This study examines the relationship between paralinguistic measures (pitch range and speech rate) and sociolinguistic variables (sex, age, and educational level) in the 2020 PRESEEA–Valencia corpus. We analyzed 58 interviews using the XXX tool and Generalized Linear Mixed Models (GLMM) with a Gamma distribution and log-link function. The results reveal that pitch range is not determined by isolated factors but by complex three-way interactions. While a general reduction in pitch range is observed with aging, this effect is heavily moderated by sex and education. Notably, older males with medium education levels exhibit a drastic collapse in pitch range, resulting in the most monotonic speech patterns in the sample. In contrast, for female speakers, educational level serves as a differentiating factor in how pitch range is maintained or reduced during the aging process.
Knowledge about vowel-related acoustic features and their role in gendered speech remains limited, especially in Hijazi Arabic. While dynamic analysis, particularly vowel-inherent spectral change, offers advantages, Arabic vowel studies have focused on single-point acoustic features. This study analyzes Hijazi Arabic vowels across genders. It examines F0, F1, F2, and duration using static and dynamic approaches, including trajectory direction, trajectory length, and rate of change. Data were collected from 20 native speakers producing vowels in different consonantal contexts. Measurements used one midpoint for static analysis and five points for dynamic analysis. Results show gender differences, with female speakers generally exhibiting higher F0 and formant values and longer durations. Dynamic features provided additional descriptive insights and improved classification accuracy by 6.6% across F0, F1, and F2, with F0 as the strongest predictor.
This study analyzed acoustically the flaps of American English and Spanish. While the flap in both languages has been described as the same sound phonetically, acoustic studies comparing the flap in these two languages are scant. In this study four acoustic measures, closure duration, intensity difference, devoicing, and inferred place of articulation were employed to analyze contrastively the flaps of American English and Spanish. Results suggest that the flap differs in duration between the two languages, shares the same place of articulation, and that American English flaps show more devoicing and are considerably less constricted than their Spanish counterparts. Additionally, speaker gender contributed to some extent to the differences observed between both languages.
Changes in speaking rate and phonological length contrasts both involve temporal variation, with vowel quantity often serving as the main cue. This study examines how these two sources of temporal variation interact in Jordanian Arabic, focusing on vowel duration, spectral information, and the relationship between long and short vowels. The results show that changes in speaking rate produce temporal overlap between long and short vowel durations. Substantial spectral overlap is also observed in the F1–F2 plane, especially when both vowel types are produced at a fast speaking rate. Overall, the findings indicate that speaking rate and phonological length affect not only vowel duration but also spectral information and the separability of long and short vowels, although to different degrees.
This study compares the VOT perception of Mandarin /p/ and /ph/ by children (aged 5;08 to 13;11) acquiring Mandarin as L2 or L3 in immersion school. L2 participants (N = 15) spoke English as L1, while L3 learners (N = 12) spoke English and Spanish. In an AX discrimination task, all participants showed categorical perception of Mandarin /p/ and /ph/ with the same perceptual boundary and slope of identification as for English /b/ and /p/. No effects of age, length of exposure and current amount of exposure to Mandarin were observed. However, there was a more gradient transition between categories (more Mandarin-like) for L3 learners. This result is interpreted in light of the predictions of the PAM-L2 model and recent perspectives on gradient categorization in multilingual language perception.
