A speech production experiment with electroglottography investigated how voicing is affected by consonants of differing degrees of constriction. Measures of glottal contact [closed quotient (CQ)] and strength of voicing [strength of excitation (SoE)] were used in conditional inference tree analyses. Broadly, the results show that as the degree of constriction increases, both CQ and SoE values decrease, indicating breathier and weaker voicing. Similar changes in voicing quality are observed throughout the course of the production of a given segment. Implications of these results for a greater understanding of source-tract interactions and for the phonological notion of sonority are discussed.

Formant contours contain essential perceptual cues for vowel discrimination and distinguish ethnolectal and regional varieties that show superficial alignment in steady-state measures (Hillenbrand et al. 1995, Jacewicz et al. 2011). The analysis of contour information allows for more fine-grained investigation of dialectal differences in effects of coarticulation and duration. Additionally, it obviates arbitrary selection of measurement landmarks under traditional static F1/F2 analyses. Although acoustic vowel analysis of static F1/F2 values are foundational to sociolinguistic analysis and dialectology, vowel trajectories remain neglected despite their value in these domains (Koops 2010b, Scanlon and Wassink 2010, Thomas 2002:pg. 172). Our study uses functional data analysis of multiple formant measurements across the duration of each vowel to examine variation in Southern English of Piedmont, North Carolina. We show that differences in vowel formant trajectories are a key marker of participation in regional sound systems and ethnolectal vowel variation, illustrating the effectiveness of using functional data analysis to incorporate trajectory information into traditional sociolinguistic analyses. Our research investigates dialectal differences in front lax vowel contours among speakers of Southern varieties of African American and European American English (AAE and EAE respectively) in Piedmont, North Carolina (see Figure 1) using speech from sociolinguistic interviews. The diphthongization of the front lax vowels is a primary component of the Southern drawl thus serving as the perceptually salient cue of interest when looking at European American speakers in Raleigh. The Southern Vowel Shift (SVS) is receding in Raleigh, NC, leading to generational differences (Dodsworth and Kohn 2012). Among older speakers who participate in the SVS the front lax vowels BIT, BET, and BAT can be raised and subject to Southern breaking or diphthongization (e.g., [ae] becomes [aej@]) while these vowels are lowered and monophthongal among younger European Americans in the region. In contrast, the African American Vowel System (AAVS) has remained relatively stable for significant portions of the 20th century in this region (Kohn 2013). Among speakers who participate in the AAVS, front lax vowels are raised, in superficial alignment with SVS patterns; yet their vowel trajectories are distinct from European American raised variants. Generally speaking, these systems differ in that Southern EAE vowels are more subject to breaking resulting in greater diphthongization than AAE vowels (Holt 2011, Koops 2010a, Risdal and Kohn 2013). Although front lax vowel height of the AAVS and SVS are similar according to static F1 measurements, formant trajectories are a central component of linguistic diversity in the region. This comparison illustrates that analyses of formant trajectories capture sources of variation that may be missed by more common static measures employed in sociolinguistic analysis. From a methodological standpoint, we believe that functional data analysis as a method for comparing aspects of curve-shaped data improves upon traditional formant analyses for a number of reasons. First, it does not require pre-defined landmarks for measurement, e.g., at a certain proportion of the vowel or a maxima/minima. Static analyses can be problematic as methods for comparing dialectal differences in F1/F2 values using measurements at fixed landmarks, for example Euclidean distances, are impacted by such a priori decisions. Instead, functional data analysis allows for more holistic descriptions and contrasts to be made using measurements of intrinsically smooth vowel trajectories. This approach eliminates the need to make a priori decisions about which section of the