Electropalatographic and acoustic data reported in this study show differences in closure location and degree, dorsopalatal contact size, closure duration, relative timing of events and formant frequency between clear /l/ and dark /l/ in two dialects of Catalan (Valencian and Majorcan). The two Catalan dialects under investigation differ also regarding degree of darkness but essentially not regarding coarticulatory resistance at the word edges, i.e. the alveolar lateral is equally dark word-initially and word-finally in Majorcan, and clearer in the former position vs. than the latter in Valencian, and more resistant to vowel effects in the two positions than intervocalically in both dialects. With reference to data from the literature, it appears that languages and dialects may differ as to whether /l/ is dark or clear in all word positions or whether or not initial /l/ is clearer than final /l/, and that articulatory strengthening occurs not only word- and utterance-initially but word- and utterance-finally as well. These and other considerations confirm the hypothesis that degree of darkness in /l/ proceeds gradually rather than categorically from one language to another.

In a study of European Portuguese fricatives it was noted that /R, / were often realized as [χ, ], which are, respectively, unvoiced and voiced uvular fricative, and voiceless tapped alveolar fricative. Although these phones have not previously been recognized as occurring in European Portuguese, they occurred in 115 words out of a corpus of 1304 words, out of which 107 words could be analysed, in recordings made by four native speakers (two men, two women). Time- and frequency-averaged power spectral characteristics are presented, together with a detailed temporal analysis. The frequencies of high-amplitude peaks in the [χ] and [] spectra clearly indicate a back place of articulation, with median duration across subjects and words of 69 ms and 35 ms. The short (approximately 20 ms), noisy acoustic signal in [] suggests a stop-like manner of articulation, but the turbulence noise characteristics, similar to fricatives, differ from the transient burst noise of plosives.

Formulations of flapping as a symbolic phonological rule suggest clear articulatory differences between flaps and stops, and often offer no overt explanation for why phrase boundaries should block the alternation. The present study explores the articulatory foundation of the distinction between flaps and non-flaps in word-final position. We examine kinematic and acoustic data for these articulations in phrase-final and phrase-medial positions and in falling- and level-stress contours. It is shown that a discrepancy exists between acoustic and articulatory durational patterning – while acoustic durations of flaps are shorter than those of non-flaps overall, their articulatory durations are not uniformly so. It is important to consider multiple potential articulatory sources – both spatial and temporal – for the acoustic shortness that characterizes flaps, including spatial reduction, temporal articulatory shortening, and changes in intergestural coordination. The kinematic data indicate that different sources of word-final flap shortness exist for different speakers and different prosodic conditions, suggesting that gradient variability in the spatiotemporal patterning of tongue-tip constrictions yields acoustic shortening in word-final flaps.

We compare the vowel systems of Finnish, Mongolian (in China) and Udmurt, and illustrate the average placements of their monophthongal vowel types on F1/F2 plots. Mongolian has more vowel phonemes (12 long and 12 short ones) than Finnish (eight long and eight short) and Udmurt (seven). Some basic linguistic characteristics and key word lists of the three languages are presented. For comparison we utilise psychoacoustical F1/F2 formant charts which are fairly good approximations to the vowel space. The phoneme distances are indicated by means of circles of 1 Bark diameter centered on the mean F1/F2 points of the vowel types. This kind of representation allows one to draw conclusions about qualitative vicinity, partial overlapping or even merging of phoneme qualities on F1/F2 plots and about the necessity of further acoustic parameters for vowel differentiation. We also discuss some centralisation phenomena in the three languages.

Phonologists have long discussed the properties of ‘stress’ in Turkish, though there are some authors that suggest that Turkish is actually a pitch-accent language. As a first step in determining whether Turkish should be grouped with pitch-accent languages such as Basque and Japanese or with stress-accent languages such as English, this experimental study provides a detailed phonetic examination of word-level accent in Turkish. Seven female native speakers of Turkish were recorded saying 40 target words in three repetitions. The results show that there are significant differences in duration, peak amplitude, and peak F0 between words with final vs. non-final lexical accent. The magnitude of the duration differences, however, is not likely to be sufficiently perceptible to be a reliable cue to accent placement. The F0 peaks were dramatically different between lexically accented and unaccented syllables. A discriminant analysis confirms that F0 is the most robust cue, followed by intensity and then duration. The current study also provides a method for determining lexical accent placement in words where accent placement has been disputed. That is, a marked drop in F0 signals that the syllable preceding the drop contains the lexical accent.

Arop-Lokep, an Austronesian language of Papua New Guinea, has a contrast between a trilled /r/ and a geminate /rr/ in both intervocalic and word-initial positions. This paper presents the results of acoustic and perceptual experiments involving these segments. Specifically, we probe the statistical characteristics of the /r/ vs. /rr/ contrast across prosodic positions from the perspective of both the speaker and the listener. We measure trills in 24 different Arop-Lokep words, so the resulting data are quite robust and give an excellent idea of the variation inherent in the language. A novel finding is that both lengths of trill consist of more apical contacts word-initially than intervocalically. The scope of this work is unprecedented in that no previous instrumental study has examined both initial and medial geminate trills in the same language.

Seri (self-designation: cmiique iitom [kwĩki:tom]) is a language isolate spoken in the state of Sonora, Mexico, by about 800 people. In this description we present a variety that is typical of the majority of speakers. The recordings are of one of the authors (XMH), a male speaker born in 1964. The technical description is based on Moser & Moser (1965), although the analysis has been updated to agree with the data and arguments presented in Marlett (1981, 1988) and Moser (1996).