The majority of studies on phonation types have focused on low vowels due to the minimal effects of their first formant on harmonic amplitude. In studies of multiple vowel qualities, reports on the relationship between vowel and voice quality are mixed: some show similar formant frequencies across phonation types (e.g. Abramson, Nye & Luangthongkum 2007, Khan 2012), while others show different formant frequencies depending on voice quality (e.g. Ren 1992, Kuang 2011). Results differ as to whether the degree of non-modal phonation varies (Andruski & Ratliff 2000, Kuang 2011) or does not vary (Esposito 2012, Khan 2012) across different vowel qualities. The present study draws on innovations which allow for more accurate corrections for the effects of formant frequencies on spectral measures (i.e. Hanson 1995, Iseli, Shue & Alwan 2007) to examine the relationship between vowel quality and voice quality, in eight languages – !Xóõ, Burmese, Gujarati, Jalapa de Díaz Mazatec, Mon, Santa Ana del Valle Zapotec, White Hmong, and Yi. While no significant difference in the degree of non-modal phonation due to vowel quality was found, results showed a crosslinguistic pattern in the relationship between vowel quality and voice quality: vowels with higher log(F1) and log(F2) values tended to be produced with creakier phonation, while vowels with lower log(F1) and log(F2) values tended to be produced with breathier phonation, but only on the measure H1*-H2*.

In the current study, we explore the factors underlying the well-known difficulty in acoustic classification of front nonsibilant fricatives (Maniwa, Jongman & Wade 2009, McMurray & Jongman 2011) by applying a novel classification method to the production of Greek speakers. The Greek fricative inventory [f v θ ð s z ç ʝ x ɣ] includes voiced and voiceless segments from five distinct places of articulation. Our corpus contains all of the Greek fricatives produced by 29 monolingual speakers, but our focus is on the distinction between the front nonsibilant fricatives [f v θ ð]. For comparison, we also discuss the other places of articulation where relevant. We apply a relatively novel classification method based on cepstral coefficients, previously successful in categorizing English obstruent bursts (Bunnell, Polikoff & McNicholas 2004), English vowels (Ferragne & Pellegrino 2010), Romanian fricatives (Spinu & Lilley 2016), and Russian fricatives (Spinu, Kochetov & Lilley 2018). For this study, fricative boundaries were automatically aligned using Hidden Markov Models (HMMs) and then manually checked. Six Bark-frequency cepstral coefficients (c0–c5) were extracted from 20-millisecond Hann windows. HMMs were used to divide the fricatives and adjacent vowels into three regions of internally minimized variance. A multinomial logistic regression analysis then used the mean cepstral coefficients from each region as predictors for classification by consonant identity. Our method yields highly successful classification rates, exceeding the performance of previous methods. We discuss these results in light of the differences of the phonemic distributions of fricatives between English and Greek.

Presently there is no consensus regarding the interpretation and analysis of the stress system of Moroccan Arabic. This paper tests whether the acoustic realisation of syllables support one widely adopted interpretation of lexical stress, according to which stress is either penultimate or final depending on syllable weight. The experiment reports on word-initial syllables that differ in presumed stress status. Target words were embedded in a carrier sentence within a scripted mock dialogue to ensure that the measurements reflect lexical stress rather than phrase-level prominence. Results from all four acoustic parameters tested (f0, duration, Centre of Gravity and vowel quality) showed that there were no differences as a function of presumed stress status, thus failing to support an interpretation according to which stressed syllables are acoustically differentiated. We consider the results in relation to previous claims and observations, and conclude that the absence of acoustic correlates of presumed stress is compatible with the view that Moroccan Arabic lacks lexical stress.

Belarusian (ISO 639-3 BEL) is an Eastern Slavic language spoken by roughly seven million people in the Republic of Belarus (Zaprudski 2007, Census of the Republic of Belarus 2009), a land-locked country in Eastern Europe, bordered by Russia to the north and east, Ukraine to the south, Poland to the west and Lithuania and Latvia to the northwest (Figure 1). Within the Belarusian language, the two main dialects are North Eastern and South Western (Avanesaǔ et al. 1963, Lapkoǔskaya 2008, Smolskaya 2011). Two additional regional forms of Belarusian can be distinguished: the Middle Belarusian dialectal group, incorporating some features of North Eastern and South Western dialects together with certain characteristics of its own, and the West-Polesian (or Brest-Pinsk) dialectal group. The latter group is more distinct linguistically from the other Belarusian dialects and is in many respects close to the Ukrainian language (Lapkoǔskaya 2008, Smolskaya 2011). The focus of this illustration is Standard Belarusian, which is based on Middle Belarusian speech varieties. For details on the phonetic differences across dialects, the reader is referred to Avanesaǔ et al. (1963) and Lapkoǔskaya (2008).

Kalasha (ISO 639-3: kls), also known as Kalashamon, is a Northwestern Indo-Aryan language spoken in Chitral District of Khyber Pakhtunkwa Province in northern Pakistan, primarily in the valleys of Bumburet, Rumbur, Urtsun, and Birir, as shown in Figure 1. The number of speakers is estimated between 3000 and 5000. The Ethnologue classifies the language status as ‘vigorous’ (Eberhard, Simons & Fennig 2019) but some researchers consider it ‘threatened’ (Rahman 2006, Khan & Mela-Athanasopoulou 2011). Kalasha has been in close contact with Nuristani and other Northwestern Indo-Aryan languages. Among the latter, the influence of Khowar has been particularly strong because it functions as a lingua franca of Chitral District (Liljegren & Khan 2017). The Kalasha lexicon includes many loanwords from Khowar, as well as from Persian, Arabic, and Urdu (Trail & Cooper 1999). Early efforts to put the language in writing employed Arabic script but a Latin-based script was adopted in 2000 (Cooper 2005, Kalash & Heegård 2016).

The Bai language () is spoken by approximately 1.6 million people in northwest Yunnan Province, China. Of the 25 minority languages spoken in Yunnan, where 33% of the population are ethnic minorities and 67% are Han Chinese, the Bai ethnic minority is second in population only to the Yi (Wiersma 1990, 2003; 2010 census). Bai is classified as a Tibeto-Burman language (Xu & Zhao 1964, 1984), although arguments have been raised as to its possible early Sinitic origins (Starostin 1994, 1995). A summary in French reviews Chinese loanwords, ancient Bai, and comparative Bai dialects (Dell 1981). The historical influence of Chinese on Bai has been significant, but evidence is not compelling that Bai is Sinitic (Norman 2003: 73). There are three major dialects of Bái: Jiànchuān (), Dàlĭ (), and Bìjiāng (). The data in this illustration represent the variety of Jianchuan (jian1239, BCA). The third author (), who was about 60 years old at the time of recording, is a male native of the Jianchuan region, originating from QiÁohǒu, a mountain village some 50 km southwest of Jianchuan city – a remote area known for salt mining and where the language has been less influenced by modern Chinese. These locations are indicated on the map of Yunnan (the southwesternmost province of China in an intensely minority-language-populated area) in Figure 1. The traditional geographical link from Qiaohou is to Jianchuan to the north rather than to Dali to the south, and many of the most distinctive characteristics of Jianchuan Bai described here are not found in Dali Bai.