefaction of pressure and thus be released with a click-like burst. Listeners would be likely to interpret this stop burst auditorily as a pulmonic [p] and this would be the basis of their own pronunciations.

This paper reports an experimental phonetic study of the acute and grave word accents in Central Standard Swedish. The hypothesis was that the fall in fundamental frequency (F₀) observed on the primary stressed syllable in grave words is a direct consequence of the word accent, while the corresponding fall in F₀ in acute words reflects sentence level intonation. Three male subjects read speech material designed to illustrate how these falls in F₀ are influenced by intonational context. The main finding was that the grave fall was substantial and consistent, whereas an acute fall appeared only in certain contexts. The results indicated that an F₀ fall is a positive acousticperceptual requirement for the grave accent, while the acute accent constitutes the unmarked member of the contrast, i.e., ‘acute’ equals ‘non-grave’.

This study investigates the influence of fundamental frequency (F₀) contour on perceived vowel duration in Norwegian. In the first of four experiments, a falling vs. flat contour in short vs. long vowels in isolated monosyllables was shown to cause a perceptual lengthening of the vowel. Contrary to the expectations, embedding the monosyllables sentence-medially in experiment 2 did not turn the lengthening into a shortening effect. Also unexpectedly, a rising vs. flat contour did not influence perceived duration in monosyllables (experiment 3). Experiment 4 focused on falling as well as rising contours in disyllabic test words. Here, too, a lenghtening effect was found only for a falling pattern. In parallel with experiment 2, the presence of the test word’s second syllable did not turn perceptual lengthening into perceptual shortening. Comparison of the present results with data previously reported for German leads to the conclusion that the interaction of Fo contour and perceived vowel duration is language-specific and cannot be explained by assuming a universal psychophysical mechanism.

Previous research has suggested that the direction of short-duration fundamental frequency (F₀) perturbations following consonants helps to signal consonant [+voice]/[––voice] (abbreviated as [voice]) status. It has been proposed that the [voice] cue corresponds to the direction and extent of F₀ perturbations relative to the overall intonation contour. A competing view, the low-frequency hypothesis, suggests that F₀ participates in a more general way whereby low-frequency energy near the consonant contributes to [+voice] judgments. Listeners identified multiple stimulus series, each varying in voice onset time and ranging from /aga/ to /aka/. The series differed in overall intonation contour as well as in the direction of F₀ perturbation relative to that contour. Consistent with one version of the low-frequency hypothesis, the F₀ value at voicing onset, rather than the relative direction of the F₀ perturbation, was the best predictor of [voice] judgments.

The three-channel pitch determination algorithm (PDA) presented in this paper combines a short-term analysis PDA, which derives fundamental frequency via a periodicity criterion, and two time domain PDAs that determine the instants of glottal closure according to local signal criteria. The first of these algorithms correlates the speech signal with an estimate of the impulse response of the vocal tract; the second applies a neural network. The reliability of these time domain PDAs is increased by constraints on the range of F₀ imposed by the short-term analysis PDA. First results show that the algorithm can be applied to both accurate pitch period determination of running speech and voice quality measurements, particularly the measurements of voice jitter.

According to the distinctive region model (DRM), sequencing one vowel to another involves one, two, or three commands or phonetic movements corresponding to specific deformations of the vocal tract. Criteria pertaining to communication theory require that these movements, or gestures, be of a small number, efficient (small deformations imply large formant variations), and simple (e.g. rectilinear movements). For a given language, the vowel-vowel transitions involving one movement should then be statistically dominant. This hypothesis is verified using a large corpus of French.

The paper presents the outcome of some very recent sound changes, which add to the already considerable inventory of surface vowel contrasts in Danish. It claims that a classical structuralist phonological account of the vowel inventory is not satisfactory if psychological reality of phonological entities is a criterion. The only reasonable abstract representation is a morphophonological one.

The variability of the Finnish vowel occurrences on the F₁/F₂ chart was studied. Repetitions (n = 8 x 10 + 10 = 160) of two syllable types, /hVh/ and /tVt/, produced by a male speaker were analysed. The F₁/F₂ variability in both contexts was estimated. In the /hVh/ context, the phoneme targets were considerably more peripheral than in the /tVt/ contexts. In the /tVt/ context, the phoneme placements were nearer the upper left corner of the F₁/F₂ space in /u/, /o/, /a/, and /æ/. That is understandable on the basis of the alveolar tongue blade position: the tongue holds its position nearer the alveolar area during its movement from [t] to [V] and back to [t] compared to the [hVh] articulation. In the latter articulation, the tongue is free to move and the vowel reaches its ideal position. This interpretation was corroborated by means of the test items with an open syllable (/tVV/). In order to explain the variability of the repetitions in one context, the notion of auditory critical band window (CBW) was used. A CBW covers an area on the psycho-acoustical F₁/F₂ plot that comprises 1 bark on the F₁ and F₂ scales. Its modification CBW-F₁ implies that a critical band is presented as a circle comprising the diameter of 1 bark according to the F₁ scale. Practically all single F₁/F₂ points occur within the CBW-F₁ circle. In most cases, the variability is smaller than one CBW-F₁ circle, but we can consider that a greater amount of repetions would yield a greater dispersion. The auditorily motivated CBW-F₁ circle might appear to be a good prediction for the articulation precision.

German has two functionally distinctive vowels that only occur in unstressed position, one being schwa, the other stemming from schwa + vocalized /r/. Although traditionally acknowledged, with an established IPA symbolic representation [e] vs. [a], little attention has been paid to the phonetic basis of their functional distinction. This paper addresses their contextual variability, and discusses some of the implications for production models found in the literature.

A speaker recognition experiment is described where subjects are asked to identify 1 out of 12 unknown speakers from short samples of speech. All were male speakers without strong regional accents and unknown to the subject group. The task was to match the voice of a test utterance to one of the speakers. The test was run individually as an interactive computer-controlled task. Results on speed and precision of this task are presented. The result demonstrates that it is possible for subjects to quite accurately match unknown voices, speaking different utterances, if the test utterances are reasonably long. However, listeners differ widely in this ability. The study will be used as a baseline for synthesis of speaker characteristics.

In this paper we consider phonetic and phonological aspects of the English voiceless affricate /t∫/ as it is realised by children with developmental speech disorders. The speakers described in the study have normal /t/ but disordered /∫/ and /t∫/. Using electropalatography (Reading EPG), we compare the stop and fricative phases of /t∫/ to independent /t/ and /∫/. This comparison shows that the place of articulation of /t∫/ can be predicted from that of independent /t/. There is a strict requirement for the affricate’s stop release to be homorganic with its fricative phase, irrespective of the place of articulation of independent /t/. Sometimes, there is also an observable coronal gesture during the stop phase of a dorsal affricate indicating the influence of independent /t/. This is predicted by phonological theories in which the affricate is related to both /t/ and /∫/ but not by theories in which the affricate is merely the stop counterpart to /∫/.

Averaged spectra of the Polish voiceless fricatives are described in terms of three central moments, and the variables are analysed using both univariate and multivariate analysis of variance. The vowel context effect is strong, but that of the (male) speaker is not. /∫/ is not distinct from /ҩ/ in this experiment, but this failure should not be a handicap in automatic speech recognition.

Phonemic phonological accounts of the fricatives [ç], [x] and [Ӽ] in German must either introduce an arbitrary piece of morphological information into the phonology, or set up a controversial phoneme of open juncture. An alternative account is proposed which avoids both of these problems. I provide a phonological analysis which employs an explicit syllable structure and a phonetic interpretation of this structure is given in terms of the exponency statements of a Firthian prosodic phonology.