School of Languages, Linguistics and Film

Linguistics conference practice talks: NWAV 47

10 October 2018

Time: 1:00 - 3:00pm
Venue: ArtsTwo 3.16

PhD students Nate Young, Karen Beaman, and Ynda Jas will be practising their upcoming conference talks for NWAV 47 at NYU.

13:00-13:40 - Karen Beaman

ds beschte Daitsch wo s gib: Variation in the use of the wo-relativizer in Swabian German

This paper presents the findings from a variationist analysis of the polysemic wo-relative pronoun (‘where’, ‘who’, ‘that’), which varies with standard German relatives (der, die, das, etc.) in many southern dialects. It has generally been assumed that wo-relatives in German are used to signal some abstract notion of place and that their usage has spread to other contexts. To investigate this phenomenon, 1458 relative clauses were extracted from a panel study of 20 speakers of Swabian, a dialect spoken in southwestern Germany, who were recorded in 1982 and again in 2017. Results from the multivariate analysis show that, over a 35-year timespan, use of wo-relatives has decreased in nominative cases with animate antecedents and increased in oblique cases. External factors, most notably community (urban/rural), education, occupation and notions of ‘Swabian Identity’, also influence speakers’ choice between standard dxx-pronouns and wo-relatives.


13:40 - 14:20 - Ynda Jas

Pitch, affect and gender: the interaction of social and physiological factors in the speech of non-binary individuals

It is customary to assume that average pitch differences exist in the speech of women and men (e.g. Titze 1994). While sociolinguists have recognised the potential for these differences to interact with other factors like sexuality (e.g. Gaudio 1994; Levon 2009, 2010, 2012; Russell 2015; Smyth et al. 2003), research has rarely challenged the assumption that the pitch properties of women’s and men’s voices differ in fundamental ways. This is because of an underlying belief in the existence of inherent physiological differences across genders – differences that affect the pitch properties observed in speech. But how does this assumption map on to the speech of trans people – individuals for whom physiological differences do not necessarily correspond to actual, self-described gender? In this paper, I examine pitch variation within a population of trans speakers in Southeast England in order to investigate the role of social and physiological factors in constraining observed prosodic patterns.


Data are drawn from a study of 28 individuals from Southeast England who all identify as non-binary (i.e. in several ways other than woman or man). In this paper, I focus on a sample of eight individuals split evenly by gender/sex assigned at birth, age (older or younger), gender presentation (“femme” or “masc”) and sexuality (“LGB” or “queer”). I extracted approximately 300 intonational phrases (IPs) per speaker from speech collected during sociolinguistic interviews, resulting in a total dataset of 2,437 IPs for analysis. Initial mixed-model regressions considered how well the factors birth assignment, age, gender presentation and sexuality predicted three measures of pitch variation among speakers: mean pitch (measured in ERB), pitch range (in semitones) and pitch dynamism or slope (in st/sec). Subsequent analyses examined whether the affective content of talk (positive, neutral or negative) further conditioned observed patterns.


Analyses reveal that none of the four factors are significant predictors for mean pitch, though birth assignment is closest (p=0.0838). Birth assignment is the only significant predictor of pitch range (p=0.0234). In contrast, social factors play a bigger role in predicting pitch slope: sexuality (p=0.0260) and age (p=0.0338) are significant predictors, presentation nears significance (p=0.0597) while birth assignment is non-significant. When incorporating affect, its interaction with presentation acts as the best predictor of both mean pitch and pitch range: masc-presenting speakers have significantly lower pitch in affect-neutral IPs (p=0.0011) and greater pitch range in positive contexts (p=0.0442). While all social factors feature in at least one significant interaction with affect for pitch slope, the best predictor is the interaction with birth assignment and age: younger speakers assigned male show significantly less dynamism in neutral (p<0.0001) and positive (p<0.0001) contexts relative to negative.
All of this highlights that while physiology – better captured by birth assignment than gender itself – does play a role in determining some pitch characteristics, social and contextual factors also condition how the resource is strategically employed by speakers in interaction. This offers support for Zimman’s (2017) call for more nuanced approaches to studying gendered voices, taking account of birth assignment, identity, presentation and more.


14:20-15:00 - Nate Young

Multiethnolect moves a round u in style: F3 of /ʉ:/ is stylistically and socially stratified in Stockholm Swedish


This is a study of the LUS vowel (/ʉː/, henceforth ‘LUS’) among adult male speakers in Stockholm. Its main finding is that speakers from the immigrant-dense periphery de-round in casual speech and move toward the conservative rounded variant in formal styles. LUS is quite salient in Stockholm’s multiethnolect (aka ‘Rinkeby Swedish’), but there is no research on it to date. Is this supposed variant acoustically measurable? Is it widespread? Has it reached marker or stereotype status? 

The analysis examines three speech styles of 36 male Stockholmers: interview, reading, and word list. A mixed-effects linear regression model (n = 6206) predicts that speakers from high-immigrant neighborhoods will have a lower F3 in casual speech; however, this effect disappears in the reading and word list styles. My interpretation is that the multiethnic working class is de-rounding LUS and that there is social salience that attenuates it with formality. The analysis is still underway but it would appear that de-rounded LUS is a marker – or perhaps a stereotype – for Stockholm’s multiethnolect.